 Welcome all to this SIB Virtual Computational Biology Seminar series. Today we have the pleasure to host Sarah Mitri, who is a tenure track assistant professor at the Department of Fundamental Microbiology of the University of Lausanne and a group leader of the SIB, the Swiss Institute of Bioinformatics. So Sarah studied computer science at the American University in Cairo in Egypt and also cognitive science and natural languages at the University of Edinburgh in the UK. So she earned in 2009 a PhD in computer and communication science at the laboratory of intelligent systems of the EPFL, so just next door. And from 2004 to 2010, she, sorry, from 2010 she pursued a career with the first postdoctoral training at the FAS Central for Systems Biology at the Harvard University in the US and as well as the second postdoctoral training from 2010 to 2014 at the Department of Geology at the University of Oxford in the UK. In 2015 she came back to Switzerland and she became a met assistant with a ambitione research grant from the FNS and in 2017 she obtained a tenure track at the Department of Microbiology, Fundamental Microbiology here at the University of Lausanne and where she's leading the evolutionary biology group. So Sarah's group is interested in the understanding of how ecosystems of microbes function and evolve over time and how they are shaped by the interaction taking place between individual microbial cells. The current research goal is to develop the theoretical and experimental tools to study and control the social interactions within microbial ecosystem. The mathematical modeling and the computer simulations approaches are developed in combination with prediction testing in the laboratory. So today Sarah will share with us the work done in her group on the ecology and evolution of small bacteria communities. So thank you again Sarah for accepting the invitation and the floor is yours. Great, thank you very much Diana for the introduction and thank you for the invitation. It's really nice to speak in this seminar series. I quite like the idea that it gets recorded and is shared worldwide so it's quite an honor to participate. Okay, so if you're following science news these days then you know that the microbiome is a big deal that people really think about how the microbes in us and on us are shaping our lives. We know that they're really important for our health. Of course this picture is a little bit exaggerated. We're not just vehicles for the microbes but it does still show that it's important to try and understand how these microbial communities function and how they affect us and of course it's not just humans and health related issues but microbes are also in the soil. They affect agricultural productivity so understanding these communities would also be helpful and of course there are many other environments that we're also interested in. So we're part of these groups that are starting to study microbial communities and we know that of course within these microbial communities we're not looking at the single species but many different species living in the same environment sometimes even up to hundreds or thousands. However when we look at what people have done in microbiology in the past most of it has been studying single species or even single clonal groups. So in the field there needs to be a transition from studying single species towards trying to understand microbial communities and how they function. So we are trying to take part in this and really taking multiple species into the lab and trying to grow them together and understand how they interact with one another. So this is what we study. We focus on bacteria and not other types of microbes. And in terms of the complexity of systems that we could be looking at so the couple of examples that I gave before whether it's the microbiome or the soil communities are more on this end of the spectrum so really with many different species and quite complex in environments that are constantly changing. And instead in our work we prefer to focus on the less complex scale of things so with just a few species. So up to let's say 10 species is sort of what we're interested in also because we can understand all the different mechanisms and what's going on within a community and then the hope is in the future to build up towards understanding the more complex communities. So we study small bacterial communities and in particular we focus on how the different members of the community so whether it's species, strains or different cells how they interact. So when I talk about interactions what we're thinking of is any way in which one cell can affect the growth and the survival of another. So these broadly come in two flavors. Cooperation where this effect is positive on the fitness of another individual or competition where there's a negative effect due to the presence of another species, strain or individual. And these have been studied in quite a lot of detail so there is really a huge field of research behind these two different types of interactions. In terms of cooperation I've put up some examples of what they may be. So one species might for example have a positive effect on another because it's secreting enzymes into the environment and these enzymes might break down complex molecules. So there might be some large proteins that these enzymes can break down and the products of this breakdown can then be available to all the cells that are around and not just the cell that created or that secreted this enzyme. And similarly they can also break down toxins that detoxify the environment for their neighbors. They're also famous for producing polymers so if you've seen these large biofilms that are sort of these sticky surfaces so those are made up of polymers that the bacteria have secreted and these can protect their neighbors from the immune system for example or from antibiotics, et cetera. In terms of competition we know that microbes tend to compete for nutrients so if they land in the same environment typically the environment will have a limited amount of nutrients which they need to compete for and so they might end up doing quite aggressive or exhibiting aggressive phenotypes like producing antibiotics to kill their neighbors or even stabbing them with so-called type 6 secretion systems that punch a hole in their neighbor and then they can eat up their DNA or their cellular contents. So there's a lot of positive and aggressive interactions in microbial communities and we're interested in understanding what do these communities look like so if we take a look at a given community will we find more cooperation? Will we find more competition? And how do these types of interactions affect the community as a whole and the functioning of the ecosystem? So that's the second element of what we're looking at is interactions and then finally we're interested in their evolution. So if we start out with a community that looks like this over time what will happen? If some strains mutate for example and perhaps the interaction between them and another strain will also change and what do we expect? Can we build up our understanding in a way that we can predict how this evolution will take place? Okay so before I give some examples and go into the research that we're doing I'd like to first just tell you a little bit about my own background. So my research, the red line of my research has always been about the evolution of social interactions and I studied computer science as Diana was saying in the introduction and my PhD was building on that so I decided to look at social interactions in groups of robots so essentially how individual robots could affect each other's behavior and each other's fitness and how these interactions would evolve over time. And so this got me really interested in evolutionary biology but I did feel quite restricted with the tool that I had because as a computer scientist who was interested in biological evolution just taking my robots and trying to apply it to different biological questions was quite restrictive. So I felt like I had a hammer and I was looking for the right nail to hit so the right scientific question. So instead I decided that I should move a bit more towards biology to expand my tool set. And so I found a position in the lab of Kevin Foster who was very kind to take me on as somebody who'd never held a pipette and he was happy to teach me how to do microbiology experiments in the lab. But his condition was that I would also bring along the skills that I had from before in building computational simulations and mathematical models and then using all those three to understand something about microbial social evolution. And these are the three tools that I've been taking forward as well in my current group that we started here at the University of Lausanne. So the point I guess that I'm trying to bring across within this talk is what tools are good ones to address biological questions? How do you choose which tool to apply to which question? So my approach was to expand my tool set and have different kinds of approaches that we could apply. But I cannot say that I can already answer this question. I just want you to keep in mind that I'm thinking about this throughout the rest of the talk. So just think about whether are we using the right tools and can one combine them in a useful way. Okay, so in today's talk I present to you three small projects. So first of all, understanding how microbes find their neighbors. So how do how does the environment determine which microbe ends up next to which other one? Secondly, how the identity of these neighbors can affect the fate of each of the of these cell strains or species? So will they end up surviving better because of their neighbors or being harmed? And then finally, I'm going to talk a little bit about what we're doing today in my group. So just to give a bit of an overview of how we're starting to expand on these ideas to think of small bacterial communities. Okay. So let's start with the first part. So this is a little cartoon that I draw to illustrate the types of problems that we're thinking about when we look at microbial interactions. So these little cells are two strains of the same species, each one being an individual bacterial cell. And hypothetically, we can say that this one here is a cooperator. So it secretes some enzyme into its environment that breaks down some complex molecule and provides food for whatever cells are close enough. And the green one doesn't secrete this enzyme. So then what we can hypothesize from this, at least the way I've drawn this, is that this brown cloud is where the enzymes are going. And so the benefit of the presence of the enzyme will mostly go to the cells that are within this cloud of enzyme and cells that are a little bit further away won't benefit from it. So in this particular configuration, if you're a blue cell, you're making some effort to produce these enzymes, you and your other clone mates are the ones that will benefit from it. And so as a group, this blue group of cells should grow faster than the green group of cells. Okay. So in this case, cooperation is a good idea. But what happens if instead we have a different setup where the two cell types are mixed together? In this case, the enzymes are equally distributed amongst all of the cells. And since the green ones are not producing this enzyme, they will tend to grow faster because they don't have to pay the cost, but they can get the benefit from the presence of this enzyme. So from the broken down nutrients, for example. So the point here is that the prediction of what will happen and which strain will end up dominating in a population will differ depending on what the environment looks like and how these different cells are distributed in space. So spatial patterns should affect whether bacteria cooperate and whether cooperation evolves in the long term or not. But first, let's talk about how these spatial patterns form in the first place. So how do we get mixed groups versus segregated groups? So this was one of the questions that I started out with when I started my postdoc in Kevin's lab. And so we use these computer simulations to study spatial patterns in microbes. And these simulations were developed previously to my joining the lab. And the way they work is that you have a solid surface and on the solid surface are cells, in this case, of two different colors. So they're identical. None of them are secreting anything. They're not doing anything special. They just have two different colors. And then you should imagine that from the top here, there's a concentration of nutrients that's diffusing down. And the cells can take up these nutrients. And as they take up the nutrients, they grow in size. And then they can divide and push the cells that are next to them. And then this just keeps going for many iterations. And you get this large population of cells and these nice spatial patterns that we can then analyze. And we can do lots of things with these types of simulations such as changing nutrient concentrations, the diffusion of molecules within this environment, the growth rates of different types, etc. So we can really play around with the different cell types and have them do different things and try and explore the parameter space of what we might get. So one of the first things that I looked at was if we change the nutrient concentration, then we end up with very different spatial patterns. Okay, so in the case of low nutrients, that was the simulation that I just, the video that I just ran for you, you get these towers of cells where each tower is a clonal group. So you only have one cell type in there. Whereas at high nutrient concentration, you get a large mixture of these different cell types together. So the first thing that I did when I started working in the lab was to try and test this hypothesis. So to do this, I used pseudomonas aeruginosa, which is a model system because it's quite a dangerous pathogen that particularly affects people with cystic fibrosis. So that's why it's very well studied and we just took it because it's a well studied bug. And then tagged these two strains with fluorescent markers. So we had one that was fluorescing in green and the other one in blue. Then we grew each one of those separately in liquid. We mixed them together at equal ratios, so 50-50 of each. And then we put a little drop of this mixture onto an agar plate. So onto a petri dish. And then you can observe as this colony grows over time, you can look at it under the microscope. And basically what you get is a nice colony like this. So the part in the middle where everybody's mixed is where I put the drop. And then as the two strains grow over time, they tend to separate from one another. What we did then to test whether this would give us something similar to what we saw in the computer simulations was to change the amount of nutrient in the agar plate. So we just made a gradient of nutrients. So this is showing what the colonies look like if you do this nutrient gradient. So what you can see obviously is that, so this of course is just the top part of each colony. And obviously the more food you give the colony, the more it grows. But what we were expecting was to see something like in the computer simulations where you would get nicely mixed cells over here and then less mixed cells at the low nutrient concentration. And that's not really what we see. So what we see is that all of the colonies ended up separating into these two colors. So we got sectors in all of our conditions. But then we proceeded to do a bit more of image analysis of our different colonies. So let me just explain this with this diagram over here. So what we wanted to do was to see to what extent these colonies were mixing and how that was changing over time or over distance from the center. So in this picture here you can see in the circle in the middle is where I put the drop. So where it landed initially. And then from there we can go slowly outwards in circles. And at each circle we want to measure something called heterozygosity. So this is a term that comes from physics and it's basically measuring the variance around the circle. So if you go between if you keep switching back between green and blue as you go around along the circle the more you switch the higher your heterozygosity. So the minimum heterozygosity would be blue all the way around. Okay and the maximum is what we see at the middle is 0.5 where they're completely mixed together. So basically what you see in this on this graph here is as you go away from the center the colonies lose their diversity. So they become less mixed as you go away from the center. And you can see that the different nutrient concentrations you get slightly different patterns. So basically what we observed is that the high nutrient concentrations initially there was a flat part. So the diversity or the mixing remained for longer before the diversity was lost and then this curve was less steep afterwards when they did start to lose their diversity. And so we defined a term called the demixing distance for the point at which we decided that the colony had segregated into its different sectors which we basically took to be the maximum slope of these lines. And then we can plot these. So this is when we reached the maximum slope at what distance against the nutrient concentration and we found a nice linear relationship. So these are basically these red lines that you see in these pictures. So each one of these is the distance at which the colony demixed. Okay so the idea of all of this so it seems like a lot of detail to try and understand what's going on. But basically there are a lot of physical models that try to explain how populations change and how they demix over time. So we used these physical models and tried to find out in our case what is the main factor that's affecting this loss of diversity. And after fitting a lot of models we found that the main explanatory variable was colony expansion velocity. So how fast your colony is growing. So it doesn't matter for example diffusion rate or yeah there are different values that might be important but the one that we found correlated the best was this colony expansion velocity. And so just to illustrate this we took all of our colonies in our experiments and we measured how quickly they were expanding and then we plotted this against the rate of diversity loss which is essentially the slope in the graphs that I showed you before and we can show a very nice correlation. So the faster the colony is growing the slower it loses the strains that were there in the center at the beginning. Okay another thing that I told you at the beginning is that we were surprised that these colonies always demixed. So even when we put more nutrients they still made sectors. And you can sort of see this prediction here. So if we were to draw a line through this then you can see that it's converging it's flattening out but it's not reaching zero. So zero would mean that eventually if you add enough nutrients it would stay mixed forever. So we wanted to understand why this was the case so why did they remain mixed? And one of the hypotheses was that they're running out of food. So in a petri dish eventually the ones towards the edge of the colony are going to start to starve. And this is something that's quite difficult to test with an experiment because I guess you could try to pump nutrients into the edge of your petri dish into the agar but that's quite complicated. And so we went back to our computer simulations and used those instead. So we redid the computer simulations to resemble more of a colony so where we put our cells in the middle and then allowed them to expand. And then again we changed their nutrient concentrations and we saw results that were quite similar to the experiments. So again the more nutrients you put the further the demixing occurs from the from the inoculum. But then what's nice here is that we can actually switch off the nutrient gradients. So what you can see here is in gray or black you see the concentration of the nutrients and as you approach the colony you can see sort of this white halo and that's where the nutrients are getting depleted. So there's a gradient of less and less nutrients towards the center of the colony. So we just switched this off. We basically said okay these cells are not going to take up any nutrients they're just going to use them but then magically these nutrients will reappear in the agar below and then you see that they actually remain mixed forever. So it doesn't, the segregation is basically caused by the starvation effect or the fact that they run out of nutrients. So that was our second conclusion from this. So this might seem like a trivial project where we're really looking at a lot of detail into these small differences that are occurring in these colonies but if you think of this as a group of cells that are within a cancerous tumor for example then this might be an important type of effect that one would need to understand. So if you start with lots of different genotypes of tumor cells eventually you will start to lose different cells and this will depend on the gradients of nutrients or gradients of oxygen for example that you have in a different environment. Okay, so just to summarize this first part what we found is that abundant nutrients in an environment can make it less likely that a cell is surrounded by its clone. So you stay in a well-mixed group for longer if you're in a high-nutrient environment. Now let's move on to the second part of the talk. So why does this matter? Why does the identity of your neighbors matter? So this diagram I showed you at the beginning to motivate why we're interested in looking at spatial structure in the first place. So the argument was that if you're in a mixed group that you're less likely to succeed as a cooperator because you're being exploited by other cells that are not producing these same enzymes for example. And so one of the things that they did in Kevin's lab was to use the simulations to test this hypothesis. So basically they rerun the simulations but this time with one cell type that's secreting something useful and then another cell type that doesn't secrete it and then they looked at under what conditions do you get the success of one lineage or the other. So low-nutrient concentrations as we predict since you end up in these clonal groups where you're surrounded by cells that are the same as you, identical to you then you get the success of the cooperator cells that are secreting the enzyme. And on the other hand at the high-nutrient concentration where you ended up in a very well-mixed group what happens is the opposite is that you get more red cells I don't know if it's very visible here but you get more red cells growing than the blue ones that were secreting the enzyme. And recently so the paper just came out a couple of weeks ago was an experimental test to look at this. So we used our colony system to test whether... So the phenotype that we were looking at, this secretion was enzymes to break down antibiotics. So we could show that cells that are resistant to an antibiotic succeed or not depending on whether they're in a well-mixed or a segregated environment. But I'm not going to present this work today just if you're interested in this then I recommend you go and read this paper. Okay, instead I'd like to look at something a little bit different with the computer simulations which is what happens if your genotypes don't start at equal proportions. So here what I showed you is all the results when you start at the beginning with 50-50 of the two cell types. So the blue cooperators and the red non-secretors. But what typically happens in a natural environment is that maybe you have a mutation that occurs where suddenly a cell starts to produce some cooperative enzyme and it's the minority in its population or perhaps you have cells that are landing in an environment and again maybe that cooperator is a minority. So what happens in that case? Can they still succeed? So we tried to do that so we changed the initial proportion of the secretors to the non-secretors. So here you see one to one which is the same as the experiments I showed before. Well the simulations that I showed before and you can see that in all cases you get secretors that win. So just to explain this graph a little bit better so here on the y-axis we have the fitness of the secretor compared to the non-secretor. So any points above that means that we ran a whole simulation and by the end of the simulation there were more secretors than non-secretors. Okay then it would be above the dashed line and anything below means that it's the blue non, sorry the red non-secretor cells that ended up dominating the population. So one effect that we see is that the more cells you have in the population the more competitive it becomes. Actually you see this more here but the point of the graph is that the fewer secretor cells you have the more they start losing. Okay so the problem is with if you're a rare cooperator in an environment then you get out-competed by the non-secretors. So here's just a little diagram to illustrate why this is. So if you're a cooperator so you're secreting this enzyme and you're in an environment on your own then the more you secrete the better and you can outgrow any other strain. But if you're in a mixed group like this and you're the only cell there that's paying a cost and is secreting then at the beginning there's not much benefit to producing this enzyme but you're growing slower than everybody else so you end up being covered by these green cells that are growing a little bit faster than you. So the initial point in a competitive environment is really important in order to be able to break out in your population and that's why we see this effect here at the lower frequency. So when only a quarter of the cells are these ones that are secreting especially in very crowded environments where we have more cells then we find that in most cases they end up losing. So how can we fix this? Why do we think that cooperation might still be able to succeed in a natural environment? So we were thinking if only cells could measure their population size then maybe they could mediate this. So perhaps they could wait until they're large enough group until it's worth making this enzyme and then they could switch it on and in that case maybe they would succeed. And so the principle that I've just put into words is called quorum sensing. So quorum sensing is essentially a way in which cells while they're growing they secrete some signaling molecules and these go into the environment around them and at the same time they can sense these signaling molecules and once there are enough signaling molecules around then they start to do some behavior. And the point is that these signaling molecules are quite cheap to make and so using those you can sort of mediate your behavior and decide when to start producing something that's more costly. So the canonical example for this comes from a really interesting system so this guy here is the Hawaiian bobtail squid and it's really cool because it has an organ inside that picks up bacteria from the environment and then those bacteria they start out as a small clonal group and then they start dividing and once they reach a certain large enough population size they start producing a protein called luciferase which essentially makes the whole squid light up. And this is supposed to be the way the squid can camouflage itself from its predators. And so this is how people discovered quorum sensing was basically in these bacteria that were producing these signaling molecules to know when to light up because if they started to light up already at the beginning of the day before it's dark and there are only a few cells then they would be making all this luciferase for nothing but instead if they waited until the group was large enough then it would make it worth it. So we were thinking why could this not be also useful in the case where you're competing against other strains? So the model that we were thinking of was something like this so if now the red cell here is a quorum sensor then at the beginning it would be secreting nothing still nothing and so while it's not secreting any enzyme then it's not paying any cost and it can grow just as fast as the green cells around it and then once it has become a big enough group then it can start to secrete and break out and overtake the other cells that are around it. So that was the hypothesis and so fortunately we have these computer simulations where this is relatively easy to implement. So this is a bit of a busy slide but I'll try and explain it slowly. So what you see is the same kind of graph as before with your relative fitness of the secreter and here you have different genotypes. So the C is the constitutive producer so that's the same one as before that was losing out when it's at low frequency and N is just a control so it's non-producers so essentially here we would be doing a ratio of one to four of non-producers against non-producers. So anyway it's clonal population and then these Q lineages are the quorum sensors and the difference between them is just that they start producing the enzyme at different amounts of signaling molecules. And again we have our three initial proportions so at proportion one to one so this was the case where anyway the cooperators the constitutive producers were winning so there's no difference and it doesn't really matter if you quorum sensor or not but the interesting case here was when these quorum sensors were rare in the population and we could see that especially at this particular threshold here that this quorum sensing behavior could rescue the population and manage to make use of this enzyme that they were secreting. So just to show you what these simulations look like helps but to visualize this so here this is a simulation where we made a mixture of all three types but the two secretors so the either quorum sensors or the constitutive secretors are in a minority and you can see that the constitutive secretors they never really managed to break out of the group and they get suffocated by the cells around them whereas the quorum sensors that only start to produce later they managed to make large populations and overcome the competition. Okay so the summary then from this part is that if your neighbors are clones then you're more likely to succeed if you make something cooperative so that's what we saw from this toy example here and I showed you with the simulations and then the other point was that if you're in an environment that's highly competitive then quorum sensing can be used as something to overcome competition so this is quite different from what people thought in the field because typically quorum sensing is really put as a canonical cooperative trait but we're saying it's cooperative within the group but it can help you to compete against other groups. Okay so the third part of today's talk I just wanted to briefly introduce how we're starting to look at expanding this research from single species, different strains towards looking at different species together. So at the moment in my lab we're working on quite a different system from the pseudomonas aeruginosa experiments that I showed you before and we've moved to studying four different species of bacteria. So the idea with this research is that we want to see what happens if you start to add in complexity in the lab. So a bit like I was talking at the beginning if we start to bring many species together and try and see how they're interacting with each other. So these four species of bacteria we chose to study because they are able to grow in something called the metalworking fluid. So these liquids are used to lubricate and cool down machines in manufacturing facilities. So essentially it's an industrial oil so similar to petroleum in composition but they're used as emulsion so they're mixed with water which makes them these milky substances and then they're used in the factory and at the end of their lifetime they need to be degraded because you can't just pour them down the drain obviously so they need to be treated. And one way to treat them which was something discovered by some collaborators of ours at Oxford is that you can put in these four species of bacteria and then they seem to be able to degrade the pollutant compounds within this liquid. So that's what you see here so this is a picture from our own work where in the left you see this metalworking fluid with no bacteria and we can quantify the pollution load of this liquid and then if we leave it for a week with these four species of bacteria we see that the pollution load goes down. So for us this was quite a nice system to study because it lies at an intermediate level of complexity so we're starting to get more than two species into the lab but at the same time it's simple enough that we can really work out all the different interactions that are taking place between the species. We can look specifically at the phenotypes of the species so for us it's a first step towards these complex communities that we see in the environment. They're quite manageable in the lab so they all grow overnight and they're relatively easy to work with. They're all aerobic and we can also measure the productivity so in terms of this pollution load so we can try to study what happens if you change certain things in the environment or you add in or take away different species how will that change the efficiency of the system or the functioning of an ecosystem? So what we're trying to do then is to build this into a model system for studying different questions related to bacterial communities and in particular small bacterial communities. So we're not claiming that any of these results would necessarily apply to larger communities but it's a good place to start. So just to illustrate the type of things that we're doing with this system so one of the first questions was to try and measure how the different species interact and this is work done by Philippe Picardi where he started experiments first putting each of the different species alone into this metal working fluid and then starting to combine species. So he did all the pairwise combinations of species groups of three and then finally all groups of four and he grows these in cultures and then measures the population sizes and tries to correlate how the growth patterns change depending on who's there. We're also working to quantify all of this so Björn Vestman is doing a PhD that's more focused on the modeling side of things so he is trying to find the right equations that would capture these types of interactions and put them more into a mathematical model that will allow us to predict what would happen under different conditions and so this is an example of one of the models that he's using which is called the Lotka Volterra model where essentially you can describe the change in population size of one species of bacteria based on its own growth rate and also how it's affected by the presence of others that may be in the same co-culture with it and essentially what he's doing then is to take all the data from the experiments and try to estimate growth rates and interaction strengths that would minimize the difference between the model and the experimental data. So for example, here's some of our data where we see one of the species is Ocrobactrum that's dying on its own so this would have a negative growth rate and then if you grow that same species in the presence of another one Comomonas you see that it starts to grow quite well. So in this case you would end up with a negative growth rate and a positive interaction strength. So it's just to illustrate the kinds of models that we're starting to build and to look at in order to study and predict and also to control our system. So if we have a good model then we can also see if we manipulate the growth of Comomonas for example by changing the environment can that change the interaction between those two bacteria? And then a final thing that I wanted to show you was that we're also looking at spatial structure within our system. So you might think that this is not an interesting point because we're working in liquid culture but if you look at this culture over here you see on the surface of it that the air-liquid interface there's this thin film and this is basically bacterial biofilm so this is where they're secreting these polymers and they make these very thick groups of cells so if we take a little bit of that surface so you can see it here in a little tube, a flask then you can see that it's sort of this sticky substance and we can look at this under the microscope as well and see where the different species are and the reason why we can see the species is because of work done by these three guys which was really a lot of hard work to try and fluorescently label the different species with different colors and setting up microscopy to be able to visualize everything but the idea is from that to see whether we can map the interactions to the spatial structure of the different species and we're also setting up some experiments with the colony model so again if we can grow our different species in monoculture and then mix them together spot them onto an agar plate and if we look at that under the microscope then we start to see some nice pictures so this again here is the center where all of the species fell in the middle of the drop and then we can see how they're separating over time into these sectors like what we were studying before and of course we can start to look at parallels between that and our computer simulations where again we got these nice clonal sectors so that was a quick overview of the type of work that we're doing at the moment in the lab so I'll get back to this question so I've shown you that we've been using a combination of these different tools so whether they're computational simulations mathematical models and trying to combine them with experiments so I don't know if the way we're doing it is correct but at least it seemed intuitive to be able to have these different tools and pick them up depending on the question that we needed to address and how easy it was to use these tools in the different situations so what I think I'm trying to say is that even though I came from a different background it was nice to be able to use these different tools and to learn different tools because it allows you to expand the number of questions that you can address with your research and so I think that this is sort of my message that I think biology students should become more quantitative that at the moment people either learn the computational or mathematical side or they learn more of the biology and I think that doing such computational analyses building models is as important to biologists as it is for physicists for example so I really think that this should be a focus in education and one does see that there is more on the side of genomic analysis so people have picked up that people need skills in bioinformatic analysis but I think that's not the only thing so more general quantitative methods I think would also be very helpful and this would allow people in the future to be able to be a bit more versatile with the tools that they use and at least to be able to more effectively collaborate with people who are more specialized in more of the mathematical or computational side okay so that's all I wanted to show you today but just to finish off I wanted to especially thank people at the DMF at the department where I am here at Emil who've been extremely helpful with us setting up in the lab these are the people at the zoology department who participated in the project that I showed today especially Jonas and Armin are the two first authors on the paper with the quorum sensing and we collaborated with Joao who was one of the first developers of these computer simulations and then I want to especially thank the members of my lab who have been extremely enthusiastic and just a great group to start our research with and finally some funding bodies and thank you for your attention