 I guess initially with some other we were using experimental parameters that we could derive either from Polish work or our own work and then the immersion behaviour came out of that, of the physical model. But as we're moving into plant systems we're more explicitly using, say for example, a good example is with the mycantia, very simple asexual propagules which undergo development of the microscope, you can get very good quantitative data for cell geometries and properties that you can map onto that process and trying to describe geometries in ways that you can classify interactions is important to us, so using graphite type approaches to formalise the description of cell interactions and grouping different patterns I guess. Yeah I think it's quite, it's an early stage in the work but I think that's really important to think of more creative ways of trying to integrate genetic properties and physical properties and dynamics. And by the way I have a comment to make since we are in IHES, there's a group in France actually and in Germany trying to use category theory and we know it's birthplace here trying to model those graphs, how they're more formal to them. How do you put that, especially in plants one must, you have to have a certain cozy of the networks. And you've called them, this is your question. Because it has to, it has to cause the information which you need. Well I think that depends a little bit on the data that you're using and the abstraction, some of it's 2D abstractions of three dimensional data like surface projections and you're, we're making at this point simplifications as fast as we can to try and identify keys or things that are recognisable and to get correlates essentially that we can then explore further. So, because a lot of it is dynamic and it's not obvious by just inspection. So in your, in one of your simulations with the reconstruction of a mechanism of morphogenesis for plants, we'd like to know if you have used the epigenetic information to reconstruct this mechanism or not. The question was about whether epigenetic information has been included in some of the models and I think, in a way, because we're using genetic markers in some cases, which are governed not just by genetic but also epigenetic constraints, so some of those epigenetic information is implicit in what we're measuring but not explicit at this point. But having said that, one of the nice things about mycantia is that it has a very streamlined genome, I don't get into details, but these very primitive plants have very little genetic redundancy. So often, unlike in higher plants where you have large gene families associated with, say, often cell or other specializations, in mycantia they're very streamlined and straightforward and some of the RNA metabolism and other machinery associated with epigenetic control is accessible for mutation. And we have things like CRISPR-Cas9 techniques for knocking genes out or modifying genes directly. So there's a whole range of experiments which are suggesting themselves to start modifying directly the mechanisms for epigenetic modification in ways that might be, so there's an integration of a system where you can measure things directly but also manipulate some of the information speeding into the system, particularly epigenetic information. From the audience? Yes, when you show the video of the growing plants, can you imagine, do you think that there is a principle of like a least action principle when it's growing? So as you say, least action principle. Like variational formulation. Minimization of the... You better define it for the poor biologists. Yeah, question. Like a ray, when you, a light ray, want to minimize both time and energy product together. You know, in physical science, everything can be derived from a purchase or least action principle. I have no idea how to answer that question. There is no reason for that question. Yes, we have to answer this question, why? But the reason we have to answer this question. Because it's like a fine man put every quantum dynamic altogether into least action principle. No, it's only in physics, only in science. And you can do the same anywhere else. Because you can do the same. Who? Because you can do the same in biology. No. Why? Absolutely, there is no reason for this. Well, you could say, if I may, just because this is a personal interest. In the literature, there are some people in France and in the US who have been trying to do so. In France, for example, there is a searcher called Gilbert Chauvet, who's been trying to find a sort of equivalent fundamental principle. And he found something related to self-organization, trying to have a definition highly debatable. Like a living organism, a living complex system is something that basically conserves the service it does to its environment. And by doing so, it's organized itself. So he found something like orbitropy. It's a very close to negentropy from Schrodinger. But it's highly debatable. It's highly, highly, highly debatable. Is more people coming from theoretical? Because actually, yes, I know why I asked the question. Because at some point on your slide, you said at the beginning, it's not only your DNA, but it's not only necessary to know the code. The coding, there is more than the coding. And there are self-signalling, which is important, networks. And then there is also, and you say also, but it's like in economics, political networks. And as if the global behavior is optimizing itself like a list action principle in nature when the storm is going down, it's following the list action. And you don't know why. As if the storm knows exactly what to do. And I was wondering if there's the same in biology. From an epistemological point of view, it would be very elegant. In open US, you can make a potentials. You have mechanical energies. You could use it in some ways. It's done for the hydro. And is your plant something like the hydro of plant? In terms of regeneration, yeah. But just get back to your question. I mean, I think from the biology perspective, there's this issue of just taking a rather pragmatic approach, which is the systems are to be optimized because they're involved systems. So there's an economy there. But I guess what we're really focusing on is trying to get a handle on the mechanisms. Because there are things we can measure, things that we can get our hands on. And the idea that we can grapple with some of the emergent processes and try to find principles or underpin it. So for example, from a pragmatic point of view, we're trying to identify genetic motifs that will play out in a cellular context to create processes which are recognizable, the kind of modules that you see in genetic networks, but played out in a cellular context, for example. Say a module that will create some self-propagating cellular process or create some kind of bifurcation, a branch, or a terminated process. They're the kind of things that we're thinking of now. The question of optimization. I hope that will come out in the end. Like riboswitch, in the nature, you have the systems going back to the gene expression. You can modify gene expression using environment, I believe. And this is on this line, I was asking the question. But again, in physics, everything comes down from the terrarium formulation of invariants. And the real question behind your question is what are the invariants in living systems, which is a massive open question. Now, maybe another question for everybody, not only for Jim. Please do, please do. This is another question for Jim also. OK. You have shown, in the beginning of your presentation, you have shown us live where you had then relationship very direct between chemical information and morphogenesis, the second on the third slide. So you have two cell, and then you say that as implants this cell when don't move too much, there is a direct relationship, and you mark two arrows if you have this cell and the morphogenic development. Can you elaborate more about this? If I recall correctly, it's a diagram which shows a single cell and a time sequence. Exactly. So a cell that might be programmed to divide, and the two cells that are the two daughter cells, then cause a breaking and addition in information in the system. So I guess it's trying to represent that. Oh, it's a representation on this. Absolutely, yeah. It's not the genetic mechanism. Well, I think that there's a relationship. What the slide is trying to represent is the relationship between the genetic information. So you have a field, like in the time lapse image of the picture plant. You have a million cells. They're all isogenic. So they have the same DNA program. And yet there must be some mechanism from starting from a single cell. And as the cells proliferate to elaborate different states of gene expression. And so the breaking of symmetry and the creation of those different states has to be somehow very precisely programmed. And then that, so there's clearly a linkage between the genetic processes which encode that. So I think a lot of people in the genetic field tend to think still of the genome as a blueprint, as that describing the endpoint, rather than describing the trajectory. So all those cells are undergoing different trajectories. And that the relationship that the governing of those trajectories, the splitting of paths in information sense, is governed by cell cell interactions. So that's what that diagram is trying to represent. And I think that, in a way, Turing's ideas, you know, in his morphogenesis-based studies, the kind of simple interactions of noise, diffusion, et cetera, are the kind of things that play here. So the kind of exchange or movement of signals and the separation, the creation of asymmetries by simple processes, but played out in a population are what gives rise to the complexity of the system. I have a question for Anne. About the boundaries of the operon-like structures, we would expect that if you have a pathway that is linear, this could work to some extent. That if you have a branch pathway, then because of the branching, you may have more than one odour or one genomic locus that should own the genes. And therefore, I would like to know if you have looked at the, let's say, the genomic structure from this point of view. That is, if you have a long pathway, but it is linear, it should still be within one. But a shorter one, which is branch, may belong to different genomic loci. Is there any relation between those? That's a very good point. I mean, I think my speculation is that these clustered pathways are insulated from the rest of metabolism. And that's why when we have mutants in those genes, we get a phenotype, because normally, there's redundancy, there's a mess. But these pathways seem to be insulated. And there's a really interesting question around what constitutes a clustered pathway in terms of the starting point. And I used to think that it's very simple. It's a bifurcation with primary metabolism, for example, as I was showing you with the triterpenes. But if you look at the very nice paper from Ian Graham's lab on the noscopene cluster from Poppy, that's a 10 gene cluster that is dedicated to the synthesis of noscopene. I've been drawing a figure for a review where I've got the whole alkaloid pathway in Poppy mapped out. And making noscopene is only one part of that. There's no publicly available poppy genome sequence. But the first step in that pathway is not a branch point with primary metabolism. It's probably the siphoning off from that network. So it'd be really interesting to know if and when the poppy genome sequence is available, where all those genes are, and whether the parts of the, not network, the parts that are dedicated to specific end products form discrete clusters. So it's a really interesting question. So you mentioned that there might be a core expression going on. Is there any common transcription regulator known for these cluster genes? Because I mean, I guess the distance between these genes is too large that one large transcript is formed like in comparison with an awkward. Yeah, so these are not operands. The genes are pretty confident. The genes are independent and transcribed. The intergenic distance is very depending on the genome size. So the bigger the genome, the bigger the intergenic distance, which in itself is interesting. So there's no periodicity that we can see at that level. I didn't mention transcription factors because hardly anything is known for any of the pathways that I've mentioned, which is quite surprising. There's a transcription factor for the rice dieterpene clusters, but it's not specific to the clusters. It also regulates upstream metabolism. There's one very nice example for a cluster from cucumber, which has been reported. But apart from that, we need to know more. We were looking for transcription factors, and so are other people. I should say that in filamentous fungi, which of course are also eukaryotes, and which have lots and lots of gene clusters for natural product synthesis, including penicillin, you quite often, but not always, find a gene for a pathway-specific transcription factor in the cluster. We're not seeing that. Question for Ranz has to follow up with Francois as to the previous video. So you observe this cluster for this particular special metabolite. And as I understood, the T-synthetases are enzymes that might exist in many copies as homo-luxury. Do you observe some kind of sequence similarity on the genomes of monocots? So for the genes within these clusters? With the whole genome. Yeah. I get the whole genome. So certainly, for the genes in the old cluster, we pretty much have a template paper that we use, where, for example, the gene for the first step in the pathway is a distant relative of psychoartinal synthase. They use the same substrate. Similar structure of the compound. Yeah. So that gene has arisen either directly or indirectly from primary metabolism and has then diversified. And all of the genes within our cluster, they all belong to multi-gene families because that's a special type of metabolism is. But they've gone off and done their own thing. They're divergent members. They're generally the founder members of a new subgroup that's, you know, we can find sequences, not the same, not closely related sequences, but phylogenetically a group. They're probably all doing different things. Yeah. So just a follow-up on complementary. So the co-clustering of the synthase with the cytochrome genes, is it a specific example of the monocots? So you observe it only on the grasses and not in the diacons. So in the diacons, you observe some more. So what does it have to do? Is it like the diphenyl for the last slide? But it's co-clustering of all. Yes. I'm not too. Sorry, I'm not quite sure what you're asking. The collocalization of the terpenic synthase with the cytochrome gene is something that you observe more in the monocots. No. You observe it in the diacons too. These are all, they're not the same genes. They're not the same sequences. But there are many, many examples. So I showed the mining of the 17 plant genomes. There are lots and lots of examples. OK, so you observe it every once in a while. But then you observe some, you saw also some specific example from the diacons only. And not from the grasses. No, but they're everywhere. They're everywhere. So the difference was in one of the sequence comparisons for the terpenic synthase P450 gene pairs. The point that I was making there is you can find those across the monocots and the diacons. And that includes the known clusters and also new clusters. But what seems to be happening is that those in the diacons, the terpenic synthase P450 gene pairs are kind of duplicating like that. Whereas in the monocots, it's all mixing and matching, which suggests something really interesting in terms of micro-syntony and recombination, at least as we've looked at it. I have a question for Anne. It's rather naive. You show a picture of, you have a picture of the promoter being expressed in the tips of the dopses and dries. Do you think that this promoter will work, say, suppose in the chute of the plant also? Or is it super specific? No, it has exactly the same expression pattern in diverse plant species, as it does in oat. And in oat, it's specifically in the epidermal cells of the root meristens and the lateral root initials. So this is very strange how a recently evolved pathway has acquired an ancient regulation mechanism. And it also means it provides a readout that we can use to look at plant development across. We're now going back to, we've been doing some Mark Antio work. We're not quite sure whether the promoter is doing anything in Mark Antio at the moment, but we're going back from higher plants to towards the lower plants to see if we can delineate where this breaks down. One last question. Question for Anne. From your comparative study of the 17 genomes and the clusters, did you bring out a sort of a favorite scenario for the evolutionary dynamic that the clusters can make disappear? Well, that's the whole, that's another thing, the birth, life, and death of clusters. So there's a whole population genetics are coming around this, which I'm still trying to get to grips with. But presumably, once the clusters lock into place, they're optimized, they're delivering a selective advantage. There must be massive selection to drive these things together. But then when they become no longer necessary, do they disappear completely? Do they break up? There are examples which I didn't go into of split clusters, which is still operational. So the maize, dimboa, the hydroxamic acid cluster, wheat and rye also make hydroxamic acids. But the genes are split into two sub-clusters there. So the hypothesis is there was an ancient translocation event that occurred after maize and wheat and rye separated that split. Nevertheless, it's interesting that the pathway still works, that's one thing. And the other interesting thing is whether once something perhaps has become established and becomes indispensable, so it's essentially no longer specialized, if that happens, then could those genes be welcomed into the genome? Was there ever a point when the deciding pathway genes were tested for example? Is that what you do? Okay, I propose to thank our speakers again for their presentation this morning.