15:03:41 Good. Some way you so I'm at the Department of genetics evolution in environment. At the University College London. 15:03:49 So I was tasked by Daniel so great to give this review talk. 15:03:53 And I must say that I didn't have a lot of time to prepare for it. 15:03:57 It's that's because before right before the before I got it that's it because I was having a such a good time at a camping trip. 15:04:16 And so I want to actually first year with you and highlights on my vacation the 10 day, or the nine day camping trip, just in case the talk became a little bit underwhelming said we do see something really interesting in the beginning. 15:04:17 Right, so we went to Yosemite and you need a reservation to enter the gate of Yosemite. So this is a Half Dome. 15:04:26 Zero pointer. I just realized that. 15:04:29 Okay so okay so this, it's a Half Dome, you can see the upper and a lower false will be 70. And so much. 15:04:38 Yeah, so this is the view from a central don't. 15:04:42 And this is a cathedral lake in Yosemite. And then we went to a monolithic sauce to fall we soaked ourselves in very salty and alkaline water. 15:05:05 So this you can see the scale of the, of the twofer, and then the devil's post pile National Monument. And finally, Connie Lake columns. so all these locations about five hours drive from here so if you plan right you might even be able to visit them 15:05:14 while you're in UCSB. 15:05:16 Okay so synthetic microbial communities. Yu them. 15:05:22 So actually, I do have to address this question every single time I submit a grant proposal or paper so I'm actually, I've thought a lot about this but I'm curious, what does the audience think, what do you think about, you know, the pros and cons of 15:05:40 for anything you for anything. 15:05:43 Yes. 15:05:46 using synthetic systems 15:05:53 Okay. So you mentioned that I can maybe football synthesis of some compounds right yes ideas. What is it, very good, very good. Yes. 15:06:04 Anyone else. So I think these two basically summarize the the my next slide. Okay, so I would argue that. 15:06:17 Yes, I would argue that for any question you have in mind why you you have to choose a model system, you commit to the mathematical models, which are highly abstract. 15:06:21 You can also choose to study natural systems which are highly complex. 15:06:25 So we see the trade off right just as what's yesterday, Pankaj talking about the trade off between a different mathematical models but here we see a similar level similar type of trade off between different systems so we, as a complexity increases. 15:06:40 We have reduced cultural ability. 15:06:45 So synthetic communities. 15:06:48 I would argue, the synthetic communities can bridge these two extremes. 15:06:54 So on the one hand, we are using critical communities, or less complex and natural systems, they're more realistic. Compared to mathematical models, because after all, they are comprised of living organisms engineered or I engineered. 15:07:10 So, on my on my talk, will be two parts. The first part is applications of synthetic communities industry and in basic biology. 15:07:20 And the second hour the cover lessons we have learned. 15:07:23 Using a synthetic yeast community in there. Again, two vignettes was only addressing the question of how corporate cooperative communities can survive. 15:07:42 Okay, so first part applications. 15:07:45 So for now, if you want to produce a battlefield right so this community has been has been made so from center loss, so you could use this fungus, to the greatest settles into glucose. 15:07:58 And then you could use a metabolically engineered equalizer to convert glucose to so beautiful. So the advantage of synthetic communities, is that you could you could have foreseen volume make it very modular so instead of attribute to know you might 15:08:20 organism ghetto will take glucose to some other compound. So you could have this disadvantage. In a second example is this study, which showed that you know so the targeted restoration of the intestinal Michael powder with a simple defined, back to your 15:08:30 therapy results relapsing Clostridium difficile disease in mice. So what happens. So I just want to unpack this a little bit. So basically the C. difficile infection. 15:08:41 Infection so January treated by antibiotics. 15:08:44 But the problem is that if you treat treat some, at least for some patients, if you treat them with antibiotics. 15:08:52 For some, at least for some patients, if you treat them with antibiotics. They will be the CDF will go down, but once you stop it, it will relapse. Whereas you have this vicious cycle. 15:09:00 But in this study showed that he could have a defined Consortium, if you introduce that into, into the gut, you could prevent the. 15:09:07 You could have prevented relapse have celiac disease. 15:09:11 And I would argue that to find the community is a lot. I would say is easier to do quality control, there are complex say fecal transplant, so that it was the motivation behind this study. 15:09:24 And the third example is actually used currently use the industry. So industry to make vitamin C, you would need to the precursor called to Kj, and to make to Kj form this course deep liquor which is the byproduct of, you know, making complex industry 15:09:40 uses culture of the two organisms, 15:09:44 and both are required if you just use one, the yard will not be a very much. And in fact that this process has been optimized to such economical efficiency, that the action is replacing the chemical synthesis of to Kj. 15:09:57 So this example of synthetic or communities. We use in industry. 15:10:03 So now I want to shift the gear to talk about what is more, what is dear to my heart right is the housing gated communities and the basic biology, basic understanding about the biology. 15:10:13 And this is. 15:10:15 This book has really influenced my thinking while I was the postdoc at Rockefeller, the struggle for existence by Jeff costs. So that book is. 15:10:24 It's very thoughtfully written in our, in it's not very long, so I would strongly recommend, as soon as we're interested in quantitative understanding of ecology to read this book. 15:10:35 So Jeff costs. 15:10:38 So one example Jeff calls Packard was that was it like was how, what are the conditions under which populations in prey predator communities will oscillate. 15:10:50 So, so let me just go through very quickly the equations. So in a homogeneous environment. The population size of prey changed 15:11:00 due to growth of the prey. Right, so it's very simple growth rate, and then minus consumption of the prey predator and prey predator is similar to just be randomly colliding with each other. 15:11:15 And the ones that colliding, and overweight and once they encountered each other place will be will be captured and consumed by predator. 15:11:22 So now the predator predator population changed. 15:11:28 The function of this. So, this term is very similar to that except you know there's always the conversion factor so it's this minus the death. So if there were no zero prey predator would just die. 15:11:46 The very simple solution was obvious ways of both could become zero so of course if you plug zero in this equation so of course it's always correct. So this is what I call trivial solution. 15:11:54 But then there's also possibility of oscillations of prey predator populations. So here let's just, you know, this is computer simulations. So the prey, so the population goes up, and it's pretty goes up, the predator can grow, could they you know they 15:12:13 food to eat, and the growth of predator will bring the prey population down. And as the population goes down the predator will will come down. 15:12:18 And in fact, if you compare the simulation results with empirical observations of hair links as a classic data set. You see features of oscillations of praise editors, so also wanted to ask, wanted to understand what what could he created this kind of 15:12:40 solitary dynamics in the laboratory. 15:12:40 So he, she did some homogeneous microcosm that means he shakes the cube. 15:12:45 And so in this system, he uses bacteria, he used the bacteria to feed the pyramids here, and use permits you to feed a titanium. So here, the bacteria is in excess. 15:12:55 So basically, you can regard them as it being constantly. So we have one, pray, pray and the predator. 15:13:01 So what she observed was that the prey went up, and the predator came up, but in the end, both crashed to zero the trivial solution, you don't get any oscillations. 15:13:12 So then she asked her, Well, how could I rescue this or how could I see the non trivial solution using this prey Predator System. So he reasoned that maybe in nature. 15:13:21 There's their migration events, so he tried to reproduce migration in a laboratory. 15:13:26 So he decided that he would just periodically she would introduce you know one prey one predator into the system, and indeed he would be able to recover oscillation. 15:13:37 But then he asked me well what about if I don't have migration, could I introduce spatial refugee, we are priests can hide from predator and see where that could weather weather, also notorious dynamics Could it be observed. 15:13:50 So he indeed introduce the film in which the prey can hide. So in this case it's here to observe kind of so this is what I call it by heterogeneous environment in the sense it's not well mixed. 15:14:01 So at least for some of the microcosms what he observed is that what is the prey were able to hide until predator died off and at this point, we would go up. 15:14:13 So appraised have a chance to survive, although he did, he did not observatory dynamics. 15:14:18 So maybe a decade later, half got picked up this question, but he said well I just want to have a spatial structural seem to be important. So he tried again experiment. 15:14:41 it was at the primates and the predator mites and he was, he was looking at the prey predator dynamics. So here in this setup, you will have a four oranges, and then the rest of the rubber balls and the mice can migrate. 15:14:47 The point is that they don't just start everywhere and consume the prey predator can catch the, the thought was that he would put prey predator into one orange, and then somehow the migration the ability of these prey predator to migrate to different 15:15:02 you know niche, different locations that would allow this allows us to a patterns. 15:15:10 So what does the opposite of what he observed over that prey went up right and the predator went up and come down but there was some migration. But when the prey migrated when prey predator migrated to the different orange to pray was just not cannot 15:15:23 divide fast enough to avoid in completely predicated by a predator secret zero again trivial solution. 15:15:31 So the next experiment of Cartwright was this, so he had even more elaborate spatial restructure the environment, you know says he has, you know one 20th orange times 120 positions, and he will have barriers so he reason that that if they migrate to fast, 15:15:45 fast, it's like homogeneous a month, and then you don't have oscillations again. So she would have this Vaseline barriers and the trade barriers and he would have set the fan, too, so that the pre might can migrate more easily. 15:15:56 And it was, it's actually very entertaining to read this up. So he actually was able to observe this auditory patterns, and she argued that by utilizing the large and the more complex environment, so as to make less likely the predators contact with the 15:16:10 prey at all positions at once. It was possible to produce awesome. 15:16:30 observation that you would observe in in nature. So I so now like. So, I just want to ask, what do you think might be different if we consider microbial communities. 15:16:39 Could that those the mites know these pessimism that's all fine but if we transition to microbial communities, what do you think would be different. 15:16:59 Yeah. 15:17:10 Very easy question. 15:17:10 No lines that yes, Daniel, Daniel. 15:17:24 Because what weaker demographic No Yes Okay, Very good. And then there's another reason. 15:17:28 Each of the modular, and then what, much larger numbers and then what, why would there be much larger numbers matter stochastic effects. 15:17:31 Yes. Okay, good. So, um so this. 15:17:35 Okay, so a large, much larger numbers, because there's a standing variations right so the mutations, of course can rise during growth, but the mutations, with large population size they exist. 15:17:46 So then if you have ecological interactions that can certain mutations will be rare very rapidly selected for. So this is an example from an esky formulas case work. 15:17:57 So we have their faith and sensitive bacterium so initially you can see this in chemistry experiment. And the very rapidly, you will see the rising of the partially resistance equally, followed by completely resistance Ecolab, but it always sensitive 15:18:12 equal so that's why the age population is maintained. You see, extremely rapid generation of diversity in very short period of spam. 15:18:21 So the author was concluded that resource level, the degree in the cost of federal resistance, whether resistance can be countered by fate can affect community structure. 15:18:31 So these are. 15:18:36 So these, these examples illustrate right how you know like, the least in my mind how these studies have contributed to our understanding of basic questions in biology. 15:18:45 So now I want to switch gears on our the lessons I have learned by working with a synthetic East community, and I want to mention Andrew Murray slab also has done very similar work with a with a different type of synthetic least cooperative community. 15:19:01 So we constructed the community to understand the. So the first part of the evolution of cooperation in the second part is, is higher learned the importance of quantitative thinking. 15:19:14 So cooperation is important, because it is thought to drive major evolutionary transitions from independent of applicators, which are just pieces of nucleic acids that are capable of self replicating to being stitched together into a long chromosome, 15:19:30 where the applications are no longer separated other other coordinated. 15:19:35 The next example is from unit solidarity, to multi similarity. 15:19:49 And therefore, individuals to society by each submission are Smith argued that each major transition evolution involved. this, this cooperation between lower level units to higher level. 15:20:04 But the problem is that each of this transition there are cheaters right so in society we have thieves, that take but not give. And what you said already that cancers right that break this break the rule of multi sorority. 15:20:10 So instead of having a strength growth, they proliferate uncomfortable uncomfortable fashion, eventually destroying the host and themselves. 15:20:21 And so in for this, the transpose of that can jump around the replicant right so again they break free from the rule of quantity replication. 15:20:29 So then the question is that, how much cooperation survive tutors. 15:20:35 So I want to actually give more rigorous definition of cooperation cheating. So let's consider net gain. 15:20:42 And if there are no cooperation, I consider the next game to be zero. That's baseline. 15:20:47 So a cooperator pays a cost see to generate benefit be greater than See, he's a cost see generates a benefit. 15:21:04 If it in tax It was another cooperator, so the net gain would be d minus c right because it against me and paste costly. And because these greater than c b minus c is greater than zero. 15:21:10 And because p is greater than c b minus c is greater than zero. And cheater pays in the for simplicity, I would just say argue that children pays zero costs and it generates zero benefit. Whereas in reality can be quantitative. 15:21:20 So the tutor is. 15:21:38 So this will allow students to increasing frequency on the end Cheetos interactive with tutors, so we are back to the, the no cooperation baseline of of zero. 15:21:50 So we can see there are two constraints of cooperation. The first is partner availability. If you cannot find a partner, then this will not be profitable. 15:22:01 The second is to do in Beijing, so focused on the second constraint in this talk. 15:22:07 So how much cooperation, have evolved to cope with cheaters. 15:22:11 So what does the audience think, I mean just intuitively speaking you have operators Cheetos to take advantage of operators on my the cooperative systems survive cheaters. 15:22:26 Yes. 15:22:30 So, the memory of past interactions. Yes, Very good. Yes. 15:22:44 That's very good so special. So let me just write down right so one is like memory 15:22:57 space. 15:23:01 And what else anybody else. 15:23:06 What did you say seems as a paradox yeah so 15:23:13 yeah so spiritual structure, in some sense, oh so is it a single self Pollock. Exactly. 15:23:19 Yeah segregates those spatial structure in the sense where you have to have that. And yes net mean there's definitely cases where evolution is coupled cooperative behaviors to sort of essential functions. 15:23:31 So, very good. so I called style topi. 15:23:38 So we'll just give these, I'm just. 15:23:41 Yes, pal just I want to elaborate on that point. So Kevin foster showed right so it's one example of the stadium. So these are social Amoeba when they stopped the salsa will aggregate. 15:23:54 And then the salsa will differentiate into pre stock and the pre sport, but then he will differentiate further will become stock, and the sports sports can survive starvation, they can come to life again. 15:24:05 When you turn off the price for the stock is basically that. 15:24:08 So what we found was that there's a there's a gene, right, the mutation in which the name is not important but Kim A. 15:24:22 So the main mutation would do this so salsa was done a would not listen to the differentiation signal, so they will preferentially get into priests for region positioning themselves into supporting, but the problem is that they cannot differentiate, we 15:24:30 finish the differentiation into sports. So, basically that gene coupled self serving traits with partners of intrigued by self serving trade in the sense that the ability to, to develop fully into sports, the partner so in the sense of being able to listen 15:24:43 to instructions and get into stock. So that was one example, so proud of in a quorum sensing you can have you so far back to you and quorum sensing you could have started responding to the to to the other end users, and as a transcription downstream transcription 15:25:02 units could include the genes that make public goods and the genes that make private goods, so that's another level of coupling so that's one example. 15:25:14 And for special structure, I just want to elaborate a little bit more so what happens to the following right so if I'm cooperating with myself, for example, we're just collectively degrading some substrate. 15:25:19 So for specially structure, as I'm standing here I'm giving birth to my children, if I'm a cooperative, my children tend to be co operators, and then we can we can reap the benefit of cooperation, but only if I will cheater, I'll give birth to cheaters 15:25:32 and we don't we can zero, right so that's how special a structured environment can stabilize that. And for memory there's also related and so it's called a collaborative partner choice. 15:25:47 So if you can recognize why so if you have to. So if I were to the the chooser I'm looking at two individuals, and I can recognize what this is cooperators cheetah. 15:25:56 And I will preference will interact with a cooperator and ignore avoid or even punish students, then cooperation can be stabilized. So these. 15:26:04 so that another action, another mechanism is privatization. 15:26:11 So Jeff Corps was studying. It was studying use the word taste. So emo taste. So what happens is that support the East has sucrose, that's the, this is by sugar. 15:26:24 So the release emerges into the environment. so that can break the to sugar into tomorrow sugars, and then you can pick up these models. 15:26:44 are freely available for everyone. So what Jeff showed was that he was small, even like 1% privatization to preferential access to the model sugars would give them enough to offset the cost of cooperation, and allow operators to coexist with cheaters. 15:26:50 So that's another, another mechanism. 15:26:56 Yes, it is a little bit like spatial, but in the sense you can have still will have well mixed environment. So the encounters of. So the, so the cooperative students will still interact with each other in English like sort of like homogeneous kind of 15:27:09 In, not like sort of like homogeneous kind of fashion, but it's just, you're right i mean there's this member and by that allows this preferential access to this creates this spatial structure. 15:27:19 So um so what do we, so we wanted is that we actually wanted the system right to avoid all these. So we want to see what are the other possible mechanisms that could destabilize cooperation. 15:27:30 So we want a system that can get rid of all of them. 15:27:33 Right, all of them. So the way we thought of doing that was to have to engineer our own East cooperative community. 15:27:41 So the ones. So the, the first string cannot make a live stream because we deleted one of the essential genes for making slicing and overproduce the add any, and as this is due to a mutation in the feedback control. 15:27:54 The first enzyme of the earning power synthetic pathway. So as a result, the, the enzyme is no longer susceptible to any product feedback so as a result, the cell just overproduce over produce a lot of these metabolites, and the license, adding both essential 15:28:10 for ourselves. 15:28:11 So then the the, so I call this, we call this L minus a plus because cannot make license in overproduce adding under the complimentary string is just opposite cannot make adenine overproduce slicing and dicing, and in fact it's not add any it's a it's 15:28:28 a derivative of agony called high presenting is released, but for the, for the purpose of this talk, let's just call this avenue. So these over released from the tablets are releasing into environment yes Terry, 15:28:45 that I do not know but I do know what is released, is no saying hi presenting. 15:28:54 So now we have not done, we have not, we I don't believe we have done comprehensive analysis of the metabolites. 15:29:06 Yes yes yes yes yeah okay so uh so these are nominating, so can be regarded as different species. And we of course label them with different fluorescent proteins so we can track their dynamics. 15:29:20 And then we can also make cheater so so those ones cannot make an icing just like these, but unlike those these just the blue cells just make enough just for itself doesn't release anything. 15:29:30 And so Adam wait the demonstrated that indeed there is a cost to over to metabolite overproduction, so he can compete cheaters with CO operators, so those ones do not over produce and those of us that over produce any excess license you will see that 15:29:44 the ratio, the cheaters will take over. 15:29:46 And it's about 2% finished manage, and you can also show last thing overproduction has been this cost. So this is a real, it's a real cooperation cheating system in the sense that cooperation increase a fitness cost. 15:30:01 Someone argue that Cosmo is a model of naturally occurring cooperation. 15:30:20 And because neutral exchanges are common between legumes and the rice or via, and the state between hosts and gut microbiota, and if we generalize nutrient exchange to benefit exchange, we have many more examples for example between fix and fig wasp where 15:30:22 the fixed provide food for last. And the last provide pollination service to fix. 15:30:29 And so now let's think about these right so these are natural systems out there complex highly complex, they have been co evolving for with these for these two will co evolving for long periods of time. 15:30:40 And so, but if we look at this way. So for example I contemplated about studying the system but then I asked myself how can you get rid of spatial structure because that is the intrinsic part of the, of the biology. 15:30:53 Right. And again, the second reason is that, because these systems are a natural right so I we if we studied these, some of the lessons we learned over the be would be perhaps general or some of the lessons will be specific to that system. 15:31:08 So what I wanted was a system that basically captures the short feature of all the way which is benefit exchange. 15:31:16 And then hopefully we will learn gentle. 15:31:21 Yesterday 15:31:25 you described the natural systems are very elaborately resume right yes the interest is a lot of things no signal and so it was set set it up. I would say probably more elaborate than a host, Michael. 15:31:39 Yes, yes. 15:31:44 Are there many examples I do not know of these, you can fill me in the natural examples of the similar more similar to the type of just obstacle for you those answers will be that that that the live together. 15:31:56 Yes, this is interesting because if you look at, if you look at, like, say, obviously oxytocin is actually far more common that I had initially appreciated. 15:32:05 When I started project were also talking about it about the stars over minimal genome. So seems to me that it's not, it's not a like a laboratory artifact right I want to say you know this of course we engineered it, but I would argue that, you know like, 15:32:18 it's very common to see vitamin oxytocin in naturally isolated isolated microbes virtually that that's a very special I mean I don't mean to be 12. Non pathway and all that but the the but that yeah that's that's what I'm studying itself the. 15:32:37 But, but, the, the, let's say you know amino acid nucleotide is kind of more common metabolite sort of mutual reliance yeah so it's like an amino acid oxytocin was also observed and the Kevin foster I think we are I mean, found that in gut gut microbiota 15:32:57 some sort some organism I don't remember what pays a cost pays a cost actually it pays a cost to generate something for the neighbor. 15:33:05 So this is still some metabolite. So some of the provisional metabolites or enzymes that generate the metabolites, but it's just some paying and possibly generated benefit apartment, but we'll just say these are the real good consistent us to study I 15:33:23 I mean just as to study the same type of issues. If they naturally arise and still just involved, so to speak. Yeah, but I want to say that even for those systems were were were to happen, because the exchange the metabolites are just so many, so there 15:33:38 are other synthetic systems of like photosynthetic or bacteria was some, some other organisms under the Mondo metabolites exchange, so numerous. 15:33:48 So it's not just the one right so now we'll get to that part where the quantitative like this allows us to think really clearly, it will be numerous in your system too. 15:33:58 But you okay okay yeah for you but the but yes just hold that thought for 4444 minutes, was the reason I'm asking this is the social me a question to understand when you have this kind of naturally occurring system that's a good question is why would 15:34:13 you want to subject yourself to be relying on somebody else well while you can easily pick up a gene to do. 15:34:30 But I would argue like a driving force yeah yeah so like before the indoor same symbiotic bacteria right so they live actually inside insect cells. so they will have very elaborate metabolite exchange. 15:34:53 So when with regard to general principles we also have the help of mathematical models so so that helps us to ask on a variety of conditions so what do want to be observed for species coexistence. 15:35:05 So we show that most non cooperation. Most cities coke business, and in the system in our system we can start at the ratio of 1000 to 1121212 1000. And if we just keep culturing them. 15:35:18 The ratio goes to around the one, and this ratio is actually dictated by supplying the consumption. 15:35:23 And the reason for this call existence is very intuitive, because if you can imagine if the ratios are lopsided now. So now we have huge amount of this benefit in a very little of that benefit. 15:35:35 So these souls will have lots of metabolites so they can grow fast, and this one will be competing for very little amount of metabolite so they will grow slowly. 15:35:43 So then this is what the rebalance the growth. So, and that's why I was saying for a couple of the systems, both species in that community grow, it's identical identical growth rate. 15:35:54 And this has been observed in my thing. Method generating community and the methane oxidizing community. 15:36:04 So the second the second the general principles is strong cooperation promotes species into mixing. So argue Of course there are other mechanisms to give you into mixing right so if they are clumpy, they will intermix, but we will asking that if two types 15:36:18 of cells, only intact, to the influence of fitness and nothing else why not Columbia not the not the chemo taxes and so on so forth. Then what happened to the patterning of the community. 15:36:31 So in the simulation in simulations we showed you that if two types of cells strongly cooperate, in a sense, by strong cooperation I mean that in the presence of the part of the partner could lift your fitness, far above your base or fitness, that is 15:36:44 strong Corporation. 15:36:46 cooperation. So we can see these. The intermixed pattern and but if we we forced them to compete. 15:36:54 Do we see these patterns, so I would, so for cooperation so these, so these simulations experiments. So in experiments we do not add, add in your license, so they're forced to cooperate. 15:37:04 But in the competition, we can add access at any analyzing so they're forced to compete. And then for competitive communities. We see columns of cells, and the for cooperation, we see mixing and intuitively, this intuitively, the way I think about is 15:37:21 the following way. So if you have two types of cells. Right, so the, let's look at the vertical city the vertical cross section, but it will process. So he's just part by done do chemo taxes, it doesn't move. 15:37:28 So, but it was randomly. 15:37:34 If this, if this right is rely on that. So if this cell happens to be away from the partner, it's less likely to but, again, compared to if it had happened to the partner, so that would cause this one to power over on top of the other and and then you 15:37:49 can reverse process happens and so you can pile on top of each other. So that is why, start incorporating self intermix. 15:38:00 And those. So as a comparison to our engineered systems. So we have a software review on the mechanical caucus. So here are the two, the tool, microbes corporate attribute What is this actually not, It's not competition plus mutualism in the sensitive 15:38:16 both stimulate each other's growth. 15:38:18 So the software review in the presence of lactate, it produces hydrogen. The hydrogen is inhibitory to to dissolve a bill. 15:38:27 So, the hydrogen can be used by my paternal caucus is reducing power to generate methane, and by consuming hydrogen. 15:38:37 McConnell Cox promotes the growth of the sofa video by removing inhibitory effect so these are mutually helping. 15:38:44 We also see intermix the patterns. 15:38:49 So the first thing I think advantage is that by having some system that captures essential features of of cooperative of mutually beneficial interactions, we can learn some general principles. 15:39:03 Okay, so then. 15:39:05 So then, so the yes to evolution cooperation so we, we, so we have this Cosmos system, we want to, so this really allows us to get rid of memory, right, because these are engineers so they, they cannot have twice or memory. 15:39:19 We can get rid of specialty structural environment because we can shake it very well. And the power Toby because they have no evolutionary history so we also get rid of that, and the privatization. 15:39:30 It's not there because what you need is your partners, but tablet not yourself. So we got rid of, we could this system can get rid of all existing about all know mechanisms. 15:39:40 So we asked ourselves so if we have these three. If we have. 15:39:44 If we're well mixed lacking these supplements, so that forced to cooperate with each other. 15:39:51 Then if we just grow them right and then we presume that cheaters will rise and the tutors in our laboratory experiment we know have fitness advantage of cooperatives. 15:40:02 And we dilute we repeated this experiment. What do what do we see, what do we see all cultures, which is crash on that. 15:40:10 So admin waited to this experiment he makes the three strengths at 121 to one. And again, this one has finished advantage over that one. 15:40:19 So I'm going to show you. 15:40:22 Community Dublin's against time. So for the community grows exponentially, you will see a straight line. 15:40:29 So we, we saw is that it's actually quite heterogeneous kind of outcome. 15:40:36 And in fact, we can color them into two groups, the orange, enter the gray. 15:40:47 The orange have fast growth wait in the end, and the great flow. We can, because these three students are colored recently labeled, so we can actually look at to the compositions of these of the regional types, we know these festivals are dominated by 15:40:58 the cooperators, and the slowest are dominated by cheaters. 15:41:05 And then we can do genome sequencing on these. So we actually saw the, the bottom lines of falling, so we have cooperators tutors, they are competing. 15:41:14 They are competing for this very limited licensing released by the partner. 15:41:20 And so if she does happen to sample a mutation that is beneficial to growth in limited licensing, then we end up with to take over so it's this outcome. 15:41:29 And the same way, if the cooperators happen to sample the most adaptive mutation, then cooperators will take over, we end up with this. 15:41:38 We call this adaptive race versus cooperators, and cheaters. They're both competing for something, and then whatever is best happens to sample the best mutation would would win the game. 15:41:52 So, that patient to stress can lead to stochastic to the project, right. So, Adam now adaptive grace. 15:42:02 It's stochastic, it's not for sure but it gives chance. Yes. 15:42:07 It's a quick question. 15:42:08 Should I understand correctly that the cheater doesn't need a beneficial mutation to take over if if everything stayed the same. The cheater would have one already but yeah right and he is the one that really needs the mutation, no changes so so the the, 15:42:22 so that's a good question. So the difference between these beneficial mutations. 15:42:37 So in a sense that, in the sense of a cooperative so overwhelmed that this one had to overcome the cost of cooperation. Right. Yes. And in fact, if we started with, you know, was it was, was 10 to one ratio. 15:42:43 So outcome will be 10 to one those we're determined by our initial population size, and just out of curiosity, what is the fitness effect, fitness, so these, these fold several fold compared to ancestor, compared to, now we have more of these to say. 15:42:59 So if we look at it so this also happens in our engineered microbes, so this and Berkeley Lab showed that, you know, Pseudomonas fluorescence are an iron limitation, because operators will pay a cost to make kill it are called zero force. 15:43:14 So release their force right and then that would be able to bind to iron at very very low concentrations. And then these complex will be taken back in ourselves. 15:43:24 So, so of course if cheaters around they will compete with cooperators for this, even though the cheaters do not contribute to their Of course, it's really funky problems so they have. 15:43:34 So they saw with cooperator cheater, both at 1% the two different experiments. 15:43:39 So if there is nothing just a regular experiment, because cheaters have enormous fitness advantage over cooperators sense, Cheaters will increase as you would predict. 15:43:50 But now the fH. 15:44:04 So now, since Cheetos actually decreases. So this is a example of adaptive race in are engineered crops. 15:44:11 Yes. 15:44:24 Here's some kind of qualitative difference between something like cytera force versus something like amino acids, versus, I mean like Terry said before like vitamins, because they therefore is also i mean that's iron that's in trace amounts. 15:44:38 Do you think that that's just a question of a concentration difference or is this like is there some like deeper difference. 15:44:45 So they're all full can be recycled. 15:44:49 And then the metabolites. 15:44:50 Maybe when they die they release it but it's much harder to recycle these metabolites, so I think that is the one difference. Second, think of 15:45:00 the cost is different way so this cost of their studies seem to be much larger than the amino acid overproduction our system, but the room stay the same. 15:45:09 Yes, I would, I would, I have not thought about being able to compete with equity versus not a psycho the way I think that could make a difference. 15:45:18 Yes. 15:45:22 So here by using a system that allows us to get rid of all the know mechanisms. 15:45:28 We couldn't discover new mechanisms. 15:45:31 And the third is conceptual clarity. 15:45:34 right now want to tell you a story. 15:45:36 So now I'm so getting back to Cosmo right so so the question in a very basic question how do you quantify partner survey phenotype. 15:45:44 Right. So, again, I would like to draw an analogy. So the rise over the goo goo goo mQy Sophia follows in faith, and the rise will be fixed naturally to the goals. 15:45:58 Right so okay so if I move on to the ancestral Razzle via if I'm a mutant. 15:45:59 So the field is the literature says, if I don't make any library of course I'm, I'm a cheater, that's given right if I'm not making any digesting to give to the partner. 15:46:09 Another question that. What about if I make less. If I make less nitrogen, then I'm a cheater to its quantitative. 15:46:15 So we say okay so we just want to see whether those involved is mutants would actually release less or more than answers. 15:46:24 So, so we in fact we're looking, whether there are different mechanism in the sense that whether you could even have over released phenotype, that's what we were actually looking for right so we focused on this L minus a plus. 15:46:38 So then, by Same, same logic right so popular survey equals to release more adenine than the ancestor by myself, if I release more add any for the partner, they're more hardware serving compared to if I release, you know the ancestral. 15:46:54 Right. Okay so good. 15:46:57 Yes, sorry. Um, so most raise me at that fixed last nitrogen actually have lower fitness. So, excuse me, move via that fixed no or less nitrogen typically have lower fitness yeah that could be cloud Toby, but I'm just saying the field quantify because 15:47:13 when they isolate this cheaters is problematic. Excuse me, I think calling, calling them cheaters just because they're fixing less nitrogen is problematic. 15:47:28 Why is it problematic, because, well, so the way that that I've defined cheating, is that it is the impact on your partner's phenotype. So to be a cheerleader, you know, so if you. 15:47:35 Yeah, I have a paper for many years ago, where I argued that rather than being defectors right like everyone talks about these Rabia cheaters that don't fix nitrogen, that they're actually just effective and they don't fix nitrogen for reasons completely 15:47:49 unrelated to increases in fitness in the symbiosis. So you may be need to factor that do not engage interactions, like loners, you're talking about the loners that. 15:48:01 Yeah, yeah. So, similar to loners or you know or they're just on the wrong host or they have a deleterious mutation or, you know, there's all kinds of reasons why something might not work in symbiosis yes yes do with it cheating its host yeah but. 15:48:14 So how would you propose to think about, yeah so but I'm asking right so i over the counter by asking the question, how would you quantify partner survey phenotype. 15:48:23 It's a couple of the system by how would you quantify partner serving. How would you judge whether this mutant is more partner serving and ancestral not by growing all the different combinations together and comparing them combination together in and 15:48:36 compare compared them, compare the, the, the, yeah yes if you have variation in Libya so what people typically do is get a bunch of Arabia grow them all individually with a host and then compare the host benefit provided by each of these particular strains. 15:48:53 Yeah so so you so you can tape measure or some metric of the recipe of fitness right and so then you can get an estimate of the host fitness and the Rabia fitness and look at those relations yeah so yes, you see you say, arguing that you should measure 15:49:12 the the the the the the the, how it does work. Yeah, so Okay so, so, so the. So we have a different way of looking but eventually we'll get there get there, right. So we. 15:49:17 So we. So we just measure that. Regardless, we need to measure release weight so we measure the release rate right so we we actually find a mutant that will release at a faster rate than ancestor. 15:49:27 So turns out that mutation is a chromosome duplication. 15:49:32 Right, so it's chromosome 14 duplication and that is, I guess, one of the Vantage so working with East is that you once you have a mutation, then you will be able to identify the duplication which gy because the other 500 genes are the which gene or the 15:49:47 causes over released phenotype. 15:49:50 So at the top chromosome so so the duplication of chromosome would give you over so we say we chopped what chromosomes two different levels, when we lose now this over releasing phenotype. 15:49:59 So it took us a year, right at the end, what do we find that it's a duplication of gene that made itself, larger, 15:50:09 and was very unsettling to us, right like Mid South larger I mean, what does that mean so immediate came to us maybe those sorts of consuming more. 15:50:17 So if you can, if you have two cells, right, you can have two cells new release in normal model you fuse them, because the rate is calculated at a personal level, so you would see over release but as they actually consuming as twice as much as well. 15:50:29 So we measured consumption. So after we normalize, right, the release rate over consumption. There was no absolutely no difference for me. 15:50:37 It wasn't like that moment I felt I felt I was almost compelled to say this is a partner swimming mutation but because this is so tractable that allowed me to think, think carefully and then we now realize why so this is Exchange Commission and getting 15:50:51 back to this to your comment. So in fact, if we have thought more mathematically at the beginning we will have avoid all these right so the mathematical says the community is a steady state community growth rate. 15:51:03 As I called this back up one more step. So this is exchange ratio right exchange ratios adenine release rate overlays and consumption. So basically it's that you release at any you have to have lightning consumption to make the cell to release the right 15:51:17 to add any. So this rate. Right, so this called exchange ratio, and for the other partner that's the other exchange ratio so basically you have lightning release rate over at any consumption and adding race, race, overlaps and constantly. 15:51:27 If you to square root, the units the correct the right so it's actually the rate. So its growth rate. 15:51:33 And so then we realized we should have done this, so instead of by. Now I would argue that I would say that papers I've seen. They just say, well, you really high release rate means more part of me that I can have references and it intuitively, it made 15:51:47 it made sense but retrospectively. 15:51:50 This makes so much sense right because it is it is a feedback system so you have to do that. 15:51:55 So force us to be conceptually more clear. 15:51:59 And finally, our CEO asked us to probe questions that are very difficult to address in natural systems, and it gets back to a net point of cloud to be so proud topic change, a linkage between a self serving trade and a partnership in trade have been observed 15:52:13 in systems, was not evolutionary history of cooperation like that. Electric the stadium across sensing. 15:52:21 So the things that can we can such linkage arise easily to early stages of a corporation. 15:52:27 Right, so then if you buy natural systems, by definition, they have millions millions years of evolutionary history. So we asked in the system could we actually get that. 15:52:37 Can we get that right so again we focus on this. So the self serving so personally I just explained why to have partners who may change you have to have higher release rate over consumption. 15:52:48 So we have settled that before self self serving changed. So what does that mean by self serving as a you have to have a mutation that allows you to grow faster than your competitors have the same population that's what it means. 15:53:00 So in fact we this is actually very easy to do because if you grow them in environments unfavorable for corporations, such as the well mixed environment. 15:53:08 Right. So then, natural selection always select the, the fitness of the fastest growing was. 15:53:13 So here we. Here I'm showing you growth bait against concentrations of lysine could either require licensing. So these because it's for us simply labeled, we could actually do 96 Well, access to subject them to different concentrations of licensing and 15:53:29 the measure of the growth rate, and we can actually calibrate this, so we know this is correct, it's not, you know, it's not limited by this consumption because we could, we could calibrate this against a different way of measuring growth rate against 15:53:41 the license, so we're very confident about that. So ancestral, you can see the ancestors not normal node, but the sigmoid curve, you can see that. 15:53:49 And the four foot. So this is one type of mutation and these other types of mutations so every single mutant closely isolated. They always grow faster than the ancestor in purple at low concentrations of licensing, that's representative of the community 15:54:04 environment. 15:54:05 But so and then we have these are just showing that you could you could engineer the mutation to bi, to be similar to what's been involved across and then recapitulated, you know, so these are sufficient for for for this for this advantage of enormous 15:54:23 is. So, by a pen is the advantage is just enormous compared to ancestral retro several fold difference. 15:54:32 So, by pen. This is an advantage is just enormous compared to ancestral that's a symbol for difference. Um, so, so this is self serving so all the clothes are self serving, but the questions that can some of them, the partner serving, and it's there's no reason why you 15:54:40 think they will be part of the survey right in fact that they could even be more selfish in the sense that they can release less power consumption. 15:54:47 So we will find that that we find that that one clone. 15:54:51 So the CCM Tonya one which is very frequently observing a rising mutation. They could have, they actually have increased release power consumption. 15:55:02 Whereas another one has has no changes. So it's not okay. It's not like a mutation, all sorts of changes would be assuming but some of them. For some reason, which we don't understand a partner serving fitness benefits are so big. 15:55:17 How did you just not fix those mutations and both subpopulations to begin with back at the beginning, we had the sort of racing, I see. So these are not fixing up. 15:55:27 I actually they are double mutations later. So this is early stage, and we have not looked into the very latest status of that. So, we find that these very common because they they arise so quickly. 15:55:39 And what do we expect those the cheater and the ancestor to get them almost immediately at the same time. The. 15:55:46 Oh I see, so So for these for these, they're not they're just to the soda no cheaters there. So in fact that we were looking for naturally arising. I mean naturally in the sense that in the test you write the, so we didn't put engineer Cheetos in there, 15:56:07 I'm thinking back to your original cheater. Why don't you just get to ECM 21 mutations in both backgrounds immediately. I actually do believe because these are just isolate so these so they actually if you look at the genomes and multiple mutations, and 15:56:13 the multiple actually chromosome duplication. 15:56:15 So we went for the, the ones that are easier because it's harder to work with chromosome duplication so we just worked with easier ones, but they are phenotypes I'm sure are different because otherwise we will not have seen the stochastic outcome. 15:56:29 Do you think the stochastic calculus because they're getting different kinds of, yes. 15:56:41 fitness advantages. Cool. Yes, Terry. 15:56:43 change ratio. Oh extremely which also release rate for consumption. So, what, what do you actually measure. Okay so, so for example for for this one. 15:56:52 It's adenine release rate. The femto mo at any release the Purcell per hour divided by, comfortable life and consumed to make a new cell number so how do you, whether it was that mean what do you actually do. 15:57:04 Oh well, that's a very good point, I will get to that at the end. In one slide I will get into slides I'll get to them for details aside. 15:57:11 I'm still going back to my earlier scenario where there could be another hundred metabolite that's released, there's no you cannot do. It was very good American, I'll get to that, I'll get to that. 15:57:33 Okay what what proportion of mutations have effects on things like trade ratio rather than things like you know cell size or other things that might have other adoptive effects within the environment. 15:57:46 Um, you mean. 15:57:47 So cell size like are you like are you just pulling out. Okay so, so the chromosome 14 duplication. Yeah, so that's a good question so chromosome 14 duplication happens to be also self serving, because the licensing premise is duplicated, it's also incomes 15:58:02 chromosome 14. 15:58:16 And that is sold right so initially we thought okay so the duplication of realizing premise made self service duplication of some other g made partner serving, but turns out it was just, it's a duplication of cell size the way I just select cell cycle 15:58:16 cell cycle inhibitor. So basically cells will not start dividing until they're bigger size and. for this you have to consume more and it's just right i mean basically our thinking. 15:58:26 So that made us realize our thinking was wrong initially. 15:58:31 And so this it says, Why so we. 15:58:34 So this is shows that shows that so proud to be getting back to proud to be the definition of cloud Toby, is that a single gene can control multiple phenotypes, and it's, if you think about it's actually not hard to understand, because for example one 15:58:46 kinase can force for many substrates and if they different substrates have different phenotypic consequences, you have proud of it. So here which is showing up for something like well because the environments. 15:58:56 Right, was no was no evolutionary history of cooperation in fact we can evolve monocultures have this, so they will already have mutations that would allow them to better adapted to this lysine limited a moment but also release release more metabolites 15:59:14 power consumption. 15:59:15 So that is a but not all, not all mutants do that. Some of them do that. 15:59:20 So it's, it's a chance event. 15:59:24 And again, I want to say this is not even though the the studies are different and the level of resolution different but because this study shows it has nutrient uptake is sufficient to drive emergent cross feeding between bacteria and synthetic community, 15:59:36 bright and early adaptation your microbial community is dominated by mutualism, he has a mutation so there might be another another mechanism in the sense of this, this, this, it's not that. 15:59:46 It's not like the Cheadle mutants and will always automatically arise the crashes so it's not as it's pessimistic, as the, as you know as maybe the field had thought about. 15:59:56 So the final story is a quantitative importance of quantitative thinking and she has almost maybe it's not necessary for yourself talking to physicists, but for me it's like it. 16:00:05 It is a stage of growth I want to say this is a stage of 16:00:10 intellectual growth for this way. 16:00:12 So I actually went to a physicist lab to the postdoc, because I want to see whether math or physics is an all useful for you as a biologist, I was trained as a molecular genetics. 16:00:23 If you look at the successful examples of math invalidity, you can count them with one hand right it's not that many examples. So it was a very few vs when I went there I saw the well just give it a try. 16:00:32 Right. And in the sea. 16:00:34 And so I thought, well, like if I want to understand the system. 16:00:40 I do know it's important to understand things quantitatively right because otherwise you know no one, the GPS is gonna stop working because I do know that it's important, but how can you be quantitative. 16:00:47 And I remember my advisor was actually very against the quantitative modeling, and I do notice like lot of his to quantitative modeling bicycle qualitative behaviors and the, you know, it's in its robust against details solar system and so forth. 16:00:58 But I thought, well, if I engineer the simplest system right if I cannot quantitatively model that like why would I trust those models of like really complex systems like there's no reason that I could if I have to. 16:01:09 I just want to try and engineer cosmic actually for half of the reasons for that. 16:01:14 And then, if you do this right so it's actually, it's actually so so my dad would say okay, if you have three parameters, you can do anything you can just fit, you can just play with the amateurs match your experimental data, but that's okay fine, I will 16:01:27 not do that, I'll measure every single parameter, and I want to get this right. So that gets to steady state community growth rate. So it's basically for parameters to release rate to consumption. 16:01:49 Now the other point I want to mention just like theorists were when you do theory you make approximations, we experiment with also to this we have to do that. Otherwise we go nowhere. Right, so, so forth measuring these. 16:01:52 So we say okay so what we'll do, because it's easier to measure everything in batch culture. We're going to measure the batch cultures the consumption will give them will give them access, access will give them different types of metabolites in metabolites 16:02:07 with different levels so we'll see final density and then we can calculate, right. If I give you this little you grow to that much if I give you twice as much work twice as much. 16:02:21 Now can I can do rap regressing, and it gets consumption per cell for release wait. 16:02:22 Oh, you know, we know that we should actually probably measure this in community like environment. But for simplicity, will just measure in starvation condition and these cells happened to still release, so we can we can look at the release debate, and 16:02:36 in the tutorial that I will give maybe next week, I'll actually go through in much greater detail about, you know parameter measurements experimental measurements of model parameters. 16:02:46 So did all these measurements and. 16:02:50 And as soon as we look at the ancestral Cosmo growth Wait wait so this this things seem like huge difference. 16:02:57 But because growth is exponential. You will small difference in the in experimental model discrepancy. 16:03:05 He was small discrepancy would expand exponentially why when you look at population dynamics. So it was really was very very disappointing. 16:03:14 So I thought about this as well. There was it I cannot I cannot really understand even such a simple systems. I cannot even get to the quantity of understanding Yes. 16:03:23 What is disappointing what is no this model, an experiment, it's a 60% difference and this growth we were talking about growth, right. So, if you look at the time, based on that cultural measure, measure yes no three parameters in in the in the stationary 16:03:38 industry you know they're never in stationery because we always give them on a tablet. So, so no but so then how. 16:03:49 So, monoculture model. Exactly. A little bit of the limiting yeah yeah so so monoculture. The reason we measure monoculture is because in cultures whatever I released his immediate taken by my partner does nothing to measure, so that you have to mess 16:04:15 me monoculture for consumption we measure the different concentrations of metabolites you supply enough to keep the grocery the same know so forth for release. So for culture, how fast is growing. Yes, Yes. 16:04:17 We must chemo step right so that's what I was saying by experimental approximations. 16:04:22 I did, I know I did it, I didn't. 16:04:25 You never 16:04:34 know, we didn't, we didn't. So if we think about it that was what 16:04:35 I was saying is describing. 16:04:39 Yeah, exactly. Exponential growth right but it just like a theorist, where you make approximations I know you make approximations, all the time right so so so so it's not like it's not just like the thing committed by experiment. 16:04:52 Right. 16:04:52 approximation you're rolling. 16:04:55 Okay okay so sorry I was advancing a little too much. Yes, I mean so one reason one possibly is terrorists terrorists point right because we're not measuring them in community like environment, but to build chemistry that's highly non trivial, as you 16:05:08 probably know it's hard enough to do that's why you're asking this question right so, so, so. 16:05:15 No, I mean, in fact expensive but how would you have a monoculture in absence of the partner having this exponential growth. 16:05:24 Oh yeah I do, I do, yes yes I do, I do. Yes I do, I want to predict the 16:05:32 future you produce constantly materials you have to keep my support to keep their exactly monoculture you asking for kinesthetic slow dripping of lighting right to to to 16:05:44 have a courtesy, very nice in concentration, no so all the coaches have no license to add any supplements. You had a model, you had experiment with a prayer either know that you very nice and concentration, what am I promoted to Michael Moore and so forth. 16:06:00 Yes yes you're getting groceries. Yeah, but these are these are like us in such in such a short time span, they're not releasing enough, that's a very good point not in release enough for us to measure. 16:06:10 Okay so, yeah, so and then another reason, like another areas if you think what's kind of disappointing, but there's so there can be many possibilities why self could be evolving when we've measured these measure those rates and so on so forth. 16:06:24 So I have to give up right so I thought maybe you know, we wanted a modern empowered is just, you know, it's just it's just too hard. 16:06:33 So that one out why so we were still trying to quantify for the release rate of the, of the mutant against the ancestor. So we're just doing these measurements. 16:06:57 result. So I have the best, you know, Sam heart is just super experimenters. So he was trying to grow cells, right in the very identical fashion. So if we have the replicates with the experiment, they look very similar, but if you do them in different 16:07:07 So we're just going around circles right like for over two years. 16:07:12 Eventually it dawned on me right so if build assets, I'm not good enough to tell me this. Why would I trust all the assets. Right, so we bite the bullet made chemo stats right so that is a whole different story. 16:07:25 There's a whole different side of that right there. So to make them so that flow rate is steady enough. 16:07:31 And it's precise enough it's reproducible enough. 16:07:33 So then we mentioned all these incredible stats, because I'm sure, and the the end is really sweet and we noticed that reduced rate is a highly sensitive to growth environment, the growth conditions, and the ones we have these. 16:07:46 Now we have this. So this is a prediction model from phenotypes from chemo stats and this arrow bar is to arrow propagation of the uncertainty of measuring parameters. 16:07:57 And because of this, it will rescue the earlier experiment I talked about, because we were just so confused about what's going on between mutant that then says, and so this rescue that and the rescue the other, because once we measure compare phenotypes 16:08:07 in this chemotherapy moment expands immediately become very reproducible. So that tells us why it's actually really hard to repeat we capitulated starvation kinetics the dynamics of physiology of the cells. 16:08:20 And so that taught me kind of like ironically rightly I wanted to do quantitative modeling and I stepped away from it, and paid a huge price way for not listening to what bothers me. 16:08:32 But when you know let's say you're, you're quite lucky that in your chemo style you do not get no Usually what happens is people studied and mutations. 16:08:39 Oh yeah, no, no, we have mutations I'm not going to talk about in tutorial I mean not getting kind of late in this in this. 16:08:46 So that was another challenge where it's extremely rapid arising of mutations and there are two types of mutations, ones that slow mutation with normal mutation rate, but enormous several fold advantage. 16:08:58 So we have to cut it really quickly so that otherwise we will be measuring the phenotype of evolved Chrome. And the second one is extraordinary high mutation rate. 16:09:07 And then this one we did some approximations right so we thought if we did community experiment for the law and we did a parameter measurement for the law, we would kind of match out like a sort of cancel out to the evolutionary effects, I mean these 16:09:15 are highly non trivial experiments. So, you're absolutely correct. 16:09:24 So now the question is that, so now To conclude, of course the future direction is how might we handle complexity. 16:09:35 Right now with the welcome thoughts from the audience. The way I felt about it it's one of the future direction is to quantitatively understanding how community properties for them for the robustness of community against perturbations might change as 16:09:46 members involved. So my counter argument of okay this is too simple is a Lenski has been working on model cultural evolution for decades, right so obviously he had not even exhausted. 16:09:56 the rich, the richness of evolutionary complex the model courses. So I would argue the culture cannot be simpler model culture. 16:10:03 The second is that we're actually very interested in artificial selection of microbial communities, hope communities to develop both theory and push experiment of size as well. 16:10:12 And that's why I really look forward to interacting with the physicists here to learn to learn the mathematics behind the still stochastic process. And now we'll be here for the entire duration of all weeks, and so on. 16:10:27 And I would really appreciate it if the physicists in the audience can, you know, can you enlighten me. 16:10:33 And the third direction is this statistical causal inference from time series data so far is that if you have time, we can collect the time series data relatively easily. 16:10:42 And if we can make a causal statement right from the time series data, if the system is stationary, then we might be able to make. 16:10:51 I don't make better hypothesis for experiments, right so that's our experiments can be better design. 16:11:01 And of course you know as Casey talked about whether we could have communities that have hundreds of strains of which mimics natural community so that's another direction by that one critical 16:11:19 comments. 16:11:29 Yes, I guess I'm sort of curious about this this last slide and like the take home message you would say, coming from your quantitative modeling study so I guess I read that paper is saying that you know if you really want to predict what happens in a 16:11:41 specific community you really have to measure all the parameters sort of in situ as as best you can. 16:11:47 And as we scale out to bigger and bigger yeah it's not feasible. Yeah, it's a confident oriel problem. That's right. We've done some entire growth landscape as a function of all the instantaneous values of all the concentrations which seems hard so yeah 16:11:59 what, what would you suggest we do instead are we hoping that there are like smaller, I don't know, phenotypes that organisms can live on and it's a question of like localizing strains to one of those discrete phenotypes based on other measurements or 16:12:15 yeah just sort of curious, I, you know, so actually for the first point I was hoping, because we know the ancestral state so well now. I was hoping that when you know for example when community evolves right so if you look at this community property, 16:12:30 because the growth rate, it changed. How do you know even even though it's surprising or not. How would you know, but if you actually know what are the quantitative but like, if you can phenotype, right of the major clones, you can actually see whether 16:12:56 match matches with what do you have to predict the company. If the community is of the, of these major genotypes, and if not, you might be missing important the general types, I was hoping to use this discrepancy, but should maybe I'm just, they could 16:13:06 be it's too ambitious, but I was hoping to use this quantitative match between model prediction and expense to tell me that I'm missing something. But and how feasible, that is, I don't know where we are trying that we're trying that right and and also evolution new interactions, right because what Terry was saying, How do I 16:13:16 because as will tell you will say, How do I know that I've gotten it right, but it's because for my model. Just, just focusing only on add any nice exchange can already tell me what an experiment where I'm observing the growth rate, I will argue, even 16:13:29 for the other neutral exchange, it cannot be that important. So that is my answer to your question right and I feel quantitative thinking can be very important, but I do not know how to scale up, right. 16:13:48 So if people in the audience and I think it's actually more than just as like a maybe chemistry is that needs to be involved enter this high throughput assets right of high quality assets Bz the beyond my expertise already right so Terry. 16:13:53 What this particular problem I think technology know you're facing was the, well, measuring in faithful conditions. 16:14:02 For example, Martin Ackerman was doing this has this device where you're growing cells together in a vessel in the community but then you take some sample some other medium and pump it's rude, like to win a micro fluid so alien you can just look at the 16:14:18 cell growth in that very situation. 16:14:19 Yes, yes, put it the question is that, how can you keep right i mean so that when the cell, could you have the nutrients, I mean, the short period of time right because as long as you can maintain a defined state in your community, that you can sample 16:14:45 that medium, and you do your measurements to sample the medium one shot, but then medium more comfortable if you're continuously medium and do your study right. 16:14:47 So what they're doing is in microfluidic device. So you just need to pipe a little bit of medium pumping into microfluidic and look at the how this really yes problem is that is if you want to do metabolic measurement and yes yes yes yes so that i think 16:15:02 that that in principle, the technology can can can can can solve a matter of time but then. But this is only for Titanic cultural know everything well makes an appoint together. 16:15:14 Yes. 16:15:17 Yes. 16:15:18 So, I think I might have missed this but at the very very beginning when you're setting up the system does the growth rate of the ancestor and the CO culture Are those the same or different, like either a division of labor benefit in the synthetic mutualism 16:15:34 that you okay so we're ok so the mono cultures, they won't grow eco without partner. It is the one 16:15:49 right so because you're basically taking a yeast that has some growth rate that can do although I see a question, no the cocoa to grow much slower. So then then then Coco Coco much slower than the model quarter in excess metabolites. 16:15:58 Yes, Yes. Interesting. So I wonder, yeah I wonder if there are situations that if you've thought of if there are situations where splitting up the metabolism would lead to an increase growth rate of the CO culture relative to the ancestor. 16:16:08 Okay, so I don't know example by a horizontal from Rutgers right so what he showed was that you can't like support for industry for industry when you want to do some product, or you can have a long, long chain reactions, get to that. 16:16:22 Or you can split it into two half, you know, to have. So he showed you the yield at least for his system. 16:16:28 The, the, the division of labor gives him higher part of the yield the model. The model coach, then the superbug model. So he system he can show that he probably has some other examples, but of course I know, there must be other examples where they were 16:16:43 super Parker's better than the division of labor but in general it's a very good question. On the what circumstances would division of labor give you this advantage over monocultures, right, yes. 16:16:54 Casey, when I wanted to return to the topic Ben brought up about the strength and disadvantage and how that would affect the rate of evolution. And I guess I can't quite put my finger on it but it seems like a system like this is one where you could, 16:17:10 you know, have a really continuous change in fitness in a potential fitness advantages, at least, maybe, in theory, and ask that question I guess I'm one of your hypothesis and whether stronger fit advantages are driving more deterministic evolutionary 16:17:28 trajectories, or whether there ends up being more randomness Could I couldn't quite tell from your answer to his question whether it's like a whole sea of other mutations that are affecting fitness that we don't really understand yet and whether these 16:17:41 are the dominant ones. So those are quite common, but there are other mutations of record, you know, repeatedly arising so the mutations that we see again the independent cultures right so we see ECM 20 hours before all these actually all these mutations 16:18:04 would allow stabilization of the license premise, because it actually makes a huge amount of biological sense why are we on this, on the cell member. And part of the other, the other recurrent mutations we have not. 16:18:07 We have not studied. 16:18:10 It sounds like even with very large disadvantages, they're sort of, sir. So I think these mutations I'm fairly sure the pre existing so those ones because you know look at the outputs from there so if you when you people grow them up by so they're already 16:18:23 pre existing, but once a moment changed the definitive Vantage is so enormous that they can pick up really, they can take over really quickly. So this first wave that I'm sure the second way right The second way would like as some other mutation will 16:18:35 occur for example. 16:18:38 Yeah. So there are different types of chromosome duplications will occur. And then there are the current, so I i surmise that they are probably beneficial because I see them independent independent independent experiments, right and but we have not. 16:18:51 We have not further pursue those, those different types of mutations. 16:18:56 Yes. Angela. 16:19:01 I was just gonna say that in experiments that we've done that are less good than winnings looking at a similar problem we see mutations in the same two genes. 16:19:10 When cells are exchanging two different amino acids, sam Tanya will be if I do they also regulate the same from me, I know. 16:19:20 Yeah, I suppose, right so basically what happens is you make all of the more of all of the permeates is selling some generics and cells become both more selfish and more generous at the same time white. 16:19:33 As you showed that might change the exchange ratio but it actually pushes both numbers. 16:19:37 Yeah, yeah, it was illustrates this sort of falseness of game theory so in game theory you will never get things that are both more selfish and more generous you should just be more selfish. 16:19:48 That's right, but there isn't a genetically hardwired constraint, these are the easy mutations to find the easy thing is to just produce a shitload more of all the transporters, and it is presumably the case that the amino acids escape from the cells 16:20:02 from transporters running backwards, but I SP fine we don't see that right that's what I don't understand like why I see this increase the exchange because our SP five is actually the core. 16:20:22 yeah i think that's just the details of some oh yeah but it works. Yeah. But in in sort of an answer to Casey's question. These are the most frequent mutations, and we sort of have a basis for understanding why they appear, which you explain. 16:20:36 No, it's just thermodynamics, that like the transporters work by taking in things like sodium ions and protons so you can concentrate the amino acid against the concentration gradient.