AUSTRALIAN FRONTIERS OF SCIENCE, 2003
Canberra, 31 July to 1 August 2003
Session discussions
1. Advanced materials
2. Extinction events in the geological past – causes and effects
3. Gene silencing – from basic mechanisms to industrial and therapeutic applications
4. Protein structure as a gateway to drug design
5. New approaches to the evolution of complex biological systems
6. Utilising large astronomical datasets
7. Quantum computing
8. Molecular ecology – integrating genetics with demography to understand population biology
Session 1: Advanced materials
Chair: Professor David McKenzie
Speakers: Professor Marcela Bilek and Associate Professor Dougal McCulloch
David McKenzie – I would like to congratulate both speakers on very stimulating discussions. This is going to be the fun part of our session, where we are going to do a bit of brainstorming. We are going to challenge the speakers to answer some questions from left field. A lot of you are going to have questions coming from a different area, and it is going to be quite difficult for the speakers to field the questions, I should imagine.
Click on image for a larger version
Just to set the thing going, I have got a couple of slides to stimulate our thinking into the importance of cross-disciplinary questioning and brainstorming. This slide is rather important, because it shows the way we think of knowledge as a whole, and perhaps the way we should be thinking of knowledge.
This is the way we are taught to think of science, or knowledge. It is a very rigid sort of structure: it is a tree and it does not change that rapidly. It grows a bit; the green bits at the top are the growth area. That is the way we are taught to think about knowledge as a whole – as a tree. It builds on itself. But what is happening now is an exciting era where each area of knowledge is sending out feelers into the other areas of knowledge. I have got there what has happened to devices.
I should explain, perhaps: this tree has got two major arms, as I see it. The one on the right [Science of self organised structure] is essentially the science of things that produce themselves. They are there because of millions of years of evolution or something that has happened in the past, and we are studying what has happened. So you can see the disciplines of that; you can think of it a little bit like that.
On the left [Engineering of machined structure] are the things that we are doing, as humans, interfering with Nature. We are kind of engineering Nature. If you look at the field of devices – I am thinking there of devices that might be used to make a computer – that field has benefited immensely and the current progress is happening by feelers being sent out into the other areas of knowledge. So you may ask questions that you think are from left field, but they might actually be the beginning of something important and something new.
Click on image for a larger version
So my next slide shows the way we should be thinking about knowledge, as a web rather than as a tree – something that has lots of interconnections and thrives off those interconnections. Perhaps with that intention we can start the questions. They can be hard questions that no-one can answer – I imagine there are lots of those – but we will try. We will try and field some questions.
Question – I am an astronomer. In astronomy, when we model cluster formations – galaxies, et cetera, large-scale formations – with computers, there is a resolution problem. Because of that resolution problem, we cannot trust the simulations on small angular scales, the very small scale. I am wondering if you have a similar problem, Marcela, when you are modelling. You showed that nice computer simulation of epitaxial growth. Do you have a similar problem with resolution, that computers cannot get down into the electron cloud or even deeper, to accurately portray what is going on?
Marcela Bilek – No, we don't actually have that problem. The two models I showed are a little bit different. The one that showed the electronic cloud actually has an algorithm in it which calculates the electrons as a wave structure. So you are doing the quantum mechanics to see what the electrons are doing as a wave. The trouble with that, though, is that it is very time-consuming, as you can imagine, and so you are really limited to simulations that include of the order of hundreds of atoms. That is as big as you can get, and that is still very small. And the timescales are picoseconds, so very, very short timescales.
If you want to look at something like the second one I showed, the film growth, that takes longer and it also involves many more atoms, including the substrate, which you have to model. So to do that we use the accurate simulation – that tells us what small clusters of atoms will do together, using the quantum mechanics and the behaviour of the electrons – and then we develop our own potential, as we call it. It is basically a formula that tells you how neighbouring atoms, when they get to certain distances, will interact: how strong the forces will be between them. We develop this potential, and in the case that you saw it is called EDIP – environment dependent interaction potential. That means that it takes into account not only two atoms interacting but also the chemical environment, what else is around it. It is quite accurate. The reason it is so accurate, though, is that it has been developed or tested against that very accurate prediction including the quantum mechanics of the electrons.
So that is the approach we take. The empirical potentials on their own are not very good unless you can make sure that they predict all of the different structures you get when you do the quantum mechanics. The resolution is not a problem. We are actually down at the limit of looking at electrons as waves. I think there are different algorithms as well.
Dougal McCulloch – We have the other problem, in a sense, in that we have trouble going to big systems.
Marcela Bilek – To larger scales, yes. That is the main problem.
Dougal McCulloch – Or going for long enough.
Marcela Bilek – That is simply because of the computer power, because the sort of algorithms that we use to get that accuracy, we just cannot do large systems. It would involve incredible amounts of
computer time.
Question – As we go from branch to branch in what you were suggesting, David: Marcela, you talked about the integration of the types of materials that you work with in biological systems, in compatibility in those systems, and then at the end you talked about DNA microarrays, which are actually life systems being assembled and reused in life sciences. What do you see, either of you, in terms of applications going the other way, of the use of biological macromolecules in the sorts of nanoscale or microscale structures that you make, heading back into the world of the physical sciences – applications in that area and where that is going?
Marcela Bilek – I guess the sorts of areas that we have been looking at mainly are devices – biosensors, perhaps, so your biomolecules would do the sensing. With the specific interactions they would maybe capture an antigen or something like that. And then you would have a device off that, with its electronics and readout and so on. So that is one example.
Another one that I know of that people are exploring is using individual molecules, in some cases biomolecules, to try and assemble maybe electronic devices at a very, very small scale, so using a single molecule, for example. I am not as familiar with that work, but I know that it is happening.
David McKenzie – I think that is a really interesting area, and we should note that biomolecules – macromolecules – are very sensitive to their environments. They are actually very good at knowing what temperature the environment is at, what chemistry the environment is at. So they are the ultimate sensors, really. If you want to make a good sensor, use a macromolecule. Are there any questions related to that?
Question – This is not so much a question as continuing the conversation. I think this is very relevant. I recently attended a meeting of the semiconductor industry in the United States, where they were looking into the future of the conventional computer industry over the next seven years. They were trying to work out what technologies Intel, for example, would adopt and where they were heading. First of all they showed a very nice movie of DNA – it being the 50th anniversary of DNA – showing how life forms as a machine, as it were. Then they got onto the sorts of things Marcela was talking about. The interesting thing for me was that they actually combined two things. Marcela talked about nanostructuring in the first instance, and then self-assembly in the second instance, like top-down and bottom-up. What they are looking at is a combination of both. They actually call it guided or directed self-assembly.
The view in the industry is that self-assembly alone is not very useful, because it goes in uncontrollable directions. But if you do some low-entropy templating, with a probe, the sort of thing that they are looking at is whether you can, with such templating – say at the atom scale, using some scan probe – then allow self-assembly to build the rest of the transistor very quickly, for example. So it is exactly that technique of self-assembly to assemble an electronic functional element.
The other thing that I would say is related to one of the things you said, Marcela, that if we are doing things at the atom scale, we need to do something more quickly or more high-volume production for it to be relevant. Well, there is another interesting thing that happens there. That is that once you start getting down to devices at the atom scale, it turns out you actually don't need that many more components. You don't need so many components to make something work very powerfully.
So the other thing that came out of this discussion was that the key functional element of a future computer – what is at the core of computers – may actually be very small. It could be just using tens or hundreds of atoms. The thing you hook up to it, of course, will have all the conventional microelectronics. But there is just that key little bit. So we may not need to do high-volume.
Marcela Bilek – That is true for devices, such as computers. But also I was trying to talk more generally about materials, and it is not true for, say, coatings on devices in the body and so on. There you do need the bulk, and so assembly atom by atom is not an option. You really are looking at self-assembly if you want fine structure.
On the previous comment about the directed self-assembly: that sort of fits in to what I was trying to say about the external constraints. So we can direct it also by changing the external constraints, and that would be the structuring, I guess, that you are talking about.
Question – I was thinking that with self-assembly and your directional probing or other constraints, that gets away from the limitation of masking, doesn't it? It seemed to me that part of the masking technology is just how fine you can make the masking in order to do the thing.
Marcela Bilek – That's right. Or how you can expose the masking, and I guess the limits there are the electron writing. That also is quite expensive and slow at this stage, I think. Self-assembly also gets you past that, but of course the challenge is understanding the elements well enough to apply the correct constraints so that you can direct the assembly.
Question – And in a dynamic way?
Marcela Bilek – That's right, and also knowing what structures you are after. We are still at the stage of not understanding quite how the structure influences the properties in a lot of these materials. So we need to unravel that first and then try and build the structures as well.
Question – Dougal, in optical microscopy it has been very useful to be able to look at a plane inside something. [Dougal McCulloch – Do you mean confocal-type arrangements?] Yes. I was just wondering whether you were able to do that. Most of the things you showed, I though, were surface or just looking at an aggregate.
Dougal McCulloch – That's right. It is more of a problem in electron microscopes because your sample has to be thin anyway, so you are really looking at a transmission electron microscope. I touched on it a bit with the holography. People were trying to do holography and tomography of structures so they could reconstruct them in the material, which is sort of what you do in a confocal approach as well. Once you have got your data, you have a big dataset in your computer and you pull out bits. Hopefully, we will think of new ways of doing that, because it is certainly useful.
Question – I have a question also for Dougal, about cryo-EM. He did not touch on that, but that is clearly a very important technique for characterisation, particularly of biological samples. I would be interested in your comments on where that is going.
Dougal McCulloch – Yes, that is a very important technique. I was coming at it from the materials angle. I know that there is a new cryo-microscope going in – at Queensland, I think. I am not 100 per cent sure of this area, because of my background, but there are not cryo-TMs that allow one to look at samples that are wet and so forth. The problem, of course, when I talked about environmental SEMs or microscopes, is that they were for surface structures. So if you want to actually look at the internal structure, you need to put a thin sample – almost a thin wet sample – into a microscope. You might say that is impossible. I am not sure I can say much more than that, though.
Question – I am from entomology, so I am right up on nanotechnology. I have a question for you, Marcela. You were talking about nanomachining or something, I think, where you are making things on smaller and smaller and smaller scales. How useful is that if you get these completely different effects right down at the quantum level?
Marcela Bilek – Well, in a lot of cases you actually want to utilise those effects. A lot of devices actually rely on your getting those effects, so you do need to go that small. Obviously, if you were trying to make something that does not use that effect, then you would not go that small. It is actually harder to make it small, anyway.
Question – It is a bit hard for me to actually ask the question, because I am not quite sure what I am asking, but if you make something and it is bigger and so it does not have these effects, how useful is it on the nanoscale? Does it actually give you information about the effects that you are going to see on the nanoscale?
Marcela Bilek – No. So you can make the same thing and if you make it smaller and smaller, this is where quantum mechanics comes in. Quantum mechanics changes entirely the way things behave. When you do first-year physics, you learn about Newtonian mechanics: how does the bus go, acceleration, velocity and this sort of stuff. (You're shaking your head. Oh dear. Anyway, the sort of motions that we are used to, you see on the scale that we live in.) You find that when you do quantum mechanics, perhaps in the second year or so on as you go on through physics, it is completely different. It is very hard to adjust to it, because we just don't see those effects. They do not exist on the sort of scale we are used to. Now we are starting to see them because we are building things on that scale.
Also, that understanding was needed to understand some phenomena that we do see, that come from that sort of scale. But it is completely different. It is a completely different kind of physics. So you do find that as you get smaller and smaller the behaviour changes completely. That is a good thing, because we now have the option to design systems that utilise those behaviours, as well as systems that utilise the classical behaviours that we are used to.
Question – You have shown biological samples like the abalone shell or the butterfly wing. My question is: is there actually research into how biological systems like animals form these structures, and is it maybe possible to adapt strategies that animals and plants use to create materials along the lines that you are doing?
Marcela Bilek – I am sure there is work. I am not the person to ask, but I know that in the biological sciences they are looking at how the DNA is expressed. Basically I think all of this comes from the DNA code and how it is expressed when the animal is forming. So there is research ongoing in that field to see how that is done. It is an incredibly complicated self-assembly process, and many self-assembly processes piggybacked on each other, where the previous one I guess you can think of as setting up the constraints for the one that follows, and so on, and they all have to play out in sync. I think the sort of understanding that is gained in that field will be helpful to designing strategies in our area.
One field that is also very important for us is a field called biomimetics, which looks at understanding what those structures are, first of all – so looking at the microscopy and actually seeing what that butterfly wing is made up of, how it looks in three dimensions – and then understanding how that structure produces the properties that it has, that we see. That is also important. So there are those two very important fields that are in the biosciences. We see a big interaction there.
David McKenzie – Perhaps I could add a point to that. The field of the butterfly structures is actively being pursued at the moment, and a number of related species are being studied for the genetic expression. The related butterflies that are able to produce this photonic crystal have genes in common. Of course it comes from the genetic code originally, but the way in which it does so is a very complex problem – and an interdisciplinary one, I am sure.
Question – This question is for Marcela. With that simulation of film formation, are you able to include various sorts of forces that you can envisage, that might help to organise the formation of that film – I guess it is directed self-assembly – so bringing in some forces and trying to say, 'Okay, I'd like to grow a diamond,' or whatever?
Marcela Bilek – So you mean perhaps putting an electric field on it, or something like that. Essentially the answer is yes, if we can calculate the effect of that force in the computer. For example, with the electric field, yes, you should be able to do that. But again it depends on the algorithms. We certainly can put in a lot of forces.
One thing that is difficult, though, with the one that I showed – the environmental dependent potential – is that in terms of adding an extra element it is difficult: you have to redesign the whole potential because it depends on what chemical environment the atom is in. So if you want to add, say, silicon you have to do it all again and look at what happens in silicon carbon systems and so on. So again, with additional forces, it depends on what force you mean, essentially – which particular ones.
Question – Dougal, looking at your beautiful simulation I am interested in the fully quantum model that you do. I do a lot of quantum things myself, and I am very impressed that you can take a system of, say, 28 carbon atoms and actually do a fully quantum mechanical model. That had something like 300 antisymmetries, fermions – it is a stunningly complicated system. I am presuming there must be some fairly powerful approximations you use.
Dougal McCulloch – Oh, definitely.
Question – Can you say anything about how that is done?
Dougal McCulloch – Yes, we can. It again was using a technique known as density functional theory as the main approximation and advance to be able to do that sort of simulation on such a large number of atoms. We do a number of things. We only calculate what happens to the valence electrons, for instance, and we use a pseudo-potential, which is a way of dealing with the core of the atom. And we have to make an approximation about the interaction of electrons with one another, how the particular electron reacts with the local density. That is part of density functional theory. And there are a lot of others. If you wanted to do the highest quality or the most accurate calculation you could, then you would probably limit it – you probably know more that I do – to a few atoms, to get really high accuracy. These methods here [as spoken of earlier] are quite accurate but we still have to make some assumptions.
Question – I guess I am wondering: are those assumptions empirical, or are they based on a phenomenological point of view, or are they just a self-consistent, iterative method to try and get your dynamics? – if that makes sense!
Dougal McCulloch – They are self-consistent, yes.
David McKenzie – They are approximate, but they work amazingly well. They have been tested on a huge variety of molecules, and give the right answer for bond length, bond angles, anything you care to name. We have also found, in our work, that they can be trusted in very large numbers of atoms. We have been up to 500 atoms, with full density functional theory, and it really does predict what happens.
Marcela Bilek – There are some things that you can't get, though, because of the approximations. For example, it assumes that you are in the ground state with the electronic wave functions, so you do not get information about the higher electronic states. For example, band gaps are very difficult to predict. So you do lose some information about the electrons by treating it in this way, with the wave functions, the density functional theory. But it makes it faster.
Session 2: Extinction events in the geological past – causes and effects
Chair: Dr Charley Lineweaver
Speakers: Dr Annette George and Dr Kliti Grice
Charley Lineweaver – I think what we have just heard is a good example of multidisciplinary science – the web of science. We have a geologist looking at rocks and a geochemist looking at molecules. You two are both working on the PT boundary. Can you give us an example of how, by talking to each other, you have been able to make progress in your relative fields, and/or how talking to other people who are studying, like palynologists – people who are studying pollen and the transitions of the PT boundary – has helped you in your research?
Kliti Grice – We see correlations with the palynology. The palynology is one aspect of preservation, and biomarkers are another aspect. So when you find the two together and you know where they are from, then you can conclusively get information about the environment which existed. Also, the carbonates are particularly important in terms of biogenic carbonates which are produced in the water column, in terms of the incorporation of the isotopic composition of the carbonates into the biogenic carbonates.
Annette George – I would just add to that by saying that we actually haven't been working on the PT stuff for very long. But I think one thing that you notice about the work is that everybody is working on their geographically different areas, and you really need to be able to talk to people who bring experience from having seen other areas or – for example, Clinton Foster (Geoscience Australia) – from having worked on faunas in the different areas such as China. Really, it is a fragmentary record, and if we are going to put forward realistic and robust global mechanisms then we need to bring in the evidence from elsewhere. And not every section records all of the same kinds of things, so the geochemistry might be more important in one locality, the faunal record may be much better in another locality. So that mixing of ideas, experience and knowledge from other sections is really what is required to answer these big questions at a global scale.
Kliti Grice – With things like green sulfur bacteria, you would never be able to see those by any other method which has been established to date. You might get something else out of the geochemistry which you cannot get out of the palynology, and vice versa.
Charley Lineweaver – So there are no tiny microfossils of green sulfur bacteria anywhere?
Kliti Grice – No.
Question – I was just thinking back on the 13C, the original scale you put up, where you were showing rich and poor 13C rocks and so forth. You had an arbitrary baseline as marine bicarbonates, and you had various different kinds of rocks and so forth. I did not really get whether that was current rock production – in other words, whether different organic molecules were selecting different isotopes and therefore you were actually using the 13C as a detector of different kinds of environments, or whether you assume that the isotopes are going isotopically into different rocks and therefore it is a measure of the atmospheric carbon.
Kliti Grice – It is both. Different plants use different pathways – C3 and C4 carbon – so they contain different enzymes which will fractionate the isotopes to different extents. You see those differences amongst the species today.
You also, in the past, have got this complicating factor of the changing climate, because things can change the isotopic composition of the atmospheric CO2. And the atmospheric CO2 in the last 200 years has changed. It has become lighter, and the proposed reason for that is the input of fossil fuels, which are isotopically light, with the burning of fossil fuels. They have been derived originally from phytoplankton and so forth.
Question – Are we in the middle of a mass extinction caused by a single species, namely, us? Is there any evidence for any of the mass extinctions in the past that are caused by the dominance of one organism, whether it be sophisticated or primitive? That is, it may be a pandemic or a virus or something like that, or a toxin produced by a species.
Annette George – I will answer the second part first: There are probably people in the audience who might have a better answer than I do on this one, but certainly where there are spikes of particular things – for example, lots of fungi recognised in the rock record at particular horizons, lots of microbes coming in – generally the story for those has been that they are inhabiting a niche that is now available. So most of the evidence I have read about has tended to focus on where things have benefited from there being the demise of other organisms, rather than there being something that has caused the demise.
On the first question: yes, there are certainly a lot of people who argue that we are in a 'mass extinction' phase at present day. But I guess I go back to those comments I put up earlier, that we have problems simply measuring our species diversity from which to be able to measure species loss. I guess again the notion of time comes in, although on our timescales it is much shorter. But there certainly are issues in terms of understanding, being able to measure loss to really estimate what has happened. And whether it is truly global – whether it is occurring across widespread geographic areas – would be a second question.
We were discussing at lunch the megafauna loss. I guess there have been a lot of different comments put forward, but there was a nice article published just recently about how the diprotodon megafauna potentially coexisted with the Aborigines for quite some time. The suggestion was that potentially the Aborigines were not responsible for the early loss of, particularly, things like the megafauna. But again you have to take a lot of those articles with a grain of salt, because the dating of the deposits is often quite difficult and the results sometimes seem a bit more sensational than potentially they end up being. But it is an interesting debate.
Question – When you are looking at these patterns of extinctions, you are looking, it seems to me, to be trying to explain 75, 80, 90 per cent of the species that go extinct and that is why you get catastrophic sorts of explanations. Is it ever the condition that people look at the survivors? Can you use the fossil record to examine those species that survive, and why they do it, to give you some inference about what has caused the extinctions?
Annette George – I guess a good example that springs to mind is the microbes, which in the Early Triassic have an absolute field day. In the Late Devonian they also take over. I did not have much time to talk about it, but one of the suggestions has been that, for example, the microbes may well be able to live in much poorer conditions. Certainly in the Early Triassic there is plenty of evidence that conditions were highly stressful. It took about 10 million years for many of the big ecosystems to get going again, so big lag times. But, as to whether you can really see evidence for it, in the Canning Basin the microbial build-ups don't appear to be particularly oxygen-poor. They trap haematite in their structures, which suggests that ocean water was quite oxygenated. I guess the microbes would be one of the groups that you might argue were opportunistic enough to live in more stressful kinds of conditions.
Kliti Grice – A lot of the biomarkers come from microbes. From the samples we have looked at so far, there is evidence that the bacteria survived, and the cyanobacteria survived the event – in particular, the bacteria.
Question – I guess that has answered my question. I don't know if you have got an interest in lungfish, for example, where there were thousands of species at the end of the Devonian and now we have only got one or two. It would seem to me that lungfish swim as well in lava; one species will swim as well as another one. So what didn't make the one that lived in Australia survive?
Annette George – I guess at the end of the day people have been arguing that the extinctions really represent bad luck rather than bad genes per se.
Question – You mentioned that in the Late Devonian the bacterial record suggested that anoxia was not a reason that the extinction event occurred. Does that necessarily mean that anoxia is cut from the list, or could you have spatially different types of extinction happening around the world?
Annette George – I left anoxia on the list because I think it remains very much an open question. In well-documented sections in Germany and places like that there are well developed black mudstones, and there are at least two horizons of them, just below the Frasnian-Framennian boundary. And in a number of other sections there are also these anoxic-looking rocks, hence this emphasis on anoxia as an extinction mechanism. In the Canning Basin we don't see those same kinds of rocks, and really it does pose the question of whether the microbes themselves are showing an ability to live in a lower oxygen environment. At the moment, it is very much an open question. Generally, the published research tends to be quite conjectural with not data to really support some of those ideas. So the ideas are out there, and I guess we are trying to find ways of testing those ideas and whether the evidence is there or not.
Question – Kliti, you mentioned the methane released from hydrate decaying, with the warming atmosphere, and the runaway greenhouse effect coupled with that. Do you think there is a way to assess how much of the global warming is due to CO2 by oxidised methane in the atmosphere and how much may be due to methane itself? It is a much more powerful greenhouse gas than CO2.
Kliti Grice – I don't know. I don't think it has been established. There is always the problem with mass balance of how much methane was actually released. The isotopic signature is the lightest of all components we know of which might account for this excursion, but how much was released I don't know. It could be a cumulative effect. It might be that the Siberian Traps and maybe forest fires and then the release of methane all contributed to this greenhouse gas.
Annette George – The KT boundary work has been very much improved by very good dating on timing of the start of volcanism, and of the actual bolide impact. What really hampers the Permo-Triassic work is that it is a lot harder to define some of the ages of events. The Devonian has good microfossil control; whereas the Permo-Triassic does, but to a lesser extent. So that has added to the complication around that boundary as well.
Question – As a species we seem to be fairly interested in our own destruction. Do we know how many gigatons of methane are currently locked up in the Arctic tundra? Do we know how many gigatons of carbon dioxide are locked up at the bottom of the North Atlantic Ocean? And, given synergistic effects and mass extinction, what kind of tectonic activity would be required to bring all of that carbon dioxide fizzing to the surface?
Kliti Grice – I don't know about the CO2, but there is about 1x104 gigatons of methane trapped in gas hydrates, which I said was twice the amount trapped as fossil fuels. But I am not sure how much CO2 would be trapped.
Question – So if global warming resulted in the melting of the tundra, that would release all of that methane and it would become self-perpetuating. Is that what you mean by runaway greenhouse?
Kliti Grice – Yes. You get huge pockets of methane from gas hydrates today. It is happening all over the world, so it is a process which is occurring. But it must have been a catastrophic effect which occurred at the Permian/Triassic for this mass extinction.
Charley Lineweaver – Just as a control sample: is it ever the case that you see the type of isotopic anomalies that you showed for the PT in situations where there is no correlated extinction?
Kliti Grice – Yes.
Charley Lineweaver – As strong?
Annette George – Yes. And I guess for all of the mechanisms that we have shown, people have argued in a negative sense to say that maybe they don't really work in many cases, because, for example, there are many other times when you have major fluctuations in sea level and there aren't extinctions. Similarly, there are major shifts in isotopic composition, in response to changes in the atmosphere/ocean system, that are not linked to extinction. So people are using them to say that maybe we are not looking at things the right way, because we can find plenty of times when extinctions are not linked to them. And the same with the volcanism: there are lots of times when there are big outpourings, not necessarily linked to extinctions.
Question – You have linked the methane release to the breakdown of methane hydrates. Could there also be an effect of methanogens releasing methane, rather than a breakdown of all these hydrates?
Kliti Grice – Methanogens would produce methane. They are the biogenic organisms which produce the methane. I would not like to say that this is the answer for the Permian/Triassic release of methane. There is some evidence to suggest it is part of the process, but these organisms can also occur at volcanic vents, round volcanoes beneath the ocean, so it is a suggestion rather than a conclusion.
Session 3: Gene silencing – from basic mechanisms to industrial and therapeutic applications
Chair: Dr Wayne Gerlach
Speakers: Associate Professor Levon Khachigian and Dr Peter M Waterhouse
Question – This question may be a bit specific. It is about the heat shock response, the response of an organism to heat stress. Do you consider that that mechanism is not activated by a coding gene but by the microRNA pathway which is coming from the non-coding gene? What is your thought on that?
Peter Waterhouse – I guess my thought is: that it could well be. I think we are really just discovering the breadth of the effects of microRNAs. Some of the stuff that I just talked about was only reported three weeks ago. In the past, people have found funny phenotypes from insertational mutagenesis screens and could not map the mutation to any coding region. Now there is a possible explanation: the mutation could be in a microRNA-encoding region. I think we are going to see a whole plethora of effects that are modulated by microRNAs, and heat shock response could well be one of them.
If I may put in my sexist bid here: we just heard from Jenny Graves, at the international congress on genetics in Melbourne a couple of weeks ago, that the Y chromosome is pathetic. She said that it has only got a few genes on it, and the rest is a genetic wasteland. I would be nice to think this wasteland is not junk but actually producing microRNAs which are controlling lots of processes, and maybe the Y chromosome it isn't so wimpy after all.
Question – I was wondering if you could just comment on the two techniques. Suppose I had some RNA and I wanted to get rid of it. What is the difference in maybe the efficiency or the specificity of the two?
Levon Khachigian – No-one has actually done a side-by-side comparison, comparing DNAzymes with siRNA. We are actually doing that right now, targeting the one gene using different technologies. I think a robust comparison will provide clues to what technology is the most appropriate.
There are complications, like for example delivery, and also target accessibility. The mechanisms are quite different. So it could be that the answer might be gene-specific, or it could be combinatorial.
Session 4: Protein structure as a gateway to drug design
Chair: Dr Jacqui Matthews
Speakers: Dr Joel Mackay and Dr Bostjan Kobe
Jacqui Matthews – I will start off the discussion by asking either or both of you: when you start thinking about using protein structure for drug design and developing drugs, how do you think it compares with using things like the DNAzymes and the RNAi that were talked about yesterday? There are so many different approaches to producing drugs and things. How do they stack up against one another, and why are there so many different things there?
Bostjan Kobe – I think we should not discard any approach. Basically, the problem is that we continuously need new drugs, because bacteria and viruses mutate too quickly and become resistant. Also, there are obviously a lot of diseases that we have no drugs against. So I think we have to use every approach we can think of. For example, in the case of HIV, new drugs that have been developed, based on knowing the structure of HIV protease are now on the market. But obviously all these other methods have their advantages and disadvantages. For particular cases one may be more applicable than any others. I think they are all complementary to each other and we should pursue all of them. It is the same with other things as well – high-throughput screening of natural products and things like that all have their merits and they all should be pursued.
Joel Mackay – That is right. I think different techniques are going to be useful in different cases. If you are targeting things on cell surfaces versus inside cells, or things that are in a plasma versus things that are in the brain – all the different types of diseases you might want to treat have different characteristics that might make RNA or DNA or protein or small molecule drugs more or less appropriate, I think.
Question – I was struck by the fact that Joel finished up with something that was surprising, in his first attempt at designing a molecule, and Bostjan said that you had pretty good accuracy in finding things out. I just wonder what the difference is there. Are you finding, for any particular site that you then investigate back to the kinases, that you are not getting any ambiguities? And is that because you have had enough real biological examples to give you the better rules for that particular class of molecule?
Bostjan Kobe – I said the predictions have pretty high accuracy. That doesn't mean they always work. There are several things that I have to address here if I want to answer that question. Number one, we have to make certain assumptions, but it turns out that in this particular system those assumptions hold pretty well. One assumption is that all kinases have a similar structure and bind substrates in the same way – that is what it is all based on. So they all have to bind from N-terminus on this side to C-terminus on this side, and bind in the same pockets, otherwise our predictions will be wrong. In terms of this particular system that works, and I think there are a few other systems, for example phosphopeptide-recognising domains, where that type of thing works but it might not always work and there might always be some surprising cases.
The other point is that this recognition is not the only thing that determines what is a substrate of a kinase inside the cell. The only thing we have here is how a kinase recognises its substrate. But in fact what happens in the cell is that kinases also have other domains, and those domains recruit proteins. They bring them close together. And if they are close together already, then the kinase doesn't have to recognise optimally the substrate, but because it is locally there in a high effective concentration it can still get phosphorylated. At the moment, with what we have got, we can't take these things into account other than to say after we have done our predictions, 'Okay, in this case it is likely that this type of thing happens.'
What we would like to do in the future is actually integrate this type of information with other types of information. An integrative approach will have a higher predictive value. But of course you always have to be careful: you have to remember what the assumptions are and you have to then look at it case by case and try to identify when problems could arise. It is not going to work in 100 per cent of cases.
Joel Mackay – Another issue is that the prediction that Bostjan is doing there is based largely on the primary amino acid sequence – when you look at certain amino acid sequences you can predict that they may be phosphorylated by a certain kinase. The stuff that I was talking about is predicting the three-dimensional structure of a protein, trying to design three-dimensional structures. The rules for understanding three-dimensional structures and how they are put together are much less well understood than the one-dimensional rules that people have used to decide whether something will be phosphorylated or not. So the problem, in a way, is more difficult, I guess – in one sense, anyway.
Question – This is for Joel. I find it absolutely fascinating that in the result, your scaffold with functional bits attached, it attached to a different part of the target. Can you comment on that?
Joel Mackay – Certainly it was not the result we expected. I think it highlights why people have so far not been able to use computer modelling to determine protein structures and to predict protein interactions. You can do molecular modelling, where you say, 'Here is a protein sequence and I am just going to feed it into a computer, with no experimental data at all, and predict the three-dimensional structure.' People have been doing that for 10 or 20 years now, with limited success – and the success is still relatively limited, doing that. That is why people like Bostjan and I have still got a job determining protein structures experimentally, because the job is just so hard. Proteins are just very difficult things to deal with, because they have 20 different amino acids and there are a lot of different possibilities.
People understand a lot more about DNA structure – how DNA is put together – and how you can treat DNA and what its behaviour will be, because it has a much more limited kind of character. There are only four bases, and basically they always fit together in pretty much the same way. I think the result is just a measure of protein complexity, of how little – still – we understand about how proteins fit together. I think that is pretty much what it tells us.
Question – Bostjan, I am really interested in the macrophage work and your focus on the innate immune aspects of that cell. I should add that type 1 interferons are very important molecules in that phase, and that involves double-stranded RNA, just like what we heard about yesterday in plants. So there is a nice correlate there.
Macrophages, as well as their innate immune activity, also have a role in adaptive immunity – that is, the immunity that generates immunological memory. Obviously, you want to get structures and possibly manipulate proteins involved in innate immunity. Are there any dangers that by doing that you might compromise adaptive immune responses?
Bostjan Kobe – We have to look at this case by case. One of the reasons we are doing what we are doing is to find out exactly what certain proteins do in a certain process. For example, at the moment all we are doing is looking at the response of macrophages to lipopolysaccharide, which is one of the things on the outer surface of a gram-negative bacterium. There are many other ways it can get activated and many other things they do. So if we identify a protein that is involved in that, with all the other things that we are doing – for example, looking at structure and also at some other functional aspects – hopefully, we will identify what exactly it does in playing that role. And when we know that we can then probably, in favourable cases, have an idea of whether this is a good target or not, whether this will actually interfere with some other important processes, in which case it will be toxic and not very useful, or whether this is a good target which is specific for that particular response and we can use it in that way.
We are able here to identify new targets, but again we have to then validate the targets. If they are useful targets, then we will pursue and design drugs against them; if they are not, we will use other ones.
Question – It is very important work, because some of my own studies with colleagues have shown that viruses are actually able to specifically target transcriptional protein complexes and, basically, switch them off so particular genes are not activated. So I have a very strong interest in this. Just balancing the two roles of the macrophages I think is of a long-term fascination.
My next question is to Joel. Basically, in the cross-disciplinary nature of this meeting, I would like to know what it is about the fundamental chemistry of zinc that makes it such a popular motif in proteins.
Joel Mackay – I think one of the issues about zinc is that it is very redox-stable. There is only one common oxidation state of zinc, and I think that means that proteins that use zinc for a structural role do not have to worry about possibly being oxidised or reduced, which might change the nature of the coordination chemistry of zinc. If there were zinc 3+ or 4+, it might not coordinate cysteine and histidine amino acids in the same way that the zinc 2+ does. I think that is one issue. Zinc is also very common; it is one of the more common metals in your diet. So I think it is probably just fortuitous that when proteins evolved it was, 'Oh, there's this zinc lying around. Let's stick to the zinc' – that sort of thing.
One thing is that people do not actually know whether most 'zinc-binding' proteins are actually zinc-binding proteins. For probably 95 per cent of all things that are supposed to be zinc-binding proteins, no-one has ever actually made the protein – they have just seen the DNA sequence. So it is entirely feasible that lots of 'zinc-binding' proteins are actually iron-binding proteins or cobalt-binding proteins or something like that. We don't know for certain.
Question – By using phage display you can do antibody libraries. I was wondering: maybe it is already being done, but can you do such a thing with zinc fingers on?
Joel Mackay – Yes. We are doing phage display just at the moment, actually. We are just getting the technique going in the lab, and hopefully within the next couple of months we will have our first result. We have made our first peptide library, so hopefully we will start to see whether we can generate some diversity that way. That is right, yes.
Question – I am going to throw in probably a controversial question, comparing a solid-state characterisation method with a solution-state characterisation method.
Joel Mackay – That's all right, we've all heard this question before. We can cope with it.
Question – With small molecules we see that in solution state the aggregation can be quite different from that in the solid state. So when you are talking about your pairing between the two biomolecules, have there really been any studies to see how good the comparison between solution-state structures and solid-state structures has been?
Joel Mackay – People are intensely aware of this issue. Basically, ever since NMR started to be able to determine protein structures, which was only about 20 years ago – crystallography has been going for much longer than that – and once people started to realise that you could solve structures using NMR spectroscopy, there was very strong interest, and still has been, in whether both experimental methods give you the same kind of information. In most cases, the cases where structures have been solved using both methods, they are almost always identical. In the cases where they are not the same, it is usually clear why there is a difference. For example, the formation of a crystal may sometimes induce certain parts of a protein to pack in a way that they don't necessarily pack in solution.
So there is always that issue, and both of us always have to be wary when we determine structures: 'Okay, this particular feature of the structure may be an artefact caused by the method we are using or it may be real.' Both methods have their own artefacts that they are likely to give rise to. The community is very much aware that that is an issue.
Bostjan Kobe – I don't know if people are aware that in macromolecular crystallography, the crystal usually consists of at least 40 per cent of solvent, or sometimes up to 90 per cent of solvent. So in fact the proteins inside the crystal are in a very similar environment to the one they would be in in the cell. I would not say this was exactly solid-state; it is somewhere in between.
The other issue is that, especially when you look at complexes, crystallography cannot unambiguously tell you how complexes form. In a crystal you have a three-dimensional array of proteins interacting with each other. So usually from the interface size you can tell, 'Okay, this is the real interface from what should happen in solution, and the other ones are just happening in the crystal but not in solution.' That is why our techniques are very complementary. Also, crystallography has to be complemented with other biophysical techniques, looking at interactions, or with some mutagenesis data and things like that, to unambiguously tell how interaction is formed.
Question – They are two very elegant approaches. I have two very quick questions for Joel. The zinc finger domain serves as a host for the graft, but does the hybrid protein actually preserve any of the zinc finger function? That is the first question. Secondly, just building on two questions ago: can other cations substitute for zinc in maintaining the structure of the zinc fingers, like calcium, magnesium, et cetera?
Joel Mackay – To answer the first question, about the domain that we use: because we made all these mutations to alanine we had basically stripped off any function of that domain that there might have been. So as far as we are aware, if you just introduced it into a cell it would have no wild-type function. The only function it will have, hopefully, is the one that we have added on to it.
To answer the second question: yes, other metals can substitute often for zinc. Things like cadmium or cobalt, Fe 2 and sometimes nickel are potentially capable of stabilising these structures in the same way as zinc has. As I say, in many cases people really have no idea what the natural metal is for a particular zinc finger protein.
Session 5: New approaches to the evolution of complex biological systems
Chair: Associate Professor Christopher Daniels
Speakers: Dr Sandra Orgeig and Professor Hugh Possingham
Chris Daniels – One of the things that Sandra and Hugh wanted to do was to identify similarities and differences between their approaches in looking at evolution of complex systems. It certainly occurred to me, listening to both talks, that there are at least two similarities. Both are taking multidisciplinary approaches – and I think multidisciplinary has been the core of many of the sessions at this meeting – and, secondly, both talks focus on variation. Variation is always at the heart of any evolutionary process. Perhaps the biggest difference is looking at and using different models to look at different sorts of questions. With those sorts of things in mind, perhaps we will open up for questions from the floor.
Question – It seems to me, Hugh, that you are data-starved. Is it viable to break up your kangaroo management into 100-kilometre blocks and harvest at 50 per cent, to see how that area responds? Or is there too much movement in an area that small?
Hugh Possingham – That is a good idea and would be an excellent application of the idea of active adaptive management. That is what we have to do with natural resource management, and yes, I think we should. The trouble is that in all the things like, let's say, water reform and salinity, people building corridors for wildlife. Everybody wants to do the same thing at a particular point in time because that is current best practice. Somebody says in wildlife management, 'Build corridors,' so for 10 years we build wildlife corridors. And nobody actually knows whether or not that was better than making patches of habitat bigger, because nobody did the other thing – they all did the same thing.
How do you convince, at a high level, at a total state-property-government level, different catchment groups, landcare groups and such people to do different things on the scale that we need them to do it so we can learn? Basically we would have to say to people, 'We actually don't know how this works. Can you all do different things?' from which there would be social and political consequences as well. I have been trying to convince people to do this for a long time, and I am not getting very far.
Question – The geneticist-in-training gets drummed into them the adage, 'Treasure your exceptions, because you really learn from your exceptions.' A question to Hugh is: to what extent do you focus on those aberrations in your data, and really learn from what is happening there, compared with putting that data in perhaps a too-hard basket or putting it in an outlier basket and focusing on the overall data fit to models?
Hugh Possingham – The exception there was the 1981 kangaroo data, which has told us that our models have to have some large-scale movement, when all the data says that they don't. So yes, the exception is important. The other thing is that the problem is time. Droughts are important. We had one drought and now we have just possibly had another one, so we effectively get two data points instead of 22, if you really want to understand what is happening under poor resource conditions, which may be the most important thing in town. It may be irrelevant what we do with kangaroos. So we are data-starved, yes.
Question – Hugh, my question concerns some of the evolutionary aspects of what you are doing. On the one hand, if I understood you correctly, the protagonists are saying that the hunting of the Alpha males is going to have this negative impact, and what underpins that is the assumption that size and fitness are somehow related, and that that relationship is positive. However, you also mentioned that, to make your models work, you had to have a trade-off between size and survival, which is a negative relationship between size and fitness. There are two diametrically opposed views here. What do the data suggest?
Hugh Possingham – We don't know the genetics. We have no idea. And even though, again, people have been working on the social structure of kangaroos for a long time, we don't really know how good these Alpha males get, how many matings they really get.
Question – That is the bottom line, isn't it? You could both be wrong.
Hugh Possingham – Oh yes. We don't even say we are right. We are just saying, 'We don't think you're right. We don't know that we are right either.' We have no idea.
Question – Hugh, my question relates to the fascinating idea of kangaroos eating sheep. You suggested that maybe farmers know more about growing grass than just rainfall. The obvious things are evaporation rates, wind speeds and seasonal distribution, that sort of thing. No doubt you have thought of it, but I am just wondering how difficult that would be to model, and if you have any preliminary comments on that.
Hugh Possingham – I don't particularly. There are some wonderful models of pasture. For example, departments of primary industry and CSIRO build these detailed pasture models that have lots of other stuff in them, but they are hard to get the data for, they are very mechanistic, and they might take a long time to run. We have been looking at AussiGRASS as providing the information, rather than using this vague surrogate of rainfall, and use that to drive the modelling. Maybe then the kangaroos will stop eating sheep, which would be good.
Question – I did some marine ecology at Sydney University and did Tony Underwood's course about 20 years ago. At least as far as I remember, he had plenty of data, buckets of data, and did replicated experiments, and everyone was doing them. At that time, as I understood it, despite his overload of data, he still came up often with the conclusion that there was a random event occurring and he could not model what was happening on the intertidal rock platform. So in your data-starved experiment you might have to look at that reality, that you may never get a good model.
Hugh Possingham – Yes. The trick is, I suppose, that people managing our environment want answers now. They want answers to questions like, 'How much native vegetation should you retain – 30 per cent, 40 per cent, 50 per cent? How much of the Great Barrier Reef should have been reserved – 32 per cent, 5 per cent, 100 per cent?' And if you just say, 'Give us another hundred years of data,' or, 'Give us another $10 million and we'll do these replicated trials on a large scale,' then you're not at the table. That is why we have gone to this. Rather than null-hypothesis testing we have said, 'This is our best bet,' and we are just very honest about how stupid we really are and how little we know. We try and point out that our understanding of the world will change. But I'd rather be at the table and say something, than saying, 'We've failed.'
Question – Sandra, I was very interested to hear that you said that the processes that you are looking at pre-dated the formation of the lung. If that is so, I am just wondering whether you can do the sort of thing that you are doing, looking at external factors in the development of the lung surfactants, by looking at either the type II cell system itself in tissue culture or something of that kind, or even if there are other models, more primitive models, which might show the same sort of phenomena.
Sandra Orgeig – When I say that the surfactant system pre-dates the evolution of lungs, we actually have some evidence that it exists within the gut, that the equivalent of the type II cell and the surfactant proteins actually existed in the gut and migrated, with the evolution of lungs from the pharynx, to the lung. So perhaps we can do some sort of modelling or examination there. Did you have anything specific in mind?
Question – I was wondering whether you can get a simpler system than using the whole animal or the whole embryo, which would show the same sort of, as it were, pharmacological or chemical phenomena.
Sandra Orgeig – Well, we also have discovered that it is in invertebrates. That may be a simpler system. We have discovered type II cells and surfactant in snails. So those may be simpler systems, yes. That is potentially possible.
Hugh Possingham – It seems, in terms of data starvation, that the fact the world only evolved one biota once makes life also equally challenging.
Question – Hugh, have people considered trying explicit n-body models for the kangaroo population? You mentioned that spatial effects might be important, so have people tried to lay down a geographical area in the computer, put down a million or say 100,000 kangaroos in that area, model in some very simple way the interaction between kangaroos and their habitat and so on, and then compare the results of that model with the more global models that you were describing?
Hugh Possingham – Yes. We do that end of the modelling as well, where we build fairly complicated individual-based simulations, or agent-based models like that. They are challenging, because you have 300 parameters in there – or a large number of parameters – and you have got to then work out what they all are. So you have got to guess. But we do do that. There is a philosophy of how you can then interrogate those complicated simulations and see whether, even with a little data, the mechanisms you have put in there are plausible. So yes, we have started going from the other end, at different scales, whether it is an individual-based scale or just going down to, say, a property scale or a square-kilometre scale. It is challenging, though, because of animal movement. Populations are birth and death and movement. We have great data on birth, we have a bit of data on death. Find a population of any large organism in the world where we have good data on movement, though.
Question – Even if you do not know the parameters exactly, you might be able to consider ranges, and see what happens statistically.
Hugh Possingham – We do. That is the kind of approach we use there.
Session 6: Utilising large astronomical datasets
Chair: Associate Professor Rachel Webster
Speakers: Dr Brian Schmidt and Dr David Barnes
Rachel Webster – I am just going to start off by asking David and Brian a question. I want to pick up on the last of the ideas that we are throwing out here, which is where we, I think, link in with many other projects. This is on data, and how we store, manage and handle data. The question I want to ask is: who owns data? Take this at any level you like.
David Barnes – Well, within Australia, pretty much all of our telescopes are national research facilities. That means that anyone here, or any member of the public for that matter, can apply for time on the telescope and get some data. Typically, you 'own' that data for a period of 18 months, or you have exclusive access to that data for 18 months or something like that, and beyond that the data will be made available to anyone who wants it. The Virtual Observatory will try to make that available in a uniform and consistent way. With surveys, I guess, they are much bigger teams of people and they generally operate under the same constraints: they will do their survey, they will probably publish some key science papers from that survey and then publish a catalogue and make the data available. That is the general approach within Australia.
Brian Schmidt – The normal paradigm is that to be funded to do big things, part of your funding requirement – enforced by other astronomers, who want to keep the field as unproprietary as possible – forces you to say you will release the data in a certain form within a certain amount of time. So at this point, as you say, you get exclusivity for a period negotiable, and then it is open slather.
Rachel Webster – I want to highlight that this is a really important area, because, I think, in other disciplines, unlike astronomy, there is actually a commercial value on data. So we start with a slight disadvantage over some areas, in that – as far as we know, anyway – we haven't actually figured out a way to make money out of our data, and that is clearly not true in other areas. So I think there are other issues that come into play.
Question – David, you are talking about the grid concept and advertising it very strongly here, and you showed a picture of all the various supercomputer networks that I know about, at least, in Australia, being connected. I also happen to know that most of those things are running literally full time: they have jobs running in a queue and going through all the time. So it does not seem immediately clear why connecting them in any way increases your total amount of computing resource. So I am presuming that in fact, in order to gain more resource, you have to do something like considering a SETI-style program, and consider that a sort of proto-grid. Then you are also going to have more resource issues, in that there is only going to be a certain number of people who can control something that is at that level of distributed, and so you are going to run back into applying for time on large computer systems again.
David Barnes – Yes, that is a very accurate assessment of the situation. Within that diagram I showed, there are institutes where they are 100 per cent utilised and they are probably three oversubscribed in their proposal system for computer time. There are also systems in that diagram that are actually run and used by astronomy groups, and in general the experience so far within Australia is that there is a relatively sharing attitude towards those resources, especially where large projects and, in fact, student projects are concerned. Swinburne is a classic example of that. They have a supercomputer facility that is basically available for most Australian astronomy students to use, and for postdoc researchers and faculty to use by negotiation.
Any survey project that is going to require big computational resources will still have to think about that. At the moment, this is very much a model for how a grid might be built, and you will find that at APAC, VPAC and CSIRO CMIS they do in fact have small clusters that are available – they are basically there, their prime purpose in life is for testing in an environment like this. But sure, the hardware still has to be bought at some level and available.
Question – This is a question for both of you. I think you mentioned that the size of the datasets is something like 10 billion points, 100 parameters – something like that was mentioned. I was just wondering: are there suitable data mining algorithms in existence to handle that amount of data and to search through it in plausible amounts of time to look for correlations, cluster analysis, principal component analysis, that kind of thing?
Brian Schmidt – Well, yes and no. It depends on how sophisticated you want to be. I would say that if you do something relatively straightforward, like a PCA, it is not that far. But if you want to do some non‑linear analysis or something, then it becomes very computational. And so part of the idea of the grid is to share some of the computational resources – which you will have to apply for, presumably, but at least it is a framework where I can apply for time on these six machines and seamlessly use them together, rather than right now where I have to work on them all independently and spread myself six times thinner. There is a lot of work. Microsoft are very interested in this problem and they are using astronomy as one of their tools, because it is described by them as just about the largest free set of data around. It does have commercial interest, and so part of the goal is to develop these tools. But they are not fully formulated.
Question – You have got an orders of magnitude different problem from ours, I concede that. But I would be interested to know about computer resources that we tend to underestimate, which is the number of computers that are sitting around stagnant most of the time. For example, in our building we must have about 150 computers. Most people go home at night, in the marine world, and only the modellers really have their computers running overnight. What we have done is to put Condor on our system, which means that we now have all the computers running when people are not actually using them. So you would be amazed that PAs' computers are now actually running programs. What is the possibility of using those sorts of resources over time?
David Barnes – There is certainly a precedent for that in astronomy. It has been done at CSIRO, in fact. Typically, even when you break your problem up, it is no longer efficient or worthwhile to break it up to a size that is small enough to fit on a typical desktop PC. For example, you might have a problem where you need to be Fourier transforming a certain size of data, all at once, and often that simply won't fit in the memory of a desktop PC. The combined computational power of desktop PCs is significant. The other resources you need, such as memory and hard disc, and often a fast network interconnect between the different machines, can really hold you back in that regard. But that said, the SETI@home experiment is a classic example of that work. It has done the most processing of any computer in the world.
Brian Schmidt – These computers are very heterogeneous, so you cannot rely on the structure. Sometimes they have very slow links. There are problems where you can parallelise by dividing the problem into nice, very small ones very easily. It makes sense there. But for a lot of the really big questions it is definitely useful to have it in as big hunks as you can, just because the hardware is pretty cheap but the software to try to deal with those heterogeneous systems and the people power it takes to program it, those will kill you. The biggest dataset right now in astronomy that I know of is the Sloan dataset. They got their hardware costs to within about 10 per cent, but they missed their software cost by a factor of 100. So it turned out the software cost for them was way more than their hardware cost, and they expected it to be a tenth. You have got to be careful that way.
Session 7: Quantum computing
Chair: Professor Bob Clark
Speakers: Dr Howard Wiseman and Dr Alex Hamilton
Bob Clark – I would just make the comment, before throwing it open to the floor, that the two-atom devices have actually been built. In fact, the signal or the coupling to those two atoms should be very similar to that of a cluster. So it is not such a big step.
Question – Congratulations. These are great results. A question that strikes me when I look at Howard's predictions and your somewhat more cautious comments about predicting the future is that your results were taken at very, very low temperatures. How do you see the future of these computers in terms of their operating environments?
Alex Hamilton – One of the reasons we use low temperatures is that it is much easier to see quantum behaviour at low temperatures. If you can make devices small enough, quantum effects will persist to higher temperatures. But I think ultimately, for solid state devices at least, many of the implementations will require low-temperature operation. That may mean that you don't have one in your laptop, but there is no reason not to envisage, say, server farms of quantum computers in the future. For a supercomputer, I don't carry it round in my laptop; I just hook up to the one in Canberra or whatever. And so for a super quantum computer one might envisage something similar happening.
Question – I have a question regarding the read-out. You are measuring single electron motions, but I think the computation is done in the nuclear spin. Can you say anything about the transfer of the information from the nuclear spin to the electronic spin?
Alex Hamilton – It is a bit of a convoluted process. The information is stored on the nuclear spin, and the nuclear spin is so long-lived that it would make a very good quantum memory. But to be able to access it, we have no way of detecting what the nuclear spin is doing. So the first thing is you can use an adiabatic passage technique to transfer the nuclear spin information onto its surrounding electron, the electron that is circling that phosphorus donor, through the hyperfine interaction.
Then you just have to measure what the state of the single electron spin is. And again that is very difficult to do. It has got a very small magnetic moment; it is really pushing technology to try and detect that magnetism. So then what we use is this parallel exclusion idea. We will define this phosphorus spin to be our reference. We will say this is up, and we want to know whether this other one is up or down. Then we have this trick that if the two are the same we can't move them onto the same thing.
Question – Yes, that part I understand. I just wasn't clear on how you transferred the nuclear information to the first spin, to the electron that is with that atom. So that has been standardly done, has it?
Alex Hamilton – The proposal is there, and it is based on what we know very well about the hyperfine interaction in silicon, which has been studied since the 1950s. But no-one has yet actually detected a single spin in the solid state. You can do quantum computing with an electron spin. It has a lifetime somewhere between the 1018 seconds of the nuclear spin and the much shorter time of the charge. And electron spins you can also do quantum computing with, and there is a whole series of proposals to use those in silicon again.
Silicon is really a very good material for using spin, because silicon has no spin itself. It is basically a neutral host that just holds the phosphorus atoms in place. If you use other materials like gallium arsenide, the so-called material of the future, both gallium and arsenic have got nuclear spins, and those tend to decohere the nuclear electron spins.
Question – I have a question not on the science but on the race. You say that you are one of two groups leading this area. Who is the competition, and what is the edge that you think you have?
Alex Hamilton – What I actually said was that we were one of two groups that have developed this twin radiofrequency SET detection technology. There are five groups in the world that have these very high-frequency sensitive electrometers. UNSW is one of the only five.
In terms of who is doing what, silicon is mainly being pursued at the University of New South Wales. There are also efforts in Berkeley – we have collaborators at their Laboratory for Physical Sciences; the University of Cambridge is also working on silicon; and Ohio State also, with Tucker, I think, is working on that. So there are a number of groups working in each different area, but by far Australia has the biggest concerted effort with both fabrication technology, theory and measurement all bearing down in a focused effort.
Bob Clark – I think the edge is that you need the ability to place a single atom where you want it, and as far as we know, this is the only technology that has come out so far.
Howard Wiseman – In the optics experiments as well, which I mentioned in my talk were being done in the University of Queensland, there are two main groups in the world in that race, and the University of Queensland is ahead at the moment in that. So Australia really is at the cutting edge in this thing.
Question – I believe everything that Howard said, so I am really looking forward to being able to purchase one of these things on 2 January 2050. My questions are to Alex. How big will this computer be, how fast will it be, relative to the quick machines that are available today, and am I going to be able to afford one?
Alex Hamilton – To answer in reverse order: I am not sure what your salary will be, but I do know that it will weigh less than 1.5 tons. As to how big it will be, the actual qubits are very small. We are engineering individual atoms and putting them only tens of nanometres apart – so the qubits will be small. But the support electronics – just as with an ordinary computer, where the actual central processor unit is small but there is a lot of stuff that goes round it to support it, to drive it – may be slightly larger. And if it requires cooling, you may require some sort of cooling system as well. But our complete measurement systems are perhaps two or three times my volume.
Question – I guess the crucial issue is if you need a fridge somewhat colder than a Canberra night, though, isn't it?
Alex Hamilton – Yes and no. Those things are not so huge, and they are getting smaller. Research grade systems are about the size of a person, but if you are just using it for one very specific application and you only need to cool down to, say, 1 kelvin or something, they can be made much more compact.
Question – Do you see that you will always be dependent on the plus/minus measurement to detect the meaningful versus the random, or do you think that you will have the spatial relationship to tell you which are the meaningful ones?
Alex Hamilton – I think I would agree, actually. At the moment we are using these two detectors because we are very much in the infancy of doing this. It is the first time that anyone has really been measuring single electron motion between individual phosphorus atoms, and there are a lot of materials issues. But as the materials improve and we learn more about the system, we will get a better idea of what the signature of a correct event is, and what the signature of an incorrect event is. And so with signal processing you can look at the trace and say, 'I think that was probably noise. It had the wrong line shape, it didn't quite look right.' So then you can get away, perhaps, just with using a single detector. But using the two detectors hasn't been so harmful to us at the moment. We still get extremely good sensitivities, and it does build in automatically some redundancy.
Question – This might be a misguided biologist-type question, but I presume that the phosphorus atom can't be an endless source of electrons on one side moving to the other. Does that electron have to relax back to its original phosphorus before it can be used again in another operation?
Alex Hamilton – You're right. Each phosphorus atom only has one electron. That's why in these devices we implanted about 600, so that we could see multiple transfers happening.
When you go on to actually wanting to use a real quantum device, then really you will just have one electron, say, that is being shared. The question is: is the electron over here or is it over here? You never lose it. You keep it contained within the system. So as long as you can make sure that your electrons stay in there, then you don't need a continual source of electrons. It is a closed system, and these detectors, these external transistors, are only detecting it capacitively. There is no direct connection to those electrons.
Question – You discussed several ways of looking at the state so that would be your qubit, the interesting problem of course being that the easy ones to read are the ones that have a very short coherence time. Is it possible to choose a system where you do each operation in parallel 25 times, so that you can statistically get the right answer over your coherence time? You don't need 1018 seconds if you are going to be doing your operations very, very quickly, presumably.
Alex Hamilton – Yes, you can do that. In fact, some of the early demonstrations of charge qubits and superconductors were averaged 106 times, to demonstrate coherent operation. So yes, you can do that. For some things it makes life more difficult in terms of quantum error correction. It is difficult to do quantum error correction if you can't measure and immediately correct the error, if you are averaging – because then you can only correct an average error. But certainly for demonstrating small numbers of qubits, averaging is a very viable technique. That is essentially why the nuclear magnetic resonance computers work very well: you have got a huge number of these molecules floating round in the liquid, and you are averaging over 1012 or 1018 of them.
Question – I am pretty sure this is a dumb biologist's question, but is there any potentially measurable particle which has more than two measurable states? You would not be limited to on or off, you could actually have three different things, say.
Howard Wiseman – Absolutely. And some people have looked at quantum computing in the context of 'q-dits' – which is the term we use, as in d is the number of states. In principle it makes no difference, but I guess it is just easiest to start with the simplest thing, which is a q-bit. All that is important is that you have some finite number of levels or states or whatever, but then for it to scale well you have to have a large number of copies of that, otherwise you are just wasting your time. If you try to do everything in a single system with a large number of states, you can't win. You can basically show that that will never do anything better than you can do with the classical computer.
Question – If you had three states, would that be much more powerful than two states, in terms of building these things?
Howard Wiseman – I think, in terms of the scaling, the number that you would need in the end is not going to be very much different. So it is not going to make a huge difference. Certainly there would be some algorithm which you could show that you could do things with a three-state quantum whatever – a 'q-trit' – which you could not do with a classical 'trit' or whatever, but it is not going to make a huge difference when you are looking at a full-scale computer.
Alex Hamilton – Actually, it turns out that the optimum configuration is three levels. We have just proved that recently. The reason we have started off with two levels is that it is just the simplest thing to do. Sure, later on. But it is actually quite hard in Nature to find things that are only two-level systems. Quite often you find things that rather give three-level systems. So I think in future there is no reason why we will be stuck just with qubits rather than, as Howard says, q-dits. But you still need a collection of identical q-dits.
Bob Clark – My own opinion is that the bottleneck to building a large-style quantum computer is associated with the fact that the computer will have to work in a fault-tolerant mode. There will be errors in the computer, as there are in conventional computers. Quantum error correction can be done; it was one of the breakthrough parts of the theory in quantum computing.
To do fault-tolerant operation you have to encode the qubit in multiple copies of itself, and you have to read these copies to see what has gone wrong without looking at the real qubit you are trying to keep alive, and then apply that correction back to the qubit before it itself decoheres. What that means is that you need very fast, single-shot measurement. That can only be provided in the solid state by a single electron transistor. The most remarkable thing about those measurements Alex showed you is that they are single shot – not averaging, which is what most of the rest of the world are doing. Actually, when you think about scaling this up and the amount of circuitry you will need, probably the most difficult thing to build will be the control chip of the qubit chip, because you have to keep track of all of this error correction that is going on.
So, to answer this question, I think the processor will be about a square centimetre. The actual qubits will be, let's say, 100 atoms in an array. But around that will be all of the conventional electronics, which is essentially your control chip, and that has to operate probably at about 30 GHz. So you won't be using silicon technology for that; you will be using a technology called RSFQ, which is rapid single-flux quantum superconducting technology, combined with silicon CMOS, so we are going to be going into hybrid combinations of technologies.
Solving all of that in the scale-up is where you have to get involved with industry, but where the science is is in that 30-square array of atoms. If you can prove that you can do the entanglement read at single-shot, there is no reason why that can't be done. And that is basically the task.
Session 8: Molecular ecology – integrating genetics with demography to understand population biology
Chair: Professor Richard Frankham
Speakers: Dr Andrew Young and Dr Andrea Taylor
Question – I have a question for Andrea. This is a little bit out of the blue, but I am interested in the state of play in marsupial assisted-breeding technologies. In that context, are other species of wombat capable of carrying a northern hairy-nosed wombat pregnancy? And is that any possible solution to the problem that you face?
Andrea Taylor – Probably the most practical thing that is being looked at at the moment is pouch young transfer. It hasn't been attempted with northern hairy-nosed wombats, but between southern hairy-nosed mothers they can take extremely small pouch young from one mother and put it into the pouch of another who is in the same reproductive phase. They have got good survival rates with those sorts of experiments. So that is probably the most viable thing.
The only trouble is that we don't know anything about the reproductive cycling of northern hairy-nosed wombats yet, so assisted reproduction for this species is a very, very long way away. We just don't even know the most basic parameters. We don't know how synchronised they are, what synchronises them, whether they are synchronised. We don't even know if they have a tight breeding season, or whether it is just spread across the year. We don't have enough data to say. So that is really what is holding that sort of thing up. But I think the pouch young transfer would be the most promising of those sorts of technologies.
Question – If I could interpose a question: Andrea, what is happening as far as the reserve population of the northern hairy-nosed wombat is concerned? Is anything really happening, or not?
Andrea Taylor – They are doing a lot of work, just trying to identify appropriate sites. In fact, Steve Irwin, the crocodile hunter, took someone out from National Parks and said, 'I want to buy this land. Does this look good for wombats?' The woman gave the nod, and he said, 'Right, I'm buying it. Give me a wombat.' It is at that level now, that they are looking for appropriate sites. Then male pouch young might be the first ones that we would take, and perhaps rear them elsewhere and then release them into the new site, but they would do 'soft releases' on Epping Forest first. So there is a lot of work still to go.
Question – But they are not going to move down to Deniliquin, which is a site that Andrea's work has identified as having had a population that went extinct?
Andrea Taylor – Around the end of last century, yes. No, they're not. Probably it's a political/state boundary thing, actually.
Question – You guys use these genetic markers very effectively in small, restricted populations. How do they work in big populations? If you want to understand red kangaroo populations, for example, and mobility, can you do that? Or is it too blunt a tool?
Andrea Taylor – You can. You can just take samples from within a continuous area – for example in the graph that I showed with the genetic distance between sites. You can do that sort of analysis and look at the shape of the curve of genetic distance in relation to geographic distance between sites, and such patterns can allow you to infer the mobility of species. Yes, that sort of work can easily be done. In fact, I think there are some genetic data on red kangaroos that might provide the sort of data that Hugh needs to improve the accuracy of his models.
Andrew Young – There is another way you can go. Obviously, we are not going to try and enumerate or get a genotype for all the individuals in a population, but with a decent-sized sample, if we can get a decent idea of the shape and the allele frequency distribution, then we can start to do assignment testing and say, 'Well, this population has this frequency distribution; this has this one. If we catch an individual somewhere' – if you are unfortunate enough to work on animals that move around – 'which one is it most likely to belong to, based on its genotype?' So it is a do-able thing, and the techniques become more powerful as you rack up the number of markers you can use. It is a function of marker number and allele frequency distribution.
Question – Did the hair DNA analysis provide any direct information about wombat habitation? For example, do you know whether a given wombat used two burrows, or the number of wombats that actually shared a burrow?
Andrea Taylor – Yes, we can get all that sort of information as well. That hair collection was done over a one-week period so we could look at the short-term movements of individuals in that time, but we have collected hairs for the following two years as well, so once we have the data analysed for those we will be able to look at the more long-term positions of animals. We can also compare that back with the trapping database.
Question – Andrew, you showed a great relationship between pollinators and your grevilleas. But have you ever broken up the pollinators into individual species? Some of those birds, for example, are really specific on one bush or at one time of day, or at one time of year. Can you do that? Would you like to do that?
Andrew Young – Yes, you can have a go at it. You can take pollen swabs from the breasts of the birds, and actually see what species of pollen they are carrying, and then if we can – and a few labs have had a go at this but we have not been very successful so far. So if we can swab the pollen off the bird and look at a range of different species, and then take that pollen load, divide it up by plant species and start genotyping them, it is a doable thing. It is a case of matching how much power you use to whether you really need to know in terms of the functioning of the system. But it is a really interesting biological question.
Question – Andrew, you showed those areas of land in the brigalow that are being cleared. Are you people, or people like you, thinking of ever getting together to provide a stronger argument for why that sort of thing should be highly limited?
Andrew Young – I must admit I was a bit biased: I put all Australian land clearing into Queensland. It was not quite fair.
Andrea Taylor – We were actually at a conference two weeks ago in Sydney that was dealing with this question. It was about half ecologists and half policy makers. There was quite a lot of pressure on the ecologists to come up with numbers and answers that the managers needed to take away with them. We have a lot of sympathy with that, obviously, but I think we will be able to answer that more when we look at more data. Today the important thing is the point I tried to get across with the gliders, which is that we may have enough biological information now on some species to accurately predict what is going to happen, but I think I showed fairly well that we don't necessarily know for all species. So we do probably, for a lot of species, have to still do some measurements. But hopefully we can do it fairly rapidly nowadays. We are not going to have big time lags before we can provide the information. But the whole technology has only, really, in the last couple of years got to the stage where we can do these sorts of things and show lots of case studies and really get a body of data happening. It has not been possible until now.
Question – Andrea, I think it was the gliders that were surrounded by the pine forests, as you might say. You showed they could move up to 7 km, but then, of all those little sites, I think you said 43 per cent had no gliders – some quite large proportion had none.
Andrea Taylor – Yes, 60 per cent had no gliders.
Question – Does that mean they can only move to another site that has already got some gliders? If that is true, that is probably important.
Andrea Taylor – I don't know. That particular sub-adult female clearly had moved some time very recently, because she was only a very young animal. But we don't know for some of the other movements. I'm probably not even answering your question here.
Question – If they can move, why aren't they at all the sites?
Andrea Taylor – Well, powerful owls can come through, for example, and wipe out an entire patch population in a few nights. We are really only talking small numbers of animals in each site, anyway. It wouldn't take much to wipe out a population. Or if they all happen to be males, that is not a persistently long-term population. So there is lots of stochasticity in the system.
Question – Andrea, the dispersal of female northern hairy-nosed wombats out of Epping Forest leads to the loss of the individual to the population. Does it lead to the restriction of genetic diversity?
Andrea Taylor – I guess that is a two-part question, the first part being whether there are females leaving. I would love to know if that is why there are hardly any females left. Are they all wandering off and hoping to find somewhere to end up, but just not ending up anywhere? I would like to know whether that is the case. But having the dingo fence round now might answer that for us in a few years' time. If the females are actually leaving, that should put a stop to it. That is one reason I was glad to see the dingo fence go up, so we could test that.
The other issue is genetic diversity. Yes, the population does have quite severely depleted genetic diversity, in terms of those neutral markers. But we have no idea what effect that is likely to have on their viability.
Question – Andrew, you were doing computer modelling. How do you compare the modelling you were doing with that that Hugh Possingham was talking about?
Andrew Young – We are coming at the same question, I guess, from different ends at the moment. But I think the real power is to pick your model for the situation you have. The situation we have in this case is a reasonably small spatial scale for a lot of the organisms we are dealing with, and reasonably low numbers of individuals, and an ability to get quite a lot of detailed information on both life history and, using the markers, dispersal and pollination. So we really can build those fairly integrated, spatially explicit, individual-based models that give us really high predictive power. And luckily we are in a situation where we can actually test that – for instance, the S allele information. We think we need at least 50 S alleles, so one of the things we are doing, not this season but the following one, is to actually plant out arrays of plants in populations that have different numbers of S alleles and check that we are right: are we doing this properly?
So I think it is 'horses for courses'. If you are trying to make reasonably detailed predictions about the behaviour of a species in a landscape which is fairly spatially defined, then maybe the more detailed, stochastic models are the way to go. But if we are trying to predict how large suites of organisms work over really large scales, then the broader models are the thing to do.
Panel discussion
Chair: Dr Paul Willis
Jim Peacock – At the start of this symposium I promised that these gifted young scientists would explain what they do and why, and during this process, they would come to discover how an idea can bridge disciplines. Now, has it really happened?
What we have on our panel is a couple of the gifted young scientists and a couple of gifted older scientists, I would say, who also participated in the program. More importantly, we have the Wild Man of the ABC, Paul Willis, here to examine them and determine whether they really did get anything out of this symposium in the way that we had hoped.
Paul was asking me whether I was going to introduce him, and I started to say I was, but I didn't get a word in edgeways. So I am not going to try and do too much now. He did tell me that he was a palaeontologist and worked on long-dead crocodiles, and I think he indicated that he should be in the Academy because he certainly was the world's expert on long-dead crocodiles. He said the best thing to do is to get into a field where there is no-one else and you're the world's expert: you're in. I guess now in the ABC he is more often dealing with live crocodile-like organisms, both within and outside the ABC. He has survived, quite obviously, and I am really looking forward to his examination of our panel.
Paul Willis – Thanks very much, Jim, for stealing my best jokes – and mucking them up in the process. No; thank you very much. It is a pleasure to be invited along here today. Unfortunately, I haven't been able to be here for the whole conference, although I would love to have been. It sounds like you guys have been covering some incredible stuff. Looking through the abstracts here, I think you have all been exposed to some wonderful stimuli. To round off this particular conference I have been asked to run a forum with the four selected panellists, who I will get to in just a moment. It is not a multiple-choice examination, you'll be pleased to know. I'll be taking any answer as acceptable, and you won't even be graded at the end of it.
The idea is that we have had a couple of days of very specialist talks, of people talking about the great diversity of specialties that there are in science in Australia. And the real benefit of bringing together so many diverse disciplines under the one roof here is to cross-fertilise those ideas, to swap ideas around. It has often been claimed, 'What would Einstein's physics have been like if he wasn't a really bad violinist?' How much does playing violin really badly help you understand physics and quantum theory? By the same token, I hope that as this forum progresses it will be interesting to hear from you, the panellists, on what you have gained from this symposium.
It's a wonderful world that we live in. The main journals that we have in science these days are Science and Nature. I have yet to meet anyone who can read any given copy of either of those journals from cover to cover and understand everything inside – particularly the adverts. You have only got to get about two or three ads in, and most scientists start developing a nervous tic. It is because science has become so specialised. We are celebrating the diversity of science here, and the crossing over of those disciplines.
By way of brief introduction, in case you haven't already met them: we have here Hugh Possingham, from Queensland University, Bob Clark, Marcela Bilek and Wayne Gerlach. I would like to go across the panel one by one and ask you what you have gained from this particular conference.
Hugh, you work in ecology, looking at kangaroo populations and how to deal with kangaroo populations. What was the presentation that you heard over the last couple of days that really stood out in your mind because it was so left of field for you?
Hugh Possingham – I think I learnt a lot from lots of presentations. In fact, in terms of cross-disciplinary things I think we tend to steal things from economists, believe it or not.
Paul Willis – Ooer.
Hugh Possingham – Whenever I say that, people say, 'Well then, your models don't work either, do they.' But I am actually more interested in how people think. I was intrigued by the astrophysics perspective of these networks, that there arethree to four thousand of these people in the world and they do have international collaboration and they do have this version of a grid, where they will all share data and do things together. That started to happen a little bit in ecology, but we are a little bit too wary of it. Probably there are 25,000 of us. I am intrigued how they have achieved such international cooperation that they can build big telescopes. I wonder, as somebody who just analyses and synthesises data, how we can in ecology make all this data available – taxonomic data, time series data, climate data – so anybody could go onto the Internet, any time, and go onto this grid and do stuff, and talk and collaborate on that stuff, without people feeling that things were going to be stolen and taken. So to me that was one thing that will change how I want to operate in science.
Paul Willis – So you have been inspired to hear how the astronomers operate?
Hugh Possingham – Definitely: the fact that they can do it, and the fact that we need to do it. The point of my talk was that there is bugger-all ecological data, and if you don't get a lot from lots of places you will never understand anything.
Paul Willis – Come on, don't be shy, there must be an astronomer in the audience. There have been astronomers here for the last few days. So, quick, hands up. See, they are shy! Did you guys pick up anything from Hugh's presentation on kangaroo populations? They're using economics models. You can imagine the J curve working on kangaroo numbers, right? Did you pick up anything from the ecologists?
Rachel Webster – Yes, it's great stuff. Oone of the things I think astronomers get excited by is seeing mathematics that they can relate to and understand, and problems that are actually real problems, which is what we are pretty interested in. So the mathematical modelling I think really caught the imagination of the guys here.
Paul Willis – Hugh said that he is going to take away with him your way of interrelating with other astronomers and try and apply that to ecologists. Is there anything you have taken away from Hugh that you think, 'Oh, that's a good idea'?
Rachel Webster – I'm not sure that the specifics of the modelling are going to actually apply to any of the systems that we work with. Andrew, do you want to comment?
Andrew Melatos – Well, I think we are very interested in complex systems, where you have many independently interacting agents, and so just the general idea of trying to map the results of the mathematical modelling to even small amounts of data where you don't have access to the full story, and how you can try and see if the template carries across. I think the lessons there are really important as well, for us. Although we have a lot of data, we don't have as much as we would like in almost any case, I would say.
Paul Willis – Have you ever met a scientist who said they had enough data?
Andrew Melatos – [Laughs] Of course.
Paul Willis – Thank you very much. Let's move along to Bob Clark now. Bob, you're into quantum computing, but you were telling me earlier that something that you were finding incredibly interesting was northern hairy-nosed wombats. Why is a quantum computer nut getting inspired by northern hairy-nosed wombats?
Bob Clark – Well yes, I was inspired by the talk. I would just like to say before I specifically answer the question that the thing that has really excited me is actually listening to a group of young people who are actually at the bench, as it were, in science in Australia, talking about such a diverse area of science, where we can all learn from each other. I think that's got to say that science in Australia is alive and well. That's the message that I am getting straight off from this meeting, and I think if we have this meeting more often, then we are going to ensure that it is going to be alive and well.
The thing that has got me, the general theme – and the wombat certainly fits in to this – is that people who are studying this broad range of things are actually all coming back down to the very small level of atoms, molecules, supermolecules, DNA, cells, to really understand in a quantitative way what is going on. So, for example, tracking down the great northern hairy-nosed wombat murder mystery was done through DNA. Other people are talking about how they can manipulate DNA in new ways, or the missing mechanism in the chain that has been lost by looking further downstream at what might be going on, to actually inhibit processes that are bad for people, or design new drugs, et cetera.
The way a cell works so efficiently in its manufacturing process is something that we who are working in semiconductors, which are of course not alive, would like to mimic. There is this whole field of biomimetic engineering, where we learn what is going on in biology, which is the most efficient way of doing things by self-assembly, how life is self-assembled, and we ask ourselves, 'How can we use that to learn how to make better machines? In electronics or computing how do we track down things using this technology?'
So for me a lot of things have come together. There is a well-known quote of Richard Feynman's, 'There is plenty of room at the bottom.' We are all sort of converging on the bottom end of what is going on at the atomic level, where you are starting to answer the science in terms of how the world really works. The world works from that level, and I think just about everybody here is down at that level trying to understand their systems, whether they be big wombats, tracking down mysteries, or drug design or building quantum computers, or working out why dinosaurs are extinct. It's been just an amazing experience for me.
Paul Willis – Just moving along: Marcela, you are actually a plasma physicist. What did you make of the papers on the geochemistry of extinctions? I couldn't imagine a subject further away yet still within the bounds of science.
Marcela Bilek – I have to admit that for the ones that were particularly far away – you have, I think, picked one of the furthest – the difficulties that I had were quite often with the terminology. I think in a lot of cases that was made a lot easier, in that the speakers were prepared beforehand and they made an effort to reduce the lingo. I think that's maybe something that is very important. Even if no interdisciplinary collaborations came out of this, as I'm sure they will but even if they didn't, it's the fact that we are learning to communicate in such a way that we can understand each other. I can say that in my own collaborations – and I have started collaborating now with people who work on proteins – it has taken me two years to understand what they are saying. So I think there is a real role for learning to communicate without the lingo, and to express what is going on in our fields in a way that people can understand who are from different fields.
Paul Willis – That would be interesting to know from the audience. Just by way of straw poll: hands up if any time over the last couple of days you felt completely bamboozled by someone else's jargon. I thought that would be the answer. Hands up if you had no problem at all with everybody else's jargon. Good, there's no smart-arses in the audience! But surely it touches on a bigger area, and an area close to my heart, the idea of communicating science. Has that actually been a lesson for members of the audience about communicating your ideas, about discovering where traps about what you are trying to convey are hidden in jargon? Does anyone want to talk on that? Don't be shy. What has been your experience with the jargon?
Vaughan Monamy – I don't know whether I wanted or just expected a kind of dumbing-down of terminology. Marcela says that she was a little bit lost by other people's terminology. I've got to tell you, Marcela, you went so far over my head in your introductory talk. My field is wildlife ecology. I know what plasma physics is, but there were acronyms on the PowerPoints that weren't explained and you just had to sort of sit there and be intimidated.
Paul Willis – Does that surprise you, Marcela?
Marcela Bilek – No, it doesn't. There are some things you simply have to include. I mean, the acronyms had to be left up there for the people that do know – they need to know what technique we have used to get the results. But I was trying to present it in a way that, 'Okay, you don't need to know the techniques, this is the result. This is where we are going.' I think you can't be expected to know all the techniques, but you should be able to get a flavour of where this field is going, what are the important points.
Vaughan Monamy – I agree. There was always going to be this problem of jargon and technology coming from years and years of expertise that nobody else in the room has. Over dinner yesterday we were discussing, that maybe the dumbing-down comes when somebody approaches you from a different discipline and says, 'Look, I really think that what you are doing can be applied to my field. Can you walk me through it?' rather than trying to dumb down for such a generalist audience. There are arguments for both.
Paul Willis – So next time you want to accuse Catalyst of dumbing something down, just think twice, all right? It is the commercials that do that, not the ABC. Wayne, just moving on to you: one of the things that you said you are taking home from this conference is micropumps. Tell me your story.
Wayne Gerlach – I guess I'm a bit of a fish out of water here. I am from the world's largest healthcare company, or one of the biggest pharmaceutical corporations – and when you all hiss, that's fine.
Paul Willis – Go on, get it out of your systems!
Wayne Gerlach – It was about 11 years ago that I crossed over from the Peacock empire to the dark side, so I set myself a goal coming here of a couple of mercenary ideas – one from yesterday and one from today. I would challenge you each to think about something like that, impose that discipline upon ourselves.
Yesterday, although I didn't understand everything Marcela said either, I actually liked and thought about something. Johnson & Johnson are going out in the market now with what they call an osmotic minipump. It's like a matchstick made out of stainless whatever, and it has got two holes, one in each end. They load it up with a drug, and you implant it under the skin. By osmosis, biological fluid goes in, pushes on a membrane and slowly pumps the drug out. It is called an osmotic minipump.
I guess yesterday I was thinking about: well, when will the osmotic micropump come? Get it down to a very low size, deliver it into a tissue – not just under the skin wherever you can put the depot – and slowly, constantly release the drug? So that was yesterday's idea.
Today's actually came out of the datasets and all the computing and everything else. Again I didn't fully understand everything. The protein stuff I know a little bit about, but not too much. I was actually thinking about drug discovery, and pharmacophore chemistry, where the pharmacophore chemists think about charged distribution and lipophilicity and so on in space, and protein folding, and when – I guess it's happening, but can it accelerate? – can the protein folding guys get together with the pharmacophore chemists and start to think about proteins as simply pharmacophores: simplify it and get at that spatial distribution. And that will come as well.
So it was just, 'Can you get other ideas way outside your field? It will be interesting to see where they go.'
Paul Willis – When it comes to putting together teams of people to cross-fertilise ideas between the various scientific disciplines, is it better to have a bunch of specialists who all know their stuff and don't necessarily know much about other people's stuff, or do you need some generalists in there who know a bit about everything? Or do you need a mixture? What sort of teams do you put together? Anyone?
Hugh Possingham – It was a powerful Vice-Chancellor who once said to me that he thought the best scientists were specialists. He was talking to me as a mathematical ecologist, so I suppose I was meant to take offence. But my experience has been that as a mathematical ecologist I am good at neither but I am pretty good at translating. So I grab the mathematicians when I need them and say, 'Solve this problem for me.' Then I go to the ecologists and try and work out what their problem means, now I can turn it into mathematics. Really I don't have to do anything at all, Paul, other than translate. My belief is that multidisciplinary sciences are sometimes the glue that keeps a whole multidisciplinary team together, because of the translation issues. I could say maybe my specialty is modelling. I turn biology in words into equations. There are people who solve equations, there are people who do the field stuff. And that's a valid pursuit.
Paul Willis – What's your take, Bob?
Bob Clark – I think it very much depends on exactly what you are doing, to be honest. For example, in our own project we have physicists who have to know the quantum physics as the background of a quantum computer, we have nanotechnologists, if you like, who have to know how to build very small parts of the construction of this computer at the billionths of a metre level, which is sort of a mixture of electrical engineering and physics. Then we have to understand the surface chemistry of what is going on on these devices, because that can kill you. Mathematically, we have to understand how to run algorithms on this, because if we don't know what the algorithm is that we are eventually shooting at, we will design the wrong type of architecture. And then we would like to steal some ideas from biology; occasionally we would like to talk to the biologists and get some very neat ideas that we can think about and try and translate that into our practice.
So the sort of team of people I deal with, all young, are basically specialised in their own fields. But I agree with you: one of the qualities of the team that I work in – and I am just looking from observation – is that these young people are all prepared to take an interest in the other subject. Now, that isn't always true. Sometimes you can get so focused in physics or chemistry that you just go down that tunnel and then you pop up at the end.
But I think what is happening is that in pushing to build these increasingly complex machines that will ultimately give you these greater benefits, the complexity is getting such that no one subject can deal with it. That is recognised by the young people, and we often say in our centre that you can hardly tell the difference between the physicist, the electrical engineer and the chemist. It almost doesn't matter what school they go to; they're all part of the team and we all talk to each other. And I think it's just the preparedness to go outside of your comfort zone so that you can figure out how your piece of the jigsaw fits in to the whole.
That is happening across science, it is happening in Australia, and I think we have seen evidence of it here today. There has been a lot of what you would call interdisciplinary research, where people are talking about physics/chemistry, mathematics/biology, and there is a blurring of what actually that person standing up and giving the talk is.
Paul Willis – It was interesting, a word that you used there – actually, I didn't hear too many hackles being raised – when you mentioned 'stealing' ideas from other areas. I think earlier on, Wayne, you were talking about pinching ideas from other areas. Where does it become collaboration, and where does it become finding someone else to do the work for you?
Wayne Gerlach – Part of that came from the invitation from the Academy to participate in this panel discussion. It said, 'Be ready to be asked the question of where has your field stolen ideas from.' I would say that medicine does not actually 'steal' ideas – it sequesters them, it seconds them and it adapts other ideas. There are plenty of those sorts of examples: the maser in 1954 through to the laser in 1968 – I did my homework in case this exam question came up!
Paul Willis – This is the multiple choice one, by the way.
Wayne Gerlach – That's the one. I've got some more examples: in 1982, in the IBM labs, someone actually thought, 'Hey, rather than using these lasers to etch silicon, we could use them to cut biological tissue and not generate heat.' And that started. That was 1982; 1987 was the first laser eye surgery experiment, and it was not until 1996 that it went out into production. So that is the straight answer.
The topical answer of where do people steal ideas, or take ideas, or adapt them from: you know, we heard one of them here. RNAi comes out of plants and plant virology, and it is coming across into medicine and understanding of biological issues. Probably the most controversial one is one that we all know, DNA. The biologists probably stole that from the X-ray crystallographers, although you can actually look back earlier than that – and there is actually a book that has to be written called Before 1953 – because Schroedinger, who I guess wasn't a biologist although we would like to claim him as a biologist, thought broadly across areas, just as this conference does. In 1946 the Academy of Sciences in the US ran a conference called Borderline Problems in Physics and Biology, bringing scientists together, like this. And Watson, Crick and Wilkins have all said that Schroedinger's thoughts and ideas really influenced them into DNA and so on. That is a pretty wonderful example of the sorts of things you can get out of this cross-disciplinary approach, even forcing it on people.
Paul Willis – It is interesting that you mentioned Schroedinger. We came up with a brilliant idea the other day – so many people don't understand Schroedinger's cat but the concept of Schroedinger's wife, I think, gets the point across exactly – that this mythical creature who can at any one time hold two completely separate opinions on any given subject, and trying to find out which particular opinion they hold at any one time influences that particular state. Marcela, it's your turn to get back at me for such an outrageously sexist comment.
Marcela Bilek – I just want to make one comment before we get off stealing ideas. You can't really steal an idea that is out in the public domain, and I think pretty much everything that was discussed in this conference was. It is actually good that we are sharing ideas. They are out there in the public domain and I think discussing them in an interdisciplinary forum like this just alerts people who would not read those articles to the fact that they are out there and they are usable, to everyone's advantage.
Paul Willis – What I was going to ask, Marcela, just to start with you, is this: are we actually training the right kind of scientists for the future, if we are going to be looking at these sorts of multidisciplinary, cross-disciplinary studies? A standard university undergraduate degree starts out general and becomes more and more specific, the further you go through it. Do we need to still do that for undergraduates, starting to introduce them to ideas and multidisciplinary research after they have got their PhDs, or do we actually start earlier, when they are undergraduates? Whereabouts in education should that whole development process be?
Marcela Bilek – Actually, this conference got me thinking a little bit about what we suggest for our own students. Often in the third year we suggest that our students take mathematics as the other course. The Sydney University degree works in this way: you take four subjects in the first year and narrow it down to three in the second, and two in the third. We have traditionally recommended mathematics as the second subject in the third year, and I am wondering whether that is sensible now, given the way things have evolved. I think the past it happened because a lot of the tools that physicists use have actually come from mathematics. That is where the bulk of them have come from. But now we are beginning to see that there are huge benefits in the cross-over of, say, physics/biology, that Wayne alluded to just now.
So there are, I think, merits in advising our students to do maths up to second year but look at doing something like chemistry or, in fact, biology in their third year. We do have one student who is now doing a PhD – he did a fourth year with us last year – who did that. He had timetabling difficulties, because it is an unusual combination, but he was actually instrumental in beginning our collaboration with the protein group that we are working with, or a number of protein groups.
Paul Willis – How do the rest of the panel feel about how we go about training up the next generation of scientists to be multidisciplinary, cross-disciplinary?
Bob Clark – I would say that actually it is almost happening, because the students know about this already. They are smart enough to work it out. If you combine that with the increasing flexibility of degrees – universities are now trying to position themselves so that students can choose units across a range of what were formerly just full-on subject areas within a faculty, so in science degrees now our students have an enormous selectivity of what they can put together for their degree – I am a little bit torn sometimes.
On the one hand, I am a traditionalist in that the thing I always say to my own kids is, 'It almost doesn't matter what you do at university, as long as you learn how to think logically when you come out of that degree.' The best training you can give people is a rigorous training so that they just learn how to think clearly and how to break down a complex problem and learn how to solve it. So I've got that in the back of my mind, and deeply feel that.
On the other hand, I can see that there are so many exciting things going on that you do need this sort of diversity. But my own feeling is that it is really important to get that good, solid grounding in something. Yes, have these other things around the edges within your degree, such as taking a biology unit if you are a physicist, or a chemistry unit or a mathematics unit, but the real time that people start to bring this to bear is probably at the graduate level. There one of the things that probably we do not do so well in Australia that other countries maybe do is the graduate courses that, say, you would normally do in a United States university as part of your PhD. Our students take graduate courses but it doesn't tend to be as really broad. That is not meant to be a criticism, because I think we have a lot of advantages in what we do over that system as well; it's a sort of a balancing act. But I think the students really have worked this out, and I think by and large the system that we have got is flexible enough to allow them to get this.
Wayne Gerlach – Just on that point: it would never happen, and we've certainly gone away from it, but a three-year degree is only 150 weeks, not very long, and I put the proposition that there would probably be nothing wrong with having a science course where you studied the four basic sciences – maths, physics, chem, biology –
Paul Willis – And geology!
Wayne Gerlach – And palaeontology (that comes later, actually, and that is my point) – for each of those three years, a good, solid grounding, and then later is when things come out. It relates to your question earlier of how do you put together one of these teams that are going to do it. You do it by hiring smart people who have a passionate and intense desire to achieve – those two things. And if they have got a good grounding, they are smart and they have got that passionate, intense desire, you'll get it.
Hugh Possingham – One of the things I want to pick up from what Bob said is that I think it is actually sometimes not what you learn, it's how you learn to think. I think what you learn matters diddly-squat. Having done biochemistry and no ecology didn't really make any difference to me. But I find that if I have a physicist or a statistician come into my lab, the first 10 weeks are traumatic because we are talking about projects and I know what they are saying but I don't know how they are thinking. And eventually my thinking changes. What I have noticed particularly about this meeting in the last two days is how different disciplines have, I think, quite different ways of going about science. Maybe we shouldn't admit that, because the public think there is one way to do science. For example, the astronomers are saying, 'Now the Big Bang says that the world goes out to infinity,' whereas as far as I remember we were all going to contract. And positions have been put by certain people to say, 'This is the way the world is now,' whereas my talk was all about saying, 'Actually, I have no idea, but if you want me to give you an answer this is the answer I'll give you,' accepting that some fundamental differences have been about how you probe data, what your belief system is: is it alternative models, is it null-hypothesis testing, or is this the grand theory? And sometimes, in some sciences, obviously, people get a long way by knocking down the grand theory. So it is really the sociology of the discipline that fascinates me, and there are fundamentally different cultures – that is what I've learnt – about how people go about science.
Paul Willis – While we are talking about the idea of being cross-discipline, so far the discussions have been about cross-discipline within science. Can you see merit in proposing cross-discipline studies that go outside of science, so you start mixing in the arts, you start mixing in law? There is obviously a place for these. Should we be paying more attention to them?
Wayne Gerlach – Applied science does it now. To get science through to application, you need law, you need commercial, you need marketing, you need sales, you need the whole lot to bring it through. And it is hard work from the initial discovery. We're talking about initial discovery a lot here, but there is that whole next level to getting it out there.
Bob Clark – Part of a responsibility, I think, that people are realising is on the scientists, apart from just doing high-quality science for the funding that comes through the various research channels, for example, is that there is an onus that we should be able to explain what we are doing, why we are doing it and why it is important, to the Australian public. This is true of all other countries as well. So one thing that we have been flirting with, with our centre, for example, is actually, through the Arts Faculty, employing an artist-in-residence. We don't mean someone who is going to do a painting on our wall. What we are thinking is: can we get somebody who comes and takes a look at what we are doing as scientists, and puts it into a completely different language so that it can be communicated simply to the general public?
If we are going to carry the day in science in general, and get people to accept, as it is, that if you don't have science in your community you are not going to prosper as a nation, if we want them to accept that and be with us, and the politicians to be with us, we have to sell that message effectively. I would say that, as in all other things, we are probably not very good at that. We have a good shot at it; some people are better than others. But there are really good communicators out there, like in the arts. So, for example, we would very much like somebody who could come up with some sort of video visualisation of the inside of a quantum computer, and why that is working better and why it is going to deliver benefits. It doesn't have to be correct, but it has to get the thrust of it across.
The problem with scientists is that we all glue up and we want it to be correct, and so we always are cautious and reserved about what we will say. We will back off. It has to be accurate, and so we tend to be a bit boring. But if you give that to somebody else who doesn't have that inhibition, they will do that for you. I am starting to feel, myself, that good, effective communication is to have a collaboration with good communicators who do it through different media, rather than the way that scientists would traditionally communicate with the public.
Paul Willis – To take another example of going cross-discipline in a different direction: it is very easy, being Australian scientists all together here talking about the different disciplines that we can combine, to become nationalistic about it, but there are very real benefits, are there not, of international collaborations. All of you, in your CVs, have spent considerable time overseas working in other people's labs. Are we promoting the international side of interdisciplinary science in a way that is going to be effective in the training of the next generation of scientists? Maybe you would like to take that, Marcela.
Marcela Bilek – I think we all realise how important international collaboration is, and I think the fact that just about everybody here is involved in that is crucial within our own disciplines and also crucial in interdisciplinary work. But I think it is being done, simply because people are communicating within their disciplines.
Paul Willis – The sting in the tail, though: is it easy to convince funding bodies that are nationally based of the merit of international collaborations?
Marcela Bilek – It is difficult, because I think – and this is true of every country – no-one wants to fund science done elsewhere. It is very difficult to sell to the taxpayer that your dollars are going to scientists in Germany or so on. So the way it is done at the moment is that each of the collaborating institutions has to source funding for their part of the collaboration. It is possible to get some extra travel funding. It seems to work. The difficulty is, of course, the time scales involved in the two different grants. Somebody may be successful, the other partner not. So it is not ideal, but it is the only way that seems to be possible at the moment.
Paul Willis – Okay, it's time to throw it back out into the audience. By the way, if you have any questions, please raise your hand and let me know. I would like to ask: can anybody in the audience give us examples of cross-disciplinary collaborations that they are currently involved with? Is anybody doing something in that vein at the moment?
David McKenzie – Yes, I am definitely involved in that, and it is a very enjoyable experience. What I am doing, along with Marcela, is that we are interested in the connection between physics and macromolecules. We are interested in how macromolecules interact with their environment, how you can manage them, how you can attract them to surfaces, that kind of thing.
What I picked up from this conference, particularly, is the importance of this so-called junk DNA. That was amazing to me, the fact that we had been told that there is only a little bit of the human genome that is worth anything, and the rest is a lot of rubbish and doesn't do anything, but that's wrong. In fact, in the field that we are working in, which is to do with the effect of electromagnetic radiation on macromolecules, this could be the key to understanding what is happening – the fact that this 'junk' DNA may be the important bit rather than the junk bit. So that is something I have certainly learnt, and I think it is a very valuable lesson.
Paul Willis – Anybody else like to talk about collaborations that they are working on at the moment?
Craig Johnson – One of my areas of interest is evolution, and I have been doing some theoretical work on evolution in complex spatial systems. As in a lot of evolutionary stuff, the theory is a long way in front of the empirical work, and we have recently started some interdisciplinary stuff with people who work in sewage waste-water ponds, so I think you can see it is quite interdisciplinary. The reason that we are going to sewage waste-water ponds is that the bacteria are the important things in those systems, and being bacteria they can evolve very quickly.
These guys have some very well-developed molecular probes to identify particular kinds of bugs; we can actually measure their evolutionary rates, and they form these spatially self-organised flocculates, which is a key part of the process in terms of the evolutionary mechanism. That collaboration is helping them understand about flocculation and evolution in flocculation in sewage waste-water treatment ponds, and I'm getting to answer some very interesting questions about how this kind of evolution might actually occur in the real world, in the carbon-based world as well as in the silicon-based world where we have been running the models.
Paul Willis – Just to play devil's advocate on this one for a moment: if you had been given enough time, though, surely you would have figured out their contribution by yourself, and vice versa. Is that not true? It's all there in the literature; you have just got to go and read it.
Craig Johnson – Well, I know less about microbes than these guys do about ecology. But the bottom line is that there are some very specialised molecular techniques that are required to do this empirical work, and these guys, I think, with any amount of reading, just had not been thinking about these kinds of evolutionary processes that are affecting the community dynamics of the microbial communities that are basically running their systems. These flocculates are crucial to running commercial-level waste-water management facilities, and there is a very crude understanding about how flocculates form, how they evolve and how the formation of flocculates feeds back in to the evolution of the microbial processes.
Paul Willis – I am curious to know where you would publish that research. Is there actually a journal that is generalised enough to be able to encompass both fields of research?
Craig Johnson – There will be publications in two areas, I guess. One will be in the general evolutionary literature, and the applied spin-off from this will go into the specialist waste-water management/engineering literature.
Paul Willis – This is actually another problem with the idea of doing interdisciplinary research. Are the journals, as a means of communicating the science that is being done, able to keep up with the interdisciplinary nature of the research? Usually journals tend to be quite specific in the sorts of things they will publish.
Hugh Possingham – We tend to accept that every time we do something we publish it two or three times – which isn't just to increase our publication rate it is because we have to. If we put some modelling in the Journal of Theoretical Population Biology, which is where it should go, no ecologist will read it. And if we put that modelling in the journal Ecology then the modellers will not read it. So we have just got to translate upwards and downwards, more or less, as Craig says – which, unfortunately, proliferates the literature.
Paul Willis – How do the journal editors feel about the idea that you are publishing the same paper three times?
Hugh Possingham – Oh no, we're not doing that!
Paul Willis – But the same results, essentially.
Hugh Possingham – That's right. In fact, the papers often look completely different, because of the way we spin the argument. Then there are popular science journals. I think half of the stuff we do we could be putting in something like Bioscience, or those sorts of trends-type journals. The savvy scientist is doing it three or four times, in different forms.
Paul Willis – Have any of you encountered problems getting papers published from multidisciplinary research, simply because there is no real forum for it?
Bob Clark – I don't think this is a problem at all. If you think of the accepted high-impact journals in science, Science itself, Nature, in physics Physical Review Letters, for example, they are more than happy – particularly Science and Nature – to take interdisciplinary research. In fact, it is becoming more the norm these days.
Paul Willis – I am getting to the point now where I would like to start bringing this discussion to some kind of summation, and now is the time to really bring in the devil's advocate position. You can say, 'But we've always been doing multidisciplinary research. Scientists have always been working together, they've always been crossing-over ideas.' Is the whole concept of multidisciplinary research just some kind of Californian buzzword that has been foisted on us recently and that we are holding up, or is there something special about the way that we are treating multidisciplinary research these days that has not really been done in the past?
Wayne Gerlach – Well, I think it has always happened. I think that there are plenty of examples, and we have alluded to some of those. Can you act as a catalyst? I think you can, and I think that this sort of meeting does. If from the panel we are allowed to ask questions: Jim, when you mentioned at the start of this that the US has been holding these sorts of symposia for 10 years, have they sent McKinsey in and done the cost-benefit and all that sort of thing? What does the US Academy feel it is getting out of this, as far as you know?
Jim Peacock – I don't think they have gone as multidisciplinary as to send McKinsey in, but they are holding these every year, despite the fact that it was predicted that they would be a miserable failure. They have been just the opposite. I was reading the other day that they have found many examples of people getting together and working together as a consequence of having attended one of these things. So that is what we are hoping, and I think you guys have underlined that it is likely to happen here.
Wayne Gerlach – So they truly can act as a catalyst?
Jim Peacock – Oh, I think so.
Paul Willis – Your take on that, Marcela? Interdisciplinary research: is it overhyped, or is it a new phenomenon?
Marcela Bilek – Well, it's not a new phenomenon. I agree that it has been happening for a long time, and probably more so in the past when there was less volume of scientific literature in each of the fields and there was less distinction. If you go back far enough, at Oxford and Cambridge there was only one degree. So I think it has been happening for a long time. But I think that particularly now, where each of the fields has become so specialised and there is just so much that you have to learn within your own field that it is impossible to be across a lot of disciplines and still really know enough in each area, these sorts of forums are very important. You do need to start doing things in teams, just because of the volumes of the information and the degree of intricacy and the technical detail that you have to be aware of, and so I think that it plays a crucial role in bringing people together in those sorts of teams.
Bob Clark – I would maybe like to come at this slightly differently, to give you an answer to almost a different question. But it is sort of similar. My take on it is that with this convergence of physics/chemistry/biology we are working at similar length scales where is a lot of commonality coming out. And my own feeling is that 20 years ago, when I was studying things in the lab, we sort of studied what was lying around -we worked out its properties in great detail and we published this in journals – and we sort of knew that this would be important for various things. We tended to study things that were there, things that were in Nature. I think what this explosion of getting down to these very small length scales has brought to bear is that we now have worked out that we can interfere with all of those processes and we can actually get in there and change the way that these materials work. By doing something ourselves we can direct what is actually happening down at the bottom end.
So rather than just studying what is out there, we can actually say, 'Well, we know that's out there. We can now bring all this to bear, to actually turn it into something that we want it be, something specific,' like a new drug or a new biosensor or a new coating, or a quantum information processor, for example. I think there is now this opportunity to turn what I call fundamental science into technology, more so than there has been before, for the scientist. There is a real role for the scientist there.
And we are trying to assemble teams to do that. I would say unashamedly that there is nothing wrong with doing mission-oriented research. Providing you keep it in balance with fundamental science, both of those things are quite important for healthy science. And so I think what people are realising is that as you go for these mission-oriented targets, you are going to need all of these experts. So I think that's a bit of a driver.
It relates also to something you said before, about being nationalistic. Again I don't think there is anything wrong with being somewhat nationalistic about your science. I think we all have international collaborations, and long may it be so. We have plenty. But one question I think we have to ask ourselves is: can we bring benefit to Australia – there is nothing wrong with that question – through our science? You know, what I am finding is that this is a very powerful motivation for our young people who are working in this country. We have great collaborations with our international colleagues, but boy, does it make a difference if you think you are going to have an impact on Australia. I think that's really important.
Paul Willis – So, instead of the green-and-gold jersey for Australia, the lab coat is just as much a national motivation?
Bob Clark – Possibly.
Paul Willis – And lastly, Hugh, your thoughts on the future of interdisciplinary research. Where is it going to go?
Hugh Possingham – Well, hopefully somebody will define for me what cross-disciplinary, interdisciplinary and multidisciplinary mean, one day. This is a meaning problem.
Paul Willis – Okay, I've just been using them interchangeably.
Hugh Possingham – Yes, people do use them interchangeably. I think some people have given a sensible definition. I am encouraged, because when I came back to Australia, some people were telling me, 'No decent maths department will have you, Hugh,' and, 'No decent biology department will have you, Hugh,' when both of them in times past have employed me, and I was quite worried. But there has been a message from the top; I think there has been a message from leading scientists in Australia, from wherever, that multidisciplinary science is okay. And that has just come in time for me, to save my life. In the early '90s it was just considered okay; now it is reasonably popular. But if we had not got that message across, (a) I don't think I would have got a job, and (b) I don't think we would be having these meetings. It does give people another opportunity, and that's good.
The other thing I would like to say is that it is really about social networks and network size. How many people can you collaborate with? There is a real danger that if you don't have multidisciplinary teams, even if they are geographically focused and you are forced to talk to somebody in the same university or in the same country, and if we just specialised, then we would all go overseas. We would go to the How Many Hairs are on That Drosophila's Bottom conference, and we would just do that obsessively. That would be a disaster. I think Australia is in a fortunate situation; you can get to know a lot of scientists in your country, outside your discipline, in a finite amount of time. That makes it nice and cosy and comfortable, and there is a peer group; whereas I don't know how they do it in the United States, because it is just too big and too vast. My experience of ecologists in the United States is that they become obsessively specialised and their career is built on one idea. If they can have one good idea, they will beat it to death, whereas in Australia you can actually generalise a bit. You can have some good ideas, and you can broaden out. And that makes it a lot more interesting.
Paul Willis – Well, multidisciplinary, cross-disciplinary, interdisciplinary, call it what you will – and I am sure we can come up with different definitions for them – as long as scientists come together in meetings like this and exchange ideas, I am sure that the future of science in this country will always be vigorous and healthy.
Jim Peacock – Paul, thank you very much. That was a great finish to what turned out to be a wonderful symposium. I think the panellists have spoken for all of you, and have told us that it really has given us everything that we had hoped it might. We will certainly be holding more.
It was very exciting, cutting-edge science, all the way through. I think the cross-discipline interactions have really worked – whether at the dinner, in the sessions or in the coffee didn't really matter. I would just say to you that on your list of registrants you do have all the emails and phone numbers, so you can follow up readily. I hope you do.
I want to thank you all again for making it the success that it has been. You know, to be elected into this Academy you have to do great science, and the demographics unfortunately tell us you have to grow old as well. What I think a meeting like this does – and it is what I think Paul alluded to – is that there are no problems for this country whilst we can know that we have got scientists as good and dedicated as we have heard here.
