 Kaj smo videli odpočeniti. Čestno ter se je online, ale što se načino prišlo. Imam ti mikrofon, nekaj ne zvojete. Čestno prišlo Daniel Segrejov z Bostom Universtvu. Že nezavršijo v nim, but also kind of give the talk online. And it's not going to talk about metabolism in ancient time and space. So thanks a lot to Daniel to be in us. Thank you, Tomatoеш, mind how you all are? Hej. Just says OK, good. Sorry for not being there in person. First of all really was looking forward to that and this happened. But I am happy to have je to čas, da se pridajte način, je, da je Jakobo se zelo, da se pridajte, kaj mi zelo in zelo se vzelo. Tudi, da se je vseč, teželim taj, da je bljero, da se pošljelim taj, da srečim, nekaj pomembnji je tudi, kaj je tudi, da se pošljelim in se vseč, prizvajte z vsečenih tudi, nekaj vsečenih tudi, ki so, ampak, nekaj nekaj nekaj nekaj, vzveči vsakvih vrštih viziv, in na vseh se predstavljajo na vseh metabolizm in vzvečnjem mikrobiliko ekosistemizmu. V štih labi vseh, nekaj smo vseh v mikrobiliko komunitivnih v sojlji in vzvečnih, zelo smo na sinteticijne komunitivnih. Vseh smo inhle in vseh mikrobiljnih in insekti. However, I'm not going to start as hinted by the title from the ancient past. And I want to make this point that somehow, I like to think of metabolism as memory in the origin of life. So I'm going to die straight back to 3.8 or so billion years ago, and think about the meaning of metabolism back then and what we understand about the history of metabolism. lemonija, ga to zvrživ Ja se razstavihapti. In zelo smo se praliv eštečne učeny, prič da je občite, kako načinno ustavljamo nepo dis koncept. I sližba ta pradke koncept, ta člık, da je, da je mene, dozira med the이다 in se dozirila, in imaweight. In v glasba, kako ima v vseh inovilitejo celi in deloviteva aleske premačenje, nega so la, da je inovilitej, v bistvom doblega, And the other is to look at metabolizers as a historical record and see whether we can understand the history of metabolizers by looking at networks today and think of networks as fossils. So let me start with this first part, then a lot of this dates back to work that did during my phd many years ago, and this somehow summarizes a little bit this dichotomy of possible views tren creates a lot of Euchke organ vse protagroje. The New Town's place sees an egg and viruses. Here, as I mentioned, there's a number of ways of dreaming ancient history of life in between. I just want to point out this very basic concept that somehow we still have often represented in the text book chemistry text, je to vzvečen za ljudje, in vseščen za informacije, za štora in ki se prišlo, da se vzvečenje z tem, da je tudi iz vso nekaj polimers izvaj stavil, da ero nekaj početil načine tudi odmah, tako zelo in komponent, in nekaj problema z terzim, da je več zelo izvajen, kako se zelo izvajenje z molekjulj, zelo prišlo, da je bilo vzvečen izvajen v tem, a potem so je tamo alternativno imelje, taj jaz izglasboj, da se povedem, nekaj zdaj, da je tudi in zelo pozniraja. Zelo oz. A po to, da je tudi čest, zelo ki je bilo dobro vršt para, ja bejebo tudi, da je tudi za njih, da je nožena inkonekčne, zelo je dne. Ja bo. Prečo. To je tudi oz. Šte, da je povednije. Je to povednije, treba na ter lojs'a, na ter, plenjega. Taj pa je, gre povedniji. Na to, da je pozniraja. Povednije govršt, je tudi o tom, da se však je pomembno molekul, mjude naša katalizacija, katalizacija naša katalizacija in z vrstajšim prikratnjem, z vrstajšim. Proto, sem odmah sem počust, da zašlišimo tudi in ozrednje, da nekaj je noh spesifikčen molekul in je bilo, da je to ozrednje, tudi se zelo drugovom doličenem, En bilo je to, da videmo, na kaj je izgleda mnog pa glasboj, in trgaljo to, där ne večo se začent prav na tudi razkov, z delama, obročuje, ko jaz drugim uvijek, konek, kaj je izgleda, tudi da je posljete, nekaj, čim je, n'avi najte containing, ni pelj, n'adi držaj, where an ensemble of molecules could grow and then split. And this is a form of inheritance. So this would be, though, a very trivial form of inheritance because if there is homogeneity in the composition of this ensemble, of course, there is a form of cell for production, but it's trivial. At the opposite extreme of the spectrum, if you imagine extreme complexity of the molecular components, you could imagine something like a random walk in the space of compositions and an ensemble of molecules in an early milieu on prebiotic Earth, giving rise to ensembles that will potentially grow due to self-assembly or other processes and occasionally break because of physical forces. But if there is too much diversity or there is no mechanism for maintenance of composition, this would be a random walk again in the space of compositions and it would be nothing like what we see today in modern cells. And the idea that we had proposed back then, based on a lot of work done before, by dating back from Oparin and Harold Morović, Stuart Kaufman, Freeman Dyson and others, was that somehow if the forces and the mechanism and the processes determining how an ensemble of molecules could grow and divide has certain specific properties, then there is a chance to have a composition of molecules that will grow and then split and give rise to a progeny that is similar enough to the original ensemble so that you can have what we call compositional inheritance or compositional genome or short composome, that is somehow a potential early version of memory and inheritance based on which you could jumpstart an evolutionary process. Now, I'm skipping a lot of details on this and just to give you an idea of how such a process could arise, you could imagine either chemical reactions giving rise to different molecules in this original primordial ensemble, but even just joining of existing molecules, such as lipid or self-aggregating molecules, and all you need really is a set of molecules that selects which other molecule will grow and this can be viewed as a trajectory in the space of compositions. In fact, we call this as one of the possible embodiments the lipid world because this is based on the idea of self spontaneous aggregation, but the general concept of the trajectory in the space of composition giving rise to an attractor that looks like this self-aggregating process is much more general and I think actually it's nicely explained in a much more recent review from Doron Lancet in whose lab at the Weizmann Institute in Israel I did this work early on. They published recently this article showing in more detail how one can think of this compositional inheritance as emerging as an attractor in the trajectory in the space of composition. So in my mind, when I think now of early life and metabolism, I really visualize something like this where there are complex chemical processes and you can ask about the rise of something like an eigenvector in the space of compositions that is maintained despite noise and there is beautiful work. I don't know if Sanjay is there. He's done beautiful work in this area as well. Now, one of the reasons I'm telling you this and I'm skipping a lot of the details is that somehow this is not so dissimilar from what we do when we think of flux balance analysis and modern cellular metabolism. This is actually the molecular composition, the biomass composition of anicoli cell. You see the proportion of different amino acids, phospholipids, nucleotides and cofactors. And this is in fact exactly the numbers that we use when we make flux balance model of a cell. So I think you've heard in the past few days about flux balance modeling so I'm not gonna go into any more detail but just wanna remind you that you have the whole stoichiometry of the metabolic network, sources of the different elements coming in and then ultimately you ask how can a cell produce the precise composition of the molecules needed to produce amino acids, nucleotides, lipids and so on that can pose biomass of the cell. And essentially this is very similar to this idea of composome. It is a cell for producing composition of molecules. Let me pause here, just see if there is any question. I wanna try to make this as interactive as possible so please interrupt or I will pause occasionally to see if there are questions. Any question? Seems there are no questions for now. Okay, okay, okay. So I'll keep going. And I will then jump to another way of thinking of metabolism as memory and tell a little bit about work we've done more recently about how to look at the historical record of life by looking at metabolism as a network fossil. So obviously we can look, think of fossils of life in terms of rocks and real fossils in rocks. And it's very natural now to look at genomic sequences as fossils of life where we can read about the history of life. But I think it's less usual and less natural but still actually they're interesting and I think more and more increasingly explored. The idea of looking at networks as fossils of the ancient life. So this is the structure of all known metabolic networks, metabolic reactions across different organisms from it's a representation of a collection of all metabolic reactions. You can see here the TCA cycle and glycolysis but it's obviously very complex. And one of the questions is, what does this network tell us about the ancient history of metabolism potentially dating back to a time before the emergence of genomes and transcription translation and so on? And how do you even do this? How do you interrogate this network to try and understand the ancient history of metabolism? And one way of doing this was suggested by Oliver Ebenhol and Reynard Heinrich several years ago in a series of beautiful papers starting with the genome informatics in 2004 where they proposed what they call the network expansion algorithm. I don't know, some of you may be familiar with this. I'll tell you with this super simple example how this works and you'll see in a second why this is powerful for interrogating also complex networks. So the idea is the following. Imagine this is the equivalent of this very large metabolism I showed you. So this is the collection of all known metabolic reactions that skate, this is just two reactions. So you can ask the following question. You can ask if some set of molecules are initially present, for example, in an early environment and we call this the seed molecules. Imagine these are these two molecules you can ask what reactions are possible and you don't even ask questions about the enzymes that catalyze these reactions, you just ask whether in terms of the presence of individual molecules what reactions are potentially feasible in this world. So if these two molecules are present this bimolecular reaction can take place so you'll add these two products as an additional set of molecules in your network and that's where it ends. There are no more reactions possible. This is called the scope of the network and this set of four molecules now is the collection of all possible metabolites that are feasible under these initial conditions. And you can change of course the initial condition and ask for example if this molecule is also part of the seed set and in this case in addition to these two molecules now that this molecule is feasible and this is initially present this reaction can fire giving rise to these two molecules in this case now all possible molecules are feasible and the scope corresponds to the whole network. And now you can imagine asking similar questions based on the collection of all metabolic networks and what is nice about this is that this is equivalent to asking questions about ecosystem level metabolism because now we're asking we can ask questions about how a seed of metabolites is present in this somewhere in this initial network what other portions of metabolism can they reach and you can change the set of initial molecules and explore what portions of metabolism are feasible based on those initial molecules. And one thing that is interesting again is that where by exploring and walking along this network you're exploring not the metabolism of any individual organism but the collective metabolism of the biosphere and what is the potential given this initial molecular set. Does that make sense? Does, yes. Okay. There is a question. Hi, so I was wondering if the seed and scope I guess looking at it from a more history dependent matter this also depends on the conditions that the organisms are in, right? So how do you deal with that? So kind of what reactions are possible don't only depend on the seeds but also the environmental conditions for example that this whole network is embedded in and this could change over time or do I understand something wrong here? No, no, this is a perfect and beautiful question in your extent exactly. And I'll show in a little bit in early work we did on this we did not address this environmental dependence question except for the choice of the metabolites themselves. But I'll just mention this now. So one can do this network expansion this is very, very simple way just by asking, walking on the topology of the reaction network but you can add additional information and in particular you can ask about thermodynamic feasibility and that's one way of embedding important thermodynamic environmental information because when you introduce the thermodynamics you can ask, given this metabolites and assuming certain ranges of possible concentrations and if we know the delta G for this reaction and take into account then the temperature and pH dependent on that reaction you can ask not just whether that reaction is feasible in terms of the topology but also whether it's reasonably true that that reaction could occur spontaneously under those conditions. So you can add at least pH and temperature as environmental variables. Of course, there is a limit to how much we can do with this but this is hopefully addressing at least partially your question. And I'll show you some examples of this. Any other question? I think no, so don't move on. So I'll show you first a result that we obtained using this method several years ago. This was work done only by Jason Raymond. He asked a question about the dependence of metabolism on the rise of oxygen in the atmosphere during the great oxidation event that happened about 2.2 billion years ago and that's when cyanobacteria started accumulating oxygen atmosphere changing completely the fate of life and probably inducing the rise of multicellular organi. So it was a very interesting question and we had the opportunity to ask the following question using this network expansion algorithm and the idea there was to start with a seed set that did not include oxygen and compare it with a seed set that did include oxygen so you can look at the possible changes in global metabolism induced by the presence of oxygen. And what we found was that when you make this transition from anoxic to anoxic global metabolism there is several hundreds of reactions and hundreds of metabolites that are added. So in blue this is the structure of the anoxic network and the red portions are the branches that were added due to the presence of oxygen and what is interesting here is that oxygen here was essentially used as a molecular compound for biosynthetic processes not necessarily just as a electron acceptor as we obviously know today but a lot of these branches are branches for example sterol biosynthesis, the turbulent environment biosynthesis. These are all complex molecules that require molecular oxygen for biosynthesis and there are alternative non-oxygen dependent pathways for some of these molecules but there are several new branches that are actually associated with complex eukaryotic organics. So this is one first possible application of this network expansion algorithm to ask questions that would be otherwise very difficult to ask. And more recently, we, in a particular, Josh Goldford, a former student in my lab, thought of asking a question that would address freely something going further back into the history of life and addressing what is known as the phosphate problem. So phosphate, which is obviously ubiquitous in present-day living systems, is usually other than in biotic processes locked in rocks, such as apathides, for sure, here. And it's also a very major trigger of blooms, so it's often a limiting nutrients in metabolic processes in natural ecosystem but it's been very hard to understand how phosphate could become part of life early on because of this poor bioavailability. And Josh thought of doing the following, taking a plausible set of molecules present, thought to be present on early Earth, and leaving out phosphate. And I assume many of you are familiar with how metabolism looks today. If you look at reactions in metabolism, for example, in E. coli, there is tons of reactions that, of course, include ATP and ADP, phosphate-containing molecules that where phosphate bonds carry energy, that is enabling several reactions to occur. And in fact, our first thought was that if we were to do this network expansion in metabolism, in this global metabolism, without including any source of phosphate, there would barely be any connected pieces of metabolism. I imagine, in my mind, really just a lot of bits and pieces but no feasible network based on this metabolism. Notice that there are sources of sulfur here, what are called tyoesters, sulfur-containing compounds that were hypothesized before to be potential energy carriers prior to the emergence of phosphate. So what we found was quite surprising and I'll show you, jump straight to the result that Josh found. And essentially, what emerged from this early compound was a network of 260 metabolites, 315 reactions, all connected. You can see here in blue, the initial seed. So this is where this network expansion starts and it expands to several molecules that include a lot of the known central carbonyl metabolism intermediates, such as pyruvate, and include a lot of amino acid. And what was somehow surprising, I mean, first of all, is that there is such a network at all, that there is a, whether or not one, and I think one should be skeptic about to how relevant or how, whether this is telling us something true about the early stages of metabolism, but it is, it was surprising, it is surprising and it is true that such a sub-network exist in metabolism today. So somehow the fact itself that there is a strong connected component sub-network of reactions that do not involve phosphate in metabolism today is in itself quite stunning. You can actually look at the enzyme that catalyzed these reactions today in an attempt to really try and connect this to early life. And the idea here was to try and ask the following question. If you look at the enzyme that catalyzed this reaction today and remember that in doing this network expansion, we didn't ask anything about enzyme. We're just looking at the topology of metabolism, but now we have a network and we can look at the enzyme that catalyzed this reaction in living systems today and ask the following question. Is there any sign that the enzyme that catalyzed these reactions today have an ancient origin? And what Josh thought of doing is asking the following question. Do we see an enrichment in this enzymes today for enzymes that have as their core co-factors iron sulfur clusters, which are thought to be some of the earliest minerals catalyzing reaction. And what we found shown here is that this core network is strongly enriched relative to the full network even without presence of oxygen. So irrespective of the oxygen level in the atmosphere, there is a strong enrichment for protein that contain both iron sulfur cluster as well as zinc. So this is consistent with these ideas explored by several other researchers today and proposed before by others that iron sulfur cluster these minerals where some of the early catalyst and then could be incorporated in modern enzymes and catalyzed reaction in this early network. There is a question. Oh, yes. Sorry, I don't understand if I missed this or my question is the network that you were showing earlier. So the one, the next one that you find, yes, exactly this one. Can it synthesize all 21 amino acids like technically, could an organism use just this network to make all 20 amino acids? Ignoring, of course, the fact that this does not involve the phosphase, so there is no energy, there is no ATP. Does my question make sense? Excellent question. Yeah, yeah, I know that the question makes perfect. Actually, I don't remember exactly. I don't think all of the amino acids can be synthesized. I think two thirds or so of the amino acids can be synthesized. And I also want to highlight that it's true that this is, so this is not contained phosphase. I should say this means that really there are reactions in metabolism today that are independent of phosphate. And so this could not have been driven by phosphate, but the hypothesis that could have been driven by this thioester, so it could be still potentially energetically feasible with the thioester, the sulfur containing compounds playing the initial role of energy cares prior to the emergence of phosphate. But this would be obviously an early metabolism, incomplete metabolism. Notice also there is something else that is interesting here is that because nucleotides contain phosphate, this early putative phosphate independent network could not produce nucleotide. So this could be producing amino acids and several other molecules, including some other polymers, but would require the rise of phosphate in order to add nucleotides in DNA and RNA. But that's an excellent question. Any other question? Okay, I think we can move on. Okay. So someone was asking early about energetic constraints, and I'm not gonna go into full details, but there is a follow up paper that was published in 2019 where we incorporated this thermodynamic constraints. And this was done using the Equilibrator and I don't know if, yeah, some of the Equilibrator developers might be there. And the idea was that one could take into account in order to know whether specific reaction were feasible. One could add constraints and decide that only reactions that could get above the thermodynamic constraints would be feasible. And now through the Equilibrator we could calculate the pH and temperature dependence of those thermodynamic constraints and have a network that would incorporate some of the environmental information. And one of the things that we found in this context was that there was reduced core network. And what is interesting, and again, that would require a lot of time to go into, but I just wanna point out that this reminds a little bit of the TCA cycle and in fact it includes a lot of the reactions of the TCA cycle, but also involves other thylester dependent reactions. So this is a little bit of a mosaic of reactions that are not known to exist in any specific individual organism today, but are taken from different organism and could have been a putative central carbon metabolism early on. And this is consistent with thermodynamic constraints for a broad set of pH and temperatures. And I invite you to look at this if you wanna learn more about this. One thing that now connects us back to flux balance modeling is that once you have networks like this, you can also try to make a flux balance model of a protocell. And that's what we did here. And again, I don't wanna go into too much detail, but you can essentially take this reactions that I showed before, think of what a putative early biomass might have looked like. And again, there were no nucleotides or no DNA RNA and still no proteins really, but there could have been other polymers and lipids that were feasible under those conditions. And this is very speculative models of a protocell, but the point we wanted to make here is that there is this connection between methods that we use for modeling cellular metabolism today and the way we think about ancient networks and they are connected in ways actually can be very useful to apply the systems biology approaches such as like flux balance analysis to study the feasibility and the consistency of networks of early life. The other thing that I'll point out before jumping back to the present day modeling of communities is that one can explore other types of question using flux balance models and ask questions about early life. And I wanna point you to this paper public recently also spearheaded by Josh Goldford that shows how one can ask questions about the emergence of cofactors and redux cofactors in particular. And the idea was the following. In metabolism today, in addition to ATP mentioned before that are the energy carriers, there is a strong dependence of this electron carriers, NAD and NADP. So these are the molecules that allow electrons to move from molecule to molecule and are essential in metabolism today. And one of the things that is not clear even if again it's explained in textbook, but chemistry textbooks as just being related to biosynthetic metabolism or cannibalism and catabolism that the degradation of molecules typically NADP is associated with cannibalism and NAD with catabolism but it's not if you look at where these cofactors appear, it's not very clear that there is a universal association with cannibalism and catabolism, especially given that fluxes can change based on conditions and so on. So we asked the following question. Why are there two cofactors? Why isn't one enough? Or is one potentially enough? So one can do the following exercise using flux finance model. So you can take the whole network of reactions and substitute arbitrarily. For example, you can do different exercises which are all illustrated in this paper. You can randomly switch some cofactors and switch NADP to NAD and see what happens. Or you can try and ask what happens if you adjust one cofactor and switch all the NADP to NAD. So you'd have a network that is stoichiometrically very well defined and seems feasible. And in fact, if you switch all NADP to NAD and you have a single cofactor in the network, stoichiometrically, this seems completely feasible. So you can run flux finance model, so you'll get biomass and everything seems fine. And what is even more surprising, even if you introduce the known term with dynamic constraints, it seems feasible that an E. coli cell could potentially survive and grow just with one cofactor. So this is very surprising and this is a potentially testable hypothesis. I think it could be very challenging to build an E. coli cell where all the NADPH are substituted by NAD. But if those were possible at some point, our prediction is that it would be possible to have a single cofactor dependent in coli. So then the question was, why then would we have two cofactors in present day cells? And the answer suggested by a mathematical model developed by Josh is that this has to do with the cost of protein production. And that somehow, by having two cofactors, one can enable more efficiently running reactions in both directions of donating and receiving electrons in a much more efficient way. But this has to do with the cost of protein production, not with the fundamental invisibility of having a single cofactor. So this has to do with the protein allocation and optimality of protein allocation by having two cofactors that can simultaneously be present and available to run reactions in both directions. And I will switch gears now to go, dive back into modern metabolism. And one of the links and I want to make here is that somehow I like to think of metabolism a really multi-scale process. So we already seen this, right? Because the networks were illustrated when from individual enzymes and individual reactions to really trying to make this very ambitious estimates of global planetary biochemistry in the past. And I think it's actually, whether or not we know how to do this now, I think it's really fascinating to think that metabolism, unlike I think any other metabolic process is truly multi-scale because you can think of enzymes, cells, are all catalysts of metabolic flow and you can talk about the same molecules that are involved in single enzyme reactions and whole cells and ecosystem and the planetary scale. So there is some really continuity in buildup, which I think is an exciting challenge for all of us to think about how do we bridge these scales with models of metabolism. But I wanna go a little bit back to when we started thinking about these questions of ecosystem level metabolism and this for us started in before 2010, around that time, motivated by some early work done by Wenin, Shu and others that had designed and constructed these synthetic communities. Some of I think the earliest synthetic communities. This was an example with two yeast cells that were oxotrophic for lysine and adenine respectively. So one of these yeast was engineered not to be able to produce lysine, the other not to produce adenine, but together they could coexist and survive and grow because of the exchange of these two molecules. And this was because essentially there were tweaking of the internal circuits of the cells. One thing we started thinking about in Niels Klitzkor, the former student in the lab, is whether we could do something similar and started exploring this with the flux balance modeling, but modifying the question a little bit. So instead of tweaking the internal circuits, we wanted to tweak the environment and ask the following question. Can we design an environment and choose a nitrogen carbon source and so on so that given to microbes, to bacteria, for example, we could induce an obligate interaction between these two bacteria. So now we would like to ask whether it's possible to induce interactions without tweaking the internal circuits of the cells, but just by modifying the environment. And what we found back then using flux balance models is that for several organisms, there were in fact millions of possible solutions, millions of possible combinations of carbon nitrogen sources and so on that should induce obligate neutralistic interaction between organisms. And this, if this was true, that would mean that there are a lot of possibilities for interactions between different bacteria in micro communities. And these interactions could be mediated by the exchange of essential metabolites and also that these interactions would be strongly dependent on the environment. So this was really the early work that drove us to start thinking a lot more about metabolism in communities and whether we can model metabolism communities. And ultimately also as we do now in the lab, try to build experimentally synthetic communities and explore how these communities depend on environmental condition. So there are different aspects to this which we explore in detail, the exchange of metabolites, how the structure of environment modifies interactions and also how environmental complexity affects interaction between species. I'm not gonna cover all of these topics. I'll tell you a little bit in particular about space and time. And I'll illustrate this by describing a method that I don't know if may have been brought up before in the past few days but if not, I'll just summarize quickly. This is an approach called dynamic flux balance analysis and it's a modification of the basic concept of flux balance analysis. In fact, the original paper that suggested this was published in 2002 by Mahadevan in colleagues. And I think it's a beautiful paper that kind of was sitting there for a long time, I think before people started realizing the relevance for ecosystem level metabolism. And in fact, the paper introduces two different kinds of dynamic flux balance analysis, one that is based on the global optimization of a whole trajectory. I'm not gonna go into that, I'll describe the simplest version which has to do with just taking individual steps and discretizing time. And by discretizing time, you can imagine taking for an individual time step, solving a flux balance problem for an individual organism where you have, again, it's on metabolism, the biomass that represents growth and so on. And when you solve the flux balance problem for this individual organism, what you find is the slope of this initial portion of the growth curve. And now you make an assumption about the size of delta T. And what happens then is that you start with an initial biomass composition, biomass amounts for this organism. You also start from an initial amount of the nutrient and you have to translate the nutrient abundance into a flux by using a kinetic term like a Michaelis Menten equation. So this ends up being some kind of hybrid model where the concentration of metabolites is translated into a flux. But what you get back then is that the concentration of the metabolites will change in time because, again, you have the flux of consumption of this metabolite, so you update this. And now, at this next time point, you have a modified abundance of the nutrients, a modified biomass. You can iterate, again, flux balance and ultimately you'll obtain this piecewise linear approximation of the growth curve and also a prediction of how the environment changes due to growth of this organism. Now, what is nice about this, it has, you know, people have used this to model dioxic shift and there is a lot of research in this area. What we were curious to do was to see how we could use this to model communities in particular because imagine this organism, for example, using glucose could secrete acetate, the acetate start accumulating in the environment and now there could be a second organism that might not have been able to grow in the original glucose but could grow in this acetate and now this organism can grow because of the presence of this first organism. So what you observe in this case is that there is an emergent interaction between the green and yellow organism mediated by the exchange of this acetate molecule and what is cool is that this was a consequent of each organism doing as in traditional flux balance analysis maximizing its own biomass. So there is no assumption of an ecosystem level objective and it's an interesting question in itself whether and how to potentially look for such things but here there is still just individual organisms objective each organism is trying to do what is best for itself and still this will give rise to this exchange and interaction. So we embedded this right there. I mean we decided that we wanted to incorporate the spatial component in this so we embedded this whole thing in a spatial grid where in addition to this dynamic flux balance processes there could be diffusion between neighboring regions. We started doing this in 2D now this idea is also expanded to three dimensions. This was pioneered by Bill Reel and Will Harakam was working in Chris Marx's lab at the time and now is carried on by Ilya Dukovsky in my lab and others and see how am I doing with time. I want to leave time for questions but I'll just tell you a bit more about this system called Comets, competition microwave assistant in time and space and this was tested experimentally early on on some simple communities and I'm going to skip the details here but just so that you know this were artificially constructed synthetic communities of two or three organisms this was devised by Will Harakam early on and then Chris Marx and Will and these are very elegant and interesting systems in themselves, experimental systems and what was nice that was that by after fitting just a few parameters for individual organisms as you know flux balance doesn't have many parameters but we took some of the kinetic parameters from the literature, the comets would predict quite accurately the final abundance of the two species and also in the three species community comets did a pretty good job. So this is some early results. One thing I want to point out also is that some of the secretions in the systems are spontaneous such as E. coli, secretion, acetate others in this case were evolved imposed through evolution adaptation and this is something that we're still very interested in I think it's still largely an open question how much of these mediated interactions mediated by metabolic exchange even in communities today how many of these are spontaneous or what we also call now costless secretions that are just the outcome of cells using whatever resources that are available and as an outcome of that secreting molecules that are useless from their perspective but that can be used as useful resources by other organisms and how much of these are evolved costly production that can be also mediating interactions so I think there is this really interesting question how many of these interactions are evolved and costly and how many are costless I think this is still an open question there is evidence from work done in the Sanchez lab that a lot of these interactions are somehow due to individual organisms secreting stuff spontaneously and that stuff can help other organisms grow so I think there is probably a lot of this costless exchange out there but I want to go back to comments just to illustrate it was really exciting for us that a number of people including Alvaro and Will Harcom in their lab they started using comets for different applications and rather than they start adding different modules to comets so this is kind of an exercise in collaborative software writing so rather than developing different subversions we try to bring together all these different new components and we work they work, I didn't do really much except overseeing this but there was a lot of coding and a lot of exchange for a few years to bring this together and this gave rise to a new version of comets which we described in this Nature Protocols paper a couple of years ago and I just want to highlight that comets is written in Java so it's not a straightforward software to use but now we have a Python pool box and a Matlab tool box I think the Python pool box is particularly good and still under constant revision and development so one can write macro functions to define different environments create a Petri dish for example with different media continuous culture batch culture and add different organism from standard cobra tool box and run simulations generating growth curves and evolutionary dynamics and there are a lot of other features that are available and this is still working progress comets is available this website is free and open source so anybody interested is welcome to contribute or ask questions try to use it and give us feedback with some of the aspects that are still kind of under development and we have some preliminary work on this but just to show you possible expansion of comets in new directions one is the addition of extracellular enzymes which are difficult to model using traditional flux balance analysis but now you can model extracellular enzymes as a metabolite so if you have amino acids producing biomass in a certain proportion you can assume that some of them will produce an extracellular enzyme for example a cellulase that is being secreted and once it's secreted is treated as if it was a metabolite except that it can carry reactions so you can model Mikaelis Menten kinetiks in the extracellular environments and now for example cellulose can be degraded giving rise to glucose that can be imported in the cell and you can ask questions such as does the cell have an optimal allocation of the amino acids how much should go into biomass how much should go into producing the cellulase in order to maximize growth and you can imagine that if you produce too much of the enzyme you have no biomass left but if you produce too much of the biomass then you won't have enough cellulase to cut the cellulose that is giving rise to the only source of carbon so we predict that there is this optimal rate of allocation of the amino acids to the enzyme and this depends on the amount of initial cellulose present in the cell this is for now bioarchive preprint if there is anyone interested in fotosynthetic metabolism we also added in comets the capacity to simulate day-night cycle so one can model the presence of light as a pseudo metabolite in comets enabling simulation for example of procococcus and other cyanobacteria I have a few things but I want to leave time for questions so I'll just conclude by showing something about what the spatial component of comets these are actually a little bit outdated right now we have now see they are slightly different scales but this is just illustrating by adding collaboration with the Kiril Korolev in physics at BU we are now modeling diffusion in comets in a much more accurate way including non-linearity of diffusion because of the presence of biomass and molecules diffuse through the biomass and also adding noise in the population dynamics noise in the simulations we can recapitulate for example the different shapes of the E. coli colonies shown here in microscopy images as a function of the hardness of the agar and one of the applications of this just to give you an idea where one of the things we are going with this these are devices that our collaborators at Berkeley National Lab use for growing plants this is a little flat device called the ecofab where they can grow plants and these are the roots and there are microbes inoculated in this compartment and we can take the image of the roots simulate them in comets seed microbes in different regions we can simulate the exudation of metabolites from the root plants and simulate the abundance of different organisms around the roots and this is ongoing but I just wanted to show you one of the possible applications of comets using this spatial features and to model structurally spatial communities I will conclude just by saying you know I think there are a lot of exciting things happening in this field but also a lot of open challenges and some of the things that I think need to be taken by people like you and people the interface of physics and biology in particular I think we have general scale models that are exciting but also as you know a lot of challenges in terms of reconstructing models for new organisms and we know how to build ecosystem level models but it's not clear how scalable this will be to very large communities so all these different approaches where one can use consumer resource models and have statistical approaches that can be scaled perhaps to larger ecosystems but I think it's interesting to think can we really course grain community models in a more in a continuous way and bridge gaps in these different approaches and I think this will have very important application just to give you an idea I think there's a lot of interest in question related to climate change and accumulation and of carbon and whether or not for example strategies involving microbial metabolism could help mitigate climate change by increasing the amount of carbon that is stored in soil and the oceans this is by looking at carbon use efficiency so I think it's obviously a long shot but I think it's interesting to explore whether models can help understand the processes that max that can increase carbon retention in soils so I think this are some of the new interesting frontiers and I'm gonna stop here and acknowledge all our funding sources and my lab and collaborators and there should be a few minutes for questions and thanks for listening thanks a lot so we have time for questions there is one hi thanks for the talk one question can you comment a bit on how much curation these models needs in the sense let's say you can't always say from a genome whether a species can grow on a certain carbon source or can do something so how much of like data collected from experiments do you need to actually predict accurately predict something using flux balance analysis with let's say two species or more yeah that's a great question is that Martina yes hi so it's a question that actually we are we are kind of keeps us very busy now because we are very set about trying to apply flux balance models and come into different environments and and we hadn't done this you know for a long time we relied on the existing tools and I think there's great work being done the car meet team and and now what is happening with model seed in k-base but we realized that this automatic reconstruction and automatic upfilling are really not enough to give accurate predictions and so one thing we're doing right now is and I think that's where you know we should be thinking is that we have a lot of phenotypic data so for example we have growth on different carbon sources you know even just binary growth or non growth in many different carbon sources I think that information has not been incorporated well yet in flux balance models in this to make segmentary constructions and but I by you know what I see is that for example and that you know the other question how do you how do you test what what does it mean to have a good model so for example what I can tell you is that we can now do iteration of gap filling on different carbon sources and and we can basically match growth non growth patterns on several carbon sources and what is actually surprisingly difficult is to match the non growth right so if you do gap filling and ask okay this organ should grow on acetate on arabinoz and so on so you'll add reaction but it's easy to add reaction but then you'll grow on everything so now we're careful to make sure that or you don't grow on everything and we can match reasonably well growth on I would say few dozens carbon sources but I think that the road ahead is still tricky ultimately if you ask me what I think will happen is that we'll have to do some kind of hybrid model going beyond flux balance but I don't know how that will look like but I think I think you know what I want to say because I know there is a lot of skepticism and questions on flux balance models given the challenge of annotation and I want to say I think of course we should be very mindful about this and very careful but I think still reconstructing stoichiometry and trying to test and validate is still an excellent way of condensing knowledge testing what we know and what we don't know about an organism so I know I don't know if that you know I think that partially answers your question but I think part of this is that I think we need a lot of work we're doing some of this and but we also be open to new hybrid models very good we have time for more questions if any yes how is the diffusion of bacteria modeled in the special part of comet? so it's we essentially do a equivalent of PDEs for standard diffusion except that I actually don't have the equations here but we take into account the non-linearity so the fact that diffusivity can change in as a function of the density and the density can change because of the growth of bacteria but it's essentially PDEs of diffusion equations coupled with the with flux balance so this obviously doesn't go to the level of individual cells and it's still a mean field model but yeah I think that's that's another thing that I think would be interesting to explore in the future there is all this you know single cell agent based models and I think it would be interesting how the two could interface okay but so does I don't like does it have global properties like the viscosity of some strains or some strains that create extra polymeric substances does it have anything like these or are all different bacteria strains kind of similar in a sense and only the density counts yeah yeah yeah that's a great question so we can change the diffusivity that would depend potentially the property of the individual bacteria but we haven't looked much diversity we have not looked at extra poly extra sort of matrix and secretion although you know do you know through this processes now this is in principle possible so this is this is yeah just the people the iceberg of this and I think you know for example one thing also that is missing and we're starting to put in there now is chemotaxis so we're now adding chemotaxis to see and we can capture for example the ring structure shown in Terry was lab through a combination of comets with diffusion and and the chemotaxis but this is you know there is a long road ahead and I agree I think you know extending this to specific organism would be very interesting thank you very much any other question yes, there is one hi actually I'm interested to know a little bit more about the how you solve the growing growing in all media problem like when you have an over capable network which are the approach you are using for for including the information that now you need to non grow in a certain media and yeah yeah, that's a great great question I wish that a good algorithm the truth is that we don't I know, you know Kostas Moranas and others that developed some algorithms in the past we haven't tried those but I think you know you can email me I can point you to a couple of papers that proposed mixed integer linear programming algorithm to solve this problem what we're doing for now is basically manual loops where after adding after gaping on one set of nutrients we check growth on all other other nutrients and we revisit the addition and one thing we found, for example there are surprisingly a few reactions that cause the models to grow on everything and some of them are incorrectly annotated as reversible we send feedback to model seed and I think now the some of this has been picked so the some of this were issues with some of the reactions but so far we do it manually so I think this is still you know there is certainly a better way of doing this happy to connect if you're interested in this there is one more question hi Daniel that was a very interesting talk this is Sanjay hi Sanjay thank you hi so you know very sort of intrigued and I find it quite impressive the effort to you know interrogate the metabolic network to see what is primitive and perhaps what is the most primitive inside it so at this point could you you know so all these metabolic networks are run by enzymes today and one would like to know what is possible without you know complex molecules like proteins is can something be said on that matter and what what might be the most primitive part of the of the metabolic network that you know that can possibly be run without large molecules like proteins like very efficient enzymes and so on maybe cofactors and so on yeah yes yes yes yes beautiful question and so I think there is just amazing work being done now for example some of this is in Joseph Moran's lab this paper here is one of the earlier papers that have a few more but they've been doing this experiments with minerals and they showed that you know starting from a few precursors in fact some recapitulated what we had found in the network expansion algorithm but you can have iron minerals and a few precursors and you can generate a lot of the TCA cycle and certain carbon metabolism intermediates without any enzyme so I think there is more and more of this non-enzymatic metabolism coming up and being demonstrated experimentally but I think this is just beginning I think people just had not asked these questions in the past because we're so focused on you know other types of questions but I think now that people do this will discover more and more on this non-enzymatic non-enzymatic metabolism there's also this question right of at some point the idea is that the minerals could be the early catalysts but at some point the molecules that are synthesized could themselves act as catalysts in this you know mutually catalytic networks and and I think yeah there is gonna be I bet a lot more work in this direction in the future but this is I think you know this would be a definitely a good starting point to read about this and I'm following this with great interest because I I agree with you I think that's that's what needs to be done very good any other question okay so let's thank Daniel again thank you everyone and enjoy the rest of the conference thank you thanks a lot so very good so we managed to be at least 10 minutes late so I did my job and so what we can do now is to move to the two choices you have so there is a tutorial here led by Justus on environmental dependency of selection great thanks and on the terrace who all from will lead the discussion on on books okay so what I propose to do is to have five minutes to sort of move around and thermalize and then we start the tutorial here and whatever you want to do okay