 So welcome again to this afternoon parallel session systems biology My name is Robert Ljuanek, and I'm chairing the session together with Anastasia and I'm a bioinformatician at the Department of Biomedicine and Basel at the University of Basel and I Would first start with a few words about the session today. We have on plan five talks We have our schedule for 15 minutes and two for five minutes for the longer talks so we will try to ask few questions directly here and I would ask you therefore if you can pose your questions to the Q&A Session which can reach with the button At the bottom and please also upvote the questions and now I passed the microphone to Anastasia My colleague will introduce the first speaker Good afternoon. My name is Anastasia Burs and I'm co-chairing this session with Robert Ljuanek I'm working as a postdoc in the computational systems biology lab of Professor Mihail Azaboulan at BioCenter So the next speaker is Anastasia Burs She is a postdoc and with Mihail Azaboulan at BioCenter and she will talk today about sarcopenia Just a disease where we observe the generative of skeletal muscle Typically an older people and we are looking forward to hear your talk You can hear me now, right? Good afternoon. My name is Anastasia Burs and today I will present you the comparative analysis of skeletal muscle aging in rodents and human In 2015 we teamed up with two groups in BioCenter This is a group of Professor Christoph Hanschen and Professor Markus Ruck to study molecular mechanisms underline skeletal muscle aging I would like to thank these people also Professor Eric Van Niembegen and members of the group of Mihail Azaboulan for their help in accomplishing this project Why do we study muscle aging? As you know the average lifespan is increasing and leads to an increased frequency of aging related diseases One of them is sarcopenia, which is defined as the degenerative loss of muscle mass and function at high age Sarcopenia affects about 5% of men and 8% of women at the age of 60-70 years And it affects more than 50% of individuals at the age of 80 or higher Sarcopenia drastically limits the mobility and therefore decreases the quality of life It is the main reason of the aging related frailty, which leads to a high frequency of falls and increased hospitalization costs It is well established that the onset of sarcopenia can be delayed by caloric restriction and exercising However, there is still no medical treatment available Studying sarcopenia in humans is very challenging because of the long human lifespan, high variability in lifestyle and genetic factors And of course because of the absence of non-invasive methods for collecting muscle samples Therefore rodents, in particular mice and rats are often used as models of human sarcopenia Rodents have much shorter life expectancy, which is about 28 months And they also demonstrate morphological changes characteristic of sarcopenia This is the decreased muscle mass over aging and also the decreased grips tanks Thus, we asked multiple questions What are the molecular changes underlying muscle aging in human and rodents? Are these changes shared between species? And if so, then which ones? To answer these questions, we collected samples of the hind limb gastrocnemius muscle of mice at a various age from 8 to 28 months These samples were further sequenced and analyzed From our collaborator, we obtained a comparative dataset for gastrocnemius muscle in rats, which was also an ASIC dataset For studying muscle aging in human, we used publicly available resource called G-TEX, containing sequence in data for numerous tissues including gastrocnemius muscle The age of available samples ranged between 22 and 70 years The number of samples per age differed depending on the age We expected sarcopenia in individuals at the age of 60 or higher To understand the structure of our dataset, I first performed principle component analysis Here I depicted coordinates of the principle component 1, where each coordinate corresponds to one sample and coordinates are grouped by the age of samples Principle component 1 demonstrates two aspects The first aspect is the trend of molecular changes happening during muscle aging, which we observe here The second is that we see the increased inter-individual variability during aging Strikingly, the analysis of datasets of red and human showed similar pattern Thus, we hypothesized that individual trajectories of muscle aging differ between individuals To prove that, we colored replicates that show variation within the age group by the muscle mass available for rodent species And what we observed that indeed samples with higher preserved muscle mass during aging had coordinates similar to younger replicates In comparison, samples with increased muscle loss during aging had higher coordinates than other samples for both species Thus, we concluded that principle component 1 reflects the muscle health and not the chronological age Further, I identified genes that are responsible for muscle aging To explain the procedure, I will introduce a toy dataset which consists of three samples and about 30 genes A toy dataset can be visualized in three-dimensional space where each dot corresponds to a gene and coordinates of that gene Of that dot corresponds to mean-centered expression in sample 1, 2 and 3 respectively Black-pulled vectors correspond to principle component 1 and 2 respectively To find the contribution of each gene to principle component, I calculate the projection of a gene I calculate the projection of the vector coming from the origin and pointing at the gene on the principle component This I perform for all genes, one after the other, genes that have high absolute projection values in principle component 1 contribute to this principle component most I perform, I apply this procedure for all organisms where I operate it in the space of all samples available for that organism and all genes expressed in that samples And I collected projection values which are depicted here in the form of the distribution plot To standardize projection values across species, I calculated projection Z-scores To give you a feeling how the expression of genes with extreme projection Z-scores value look like I depicted here a gene with increasing expression, it's a co-lipping gene from the mouse dataset with high projection Z-score In comparison, I also plotted a gene with a very low negative projection Z-score having decreasing expression and these are both genes contributing a lot to the muscle aging in mouse Next, I wanted to compare the dynamics of muscle gene expression changes occurring during the muscle aging For that, I correlated projection Z-scores calculated individually for each genes pairwise What I observed that correlation was very poor for all comparisons that I performed indicating that muscle aging is not really conserved at the gene level across species Next, I identified molecular pathways underline muscle aging To do that, I performed genes enrichment analysis based on projections on principle component 1 for all organisms Here is an example how I performed that for one of pathways Basically, I got the list of all pathways and associated genes from the CAC database What I did, I ordered all my genes based on the projections on principle component 1 starting from the highest and ending up with the lowest negative And then I checked what is the distribution of genes that are annotated with this particular pathway in that list And I calculated the enrichment score for that So then, if the pathway was highly enriched on the negative or on the positive side, I translated the enrichment into the color coding Meaning that oxidative phosphorylation is enriched with genes that decrease the expression during aging This is another example of foxy signaling pathway which is positively enriched, which is enriched with genes that increase the expression during aging I performed this procedure for all available pathways and for all genes Then I kept only those pathways that were significantly enriched at least in one species and subjected them to the hierarchical clustering What you can see that we have one clear cluster of genes of pathways that are enriched with genes increasing the expression during muscle aging These are pathways regulating inflammation, cell cycle and proteostasis Also, a cluster of pathways related to metabolism Also, several pathways commonly enriched just in two species In this case, let's say ribosome and extracellular matrix pathways are enriched in red and human And also, pathways enriched only in mouse and red In this case, for example, this is mytophagian autophagian By looking at these results, I concluded that changes in the gene expression are shared actually on the pathway level between human and rodents and not on the gene level This is a well-conserved and coordinated response of pathways during muscle aging and not really well-correlated response on the level of individual genes Suggested that there might be conserved transcriptional regulators during aging across species To check this hypothesis, I applied his MARA tool, which was developed in the lab of professor Eric Van Niem Wagen, also the biosenter And I estimated the activity of transcription factors using my RNA-seq datasets for mouse, red and human Then I got commonly regulated pathways during muscle aging in all species, which were some of them already known to be regulated during muscle aging, for example, this one So then what I did next, using the SMARA output, I also estimated for each of them the dynamic of targets of these transcription factors Here I demonstrate to you the expression of top targets of transcription factors called estrogen-related receptor So as you can see, the majority of top targets have a decreasing expression during aging and the enrichment analysis in David demonstrated that these targets all regulate similar processes related to mitochondria, which is a known target of muscle aging Thus we may conclude that muscle aging is a complex process and I showed that by making the analysis we found out that the chronological age of samples, meaning the age from the birth, is not equal to the physical state of the muscle I also showed that molecular changes are conserved on the pathway level and not on the gene level. I demonstrated that there are several regulators which are conserved during sarcopenium And all presented datasets for rodents can be visualized and studied if you follow the link which I presented here. Thank you for your attention Thank you Anastasia for the talk. I again remind all the audience please post your questions to the Q&A Folks, actually I was wondering, I mean the fact that this mitochondria related pathways are shared, I mean do you have some explanation for that, why exactly this is happening or why this is the main observation? So we know that mitochondria is one of main regulators of energy storage and energy homeostasis and of course the main function of muscle of course is providing us moving capacity which is a very energy rich process and of course this might be one of the main processes which is disturbed during aging and that's why it also affects the mobility that we can't move so well as we did when we were young So does that mean that if you can somehow boost the mitochondria? It would be nice to make an experiment in that direction. Any research going in the direction? There are multiple research, so because mitochondria is a very complex right so there are so many pathways that are really involved in that for example that is the group of Christoph Hanschen at the Biosenter that I mentioned right They work for example in PGC1alpha which is also involved a lot in mitochondria homeostasis and they showed for example that the over expression of PGC1alpha improves a lot the performance of muscle in rodents and if you perform the knockout of PGC1alpha then the performance gets worse So yeah it's shown but of course as I said it's a very complex process it's very hard to target it completely young But thanks again for the talk Thank you So we will move on to the next speaker So the next speaker is Marco Pani who is a senior scientist at the Vital ID in the group of Marco Iverson in Lausanne and he will give us an update to the MetaNec X database please Hello Can you hear me? Okay So I'm going to briefly present you the latest update of the MetaNec X data base And the purpose of the database, the primary purpose is Genome Scale Metabolic Network These Genome Scale Metabolic Network are in cynical reconstruction of an organism metabolism and they have a dual nature On one side they are a repository of knowledge about an organism metabolism and on the other hand they are a numerical model that can be simulated One of the main resources for Genome Scale Metabolic Network is a big database and in this database this very valuable model comes with reaction we are just represented with symbol like this one We have over database like for example the Chaya database that also have representation of the reaction but we fully define the chemical structure for the metabolites And our initial problem was try to establish relationships between symbol and chemical structure across databases And if you can establish relationships for all metabolites of the reaction you can deduce that the reaction are similar And it was really the purpose of creating the MetaNec X database Okay so essentially when we have a set of chemicals the idea is to build a set of them to become a metabolite to identify the set with an MNX-REF identifier And also to choose in this set the best, the one that best represented Starting from this and from a collection of input reaction you can progressively merge the metabolite, rewrite the reaction, go for another step of merging metabolites And at the end you end up with reaction being merged In this project we are using three different lines of evidence to do this merging First it's a systematic usage of chemical structure and chemoinformatic tools Over the last year more and more structure has become available but it remains slightly challenging because of incomplete or missing information within the structure of our groups and of mistakes So the second line of evidence is what I call the reaction context When you have collection of reaction from two databases and already partially reconciled metabolites You can infer using for example reaction cluster reference that some metabolite might be similar even if you don't have chemical structural information The last thing that has been introduced with the new release is we are running on genome scale metabolic network An algorithm which is a variation of flux variability analysis to give a status to every reaction Then we are willing to compare the status of the reaction before and after the mapping trying to preserve it as much as possible For example this is really a case where all the status of the reaction have been preserved at least for all the reactions that can be mapped one to one Here for example you see that we have four metabolites that are not belonging to one to one mapping and nevertheless the status of the reaction could be preserved A few numbers to show you the current status of the resource On the left you see the different database that have been incorporated and the relatively large number of chemical that appear in at least one reaction that have been reconciled The data base is available from our website together with a couple of tools and we are also providing a RDF version of it accessible from our sparkling point Thank you Thanks a lot for this update on the Metanet X And because this was a short talk we don't have time for a question here but we will ask the questions later in the speaker session And now I can pass the microphone again to Manastasia who will introduce our last speaker in this session So thanks a lot So I would like to introduce Xavier Deschar who is a PhD student at the University of Freiburg and he works in the group of Christian Matza And today he will present how he uses the game engine, a very unexpected tool to perform simulations of bacteria interaction So we are looking forward for your talk Yes So So I need to share the screen Yes Perfect so hello everyone thanks for the introduction And today yes I will I will speak about the use of game engine to perform stochastic simulation So imagine we have two types of bacteria in our case Pseudomonas Putida and Pseudomonas Veroni fighting for the same food in our case succinate And when a bacteria gets some food it can divide into microcolonial and occupy the space So our goal is to make a simulation of the war between Pseudomonas Putida and Pseudomonas Veroni And to be as close as we can to experimental setup we are too constrained which has the both tip of the bacteria do not move But they must have a physical reality meaning they can push each other they are solid And then the nutrient the succinate emits from a source and then diffuse into space So those kind of simulation can be can be very tricky to to program from scratch That's why we try to use game engine to help us So what is a game engine So a game engine is basically a set of tools that help people to make video games such as Mario or Zelda So it can offer a graphical user interface to see better what we are doing They can support the day or three day rendering and really important for us They really often offer physics and collision engine Which means that everything like jumping pushing falling is just managed by the engine and you don't need to program that you said So there are really a lot of different engine available And we choose to use goto game engine which is free open source and the syntax is really close to Python So here you see a typical game that we can make with the game engine so imagine you are that little knight and you want to kill a dragon or rescue a princess But before you want to to check that chess so maybe you have like a better sort of I don't know so you need to pick up the key to open the chest So basically what you can do is you define some sprites so you define your aero sprite the key sprite and the chess sprite And then you make those sprites interact So basically you say oh if this sprite touch the key then the key disappear and you add one key in the night inventory And then if the night touch the chest it will check if you have a key or not to open or not the chest So that's exactly what we did with with with a problem so we define three sprites so the 16th which is this little blue pixel The Pseudomonas put it up and the Pseudomonas Veroni And then we make them interact saying oh if a pixel of food touch a put it up and it disappear and you add one to a food counter And for put it up we decide that with one nutrient it can divide across the sprite and the same for Veroni but the Veroni is a bit bigger so you need more food So you need two nutrients to divide and the Veroni will divide at a random angle but lengthwise And as a result here you see four frames of a simulation so one of the first frame you see the 16th immense from a source And you see the two types of bacteria so you need to imagine that the bacteria are not moving but the 16th just move randomly across the screen And when a 16th touch a bacteria it add one to a counter and the bacteria start dividing And as long as the simulation just goes on you see that the food will slowly disappear because it be all eaten by the bacteria And two micro colonies from Putida and Veroni just grow So at the end you can compare the result of your simulation with the experimental data And you can fit some parameter and predict some behavior of the microcolon So it's really a really really easy way to make such kind of simulation and maybe the use of game engine will be one more piece to an uncompleted person So thank you for your attention and I'll be more than happy to answer your question in the mid-speaker room Thank you very much for your talk. It's very interesting unusual approach I would say So I would like to thank all our speakers for this session which showed really how brood system biology can be and how different questions we ask and try to answer So it was really very interesting. I would suggest that we now all move to the mid-speaker room And I would really ask everyone just to feel free and to ask any kind of questions that raised and our speakers will be really happy to answer them So thanks a lot