Kat - Professor Jan Traas works at the ENS in Lyon, France. He and his team are using computer programmes to understand how flowers grow into their beautiful shapes. I asked him how he manages to turn the complexity of a natural structure like a flower into numbers that can be crunched.
Jan - Well, one of the major keywords is quantitative biology. If you want to use mathematical approaches, modelling approaches, you really need quantified data to go beyond qualitative conclusions, like it's bigger or it's smaller. You would like to know, you want to know how much bigger, how much smaller. That doesn't only include concentrations of proteins, sizes of cells, but also, what are the mechanical properties of these cells? What is the pressure the inside the cells? And this is really challenging.
Kat - So, you take all these measurements, these numbers. Presumably then you just stick them in a big computer and see what happens.
Jan - Well, that's what we originally wanted to do, but we realised very early on that well, modelling is actually a form of simplification. So, it's not just like sticking or putting a huge amount of data in a computer and then just push on the button. You really have to think about the processes in more abstract terms and you really have to think back how you want to simplify them. Simplification is really the keyword when you're doing modelling and when you're doing mathematical modelling. And that's of course extremely difficult to simplify complex things.
Kat - And then presumably, when you've got a mathematical model, you have to go and check that it actually does bear up with the real plants.
Jan - Exactly, yes. So, there's one thing we learned over the last years. I mean, for a while, biologists in our field which is molecular biology, cell biology, we'll tend into think in mainly qualitative terms as I've just said.
Kat - So, what something looks like, how it grows?
Jan - Yeah, just it was big or smaller, or it has changed, but you know, not really going into the details. When you're doing models that actually, all models are wrong and simplification also is necessarily going to be wrong. But it's because you can say where things are wrong in your model, that you can just go further and prove your model.
Kat - Presumably, making these models takes a huge amount of computing power as well.
Jan - Yes. Some of the models that we make currently, physical models in a form of virtual tissue, about a couple of hundred cells. To compute changes in shape can take up to 20 hours on a laptop computer. So, with just one simulation, you find out you didn't put in the right parameter of values. So, you start again, etc, etc. Well, that was really a problem, but our colleagues - computer scientists - they have apparently found other ways which can go from the initial 20 hours to about 5 minutes. These are computer programmes that are also used to train surgeons for instance where they can do sort of in silico operations and see what the effects are like you're pulling on this organ and you see effects on other organs. So, these are called live modelling methods that can also be applied to plants.
Kat - So, in your talk, you showed basically kind of small buds and just little buds growing out. Is it your dream one day to have a model that you can grow an entire flower?
Jan - Yes. Well, that definitely - the long term aim is really to have a virtual flower and to have as much information in there and hypothesis in there that we can. We'll have a flower. We'll also try to couple this model of a 3-dimensional tissue with other models that are at the level of the whole plant, trees, etc. So, the future that I think we're speaking in about 10, 20 years would be to couple all these models from molecular networks to the entire plant or animal, and to be able to go across the scales with these models. But a single model will not be possible, I think.
Kat - Is there a particular flower in mind? Do you have a favourite one that you would particularly like to see in the computer?
Jan - Well, there's many flowers in the computer. I think that, you know, so far our favourite flower is Arabidopsis, but it's not a very interesting one in terms of beauty of the structure. In our laboratory, we're working on some other types of flowers, particularly roses. I think we're also working on petunia for instance which can be beautiful flowers with very complex shapes. I think well, for me, that would be very good to have models running off of one of these flowers - roses and petunias.
Kat - That was Professor Jan Traas from ENS in Lyon.