Профиль

Thanks a lot to @saramitri.bsky.social's support in this fun journey of a project! Also to @snsf.ch for funding and Mitri lab in general for a great work environment! Reach out if you want to chat about the work! 17/17

Why are we seeing these consistent distribution shapes across ecological scales despite different mechanisms? Interaction dimensionality is much smaller than expected. What does this mean for who determines what the community does? Can we operationalize this better to predict communities in lab? 16/

Overall, we've tried to bridge what the field has understood about processes that impact microbial community assembly with what ecological theory has considered the default for decades. We've just opened many new fundamental and exciting questions during this venture! 15/

Two datasets had community assembly experiments independent from interaction measurements. We then simulated Gaussian and lognormal LVs to predict drop-one-out Shannon diversities and it confirmed what we see from models - that lognormal is better at prediction, especially in larger communities! 14/

lognormal better captures exactly the "many weak, few strong" structure we observed consistently. We constructed and simulated different "informed" Lotka-Volterra communities. We found that the lognormal captures richness better while also having more stable simulations. What about in data? 13/

..explanation for the origin of distribution shape for microbes. Does this matter? Lotka-Volterra theory routinely uses Gaussian because we didn't know a mechanistic alternative. Our consumer-resource derived interaction distribution showed that a heavy-tailed distribution like 12/

Using analytics and simulations, we found that competition and crossfeeding are sufficient to mechanistically explain what creates this shape - less niche overlap / more growth rate heterogeneity could increase skew, but so does increasing leakage of metabolites. This provided an alternative.. 11/

Food web theory explains this shape using body size differences and energy constraints in predator-prey relationships. But these don't really apply to microbes. Interactions are mostly driven by metabolite-mediated processes. So we turned back to our model and looked at the two processes in it. 10/

So now, what about the shape itself? This shape is not new -- food web theory has long identified "many weak, few strong" interaction structure but we didn't know so far that this shape extends to microbes too. Is it expected that this extends to microbes? Actually no! 9/

The skewness and kurtosis were tightly correlated. We fit this correlation and it turns out this model fit predicts the datasets with R2=0.78! Note once again that these communities have different interactions, origins, and methods of measurements, but share a common higher order relationships! 8/

uses statistics like skewness (third order) and kurtosis (fourth order), which for us will quantify a) how many weak values there are, and b) what is the proportion of strong outliers. Second big surprise was that the many communities we simulated in the model ALL showed a scaling relationship!! 7/

..with a wide variety of community origins. Turns out all 9 were significantly skewed, with nearly 78% of interactions being weak!! How do we quantify this observation? We turned to non-dimensional distributional statistics (think like coefficient of variation). Probability theory routinely ... 6/

Turns out most distributions are skewed - quite a lot of values near zero, but, simultaneously, there are some pairs of species that interact much stronger than a Gaussian null expectation. We turned to different datasets to ask is this really true. We took 9 published interaction datasets... 5/

...and crossfeeding, both impact community assembly in microbes. We simulated many model communities and computed the interactions at steady state (see @olimeacock.bsky.social 's work for details on the computation). We were surprised - more than 98% of communities showed a consistent shape!! 4/

But we don't understand these distributions well in microbes -- how they are shaped, what drives the shape, does it matter for community properties in line with theory, etc. We set out to answer these question. We started from consumer-resource models which combine competition... 3/

There has been a lot of discussions on how microbes interact and whether the interactions are negative or positive. BUT interaction strength also matters significantly! From ecological theory we know that the distribution of strengths changes MANY things - from stability to composition. 2/

Very happy to share that the work I've been doing with @saramitri.bsky.social is now out on biorxiv! www.biorxiv.org/content/10.6... We broadly ask how are microbial interactions distributed, is there a common shape across datasets, what drives it, and why it matters. Summary thread below 🧵 1/17

Metabolic processes shape microbial interaction distributionsMicrobial communities are largely driven by metabolic interactions, whereby species compete for shared resources and exchange byproducts. Such interactions are commonly classified by their sign alone....www.biorxiv.org

🎉 Félicitations aux professeures 𝗦𝗮𝗿𝗮 𝗠𝗶𝘁𝗿𝗶 @saramitri.bsky.social @dmf-unil.bsky.social et 𝗧𝗮𝗻𝗷𝗮 𝗦𝗰𝗵𝘄𝗮𝗻𝗱𝗲𝗿 au Département d’écologie et évolution @fbm-unil.bsky.social @unil.bsky.social qui reçoivent chacune un ERC Advanced Grant @erc.europa.eu 👉 www.unil.ch/news/fr/1781...

Deux ERC Advanced Grants pour la FBMLes professeures Sara Mitri, au Département de microbiologie fondamentale, et Tanja Schwander, au Département d’écologie et évolution de l’Unil, reçoivent chacune un ERC Advanced Grant de l’European R...www.unil.ch

From a Workshop to a Perspective piece (on modelling AMR within microbiomes 🦠), glad to have been part of this! Big thanks to Lisa Pagani and @rleonsampedro.bsky.social for carrying it through, and to the Collegium Helveticum for hosting the kickoff meeting 👋

It was a great pleasure to be a part of this and I learnt so much during the process as well! Check it out when you get some time and thanks to all the co-authors for great work! 🙌