As the scientific pursuit of knowledge faces intimidating challenges, collaboration becomes even more important. I'm a staunch believer in "night-science", which works best when done with others (see this great article from Itai Yanai and Martin Lercher; doi.org/10.1038/s41587-023-02074-2).

Learning about this problem has been incredibly fun and exciting. I met @akshitg.bsky.social a little over a year ago at APS 2024. Since then, we've worked together on a number of projects. Collaborating with kind and curious people is my favorite part of science and Akshit is a prime example.

When we combine all known mechanisms, we still can't quantitatively account for the level of strain diversity observed in nature. This suggests a fundamental gap in our ecological understanding that requires new experimental measurements, new theoretical frameworks, and new dialogue between them.

We discuss several potential mechanisms that maintain this diversity and highlight their limitations: (1) niche-based (nutrient specialization, physiological tradeoffs, phage interactions, spatiotemporal dynamics), (2) neutral (migration, stochasticity), and (3) evolutionary (mutation, HGT).

We call this the "Paradox of the Sub-plankton" - how do organisms with >99.9% genetic identity avoid competitive exclusion and maintain diversity?

The microbial 'Paradox of the Plankton' asks why many species can coexist on limited nutrients. While we have some understanding of how this work at the species level, another puzzle exists at a finer scale: within species, nearly genetically identical 'strains' show distinct ecological behaviors.

Happy to have this out there! I had a lot of fun working with (and learning from) @akshitg.bsky.social on this mini-review on microdiversity in microbial populations and how current ecological theory fails to quantitatively explain the observations. doi.org/10.1111/1462...

Excited to finally share this work w/ @suryaganguli.bsky.social Tl;dr: we find the first closed-form analytical theory that replicates the outputs of the very simplest diffusion models, with median pixel wise r^2 values of 90%+. arxiv.org/abs/2412.20292

Generate random draws from a Cauchy just to remind yourself that you almost never actually want a Cauchy anywhere in a probabilistic model.

Happy to have this out today in JOSS! Hplc-py is a Python software tool that quantitatively decomposes complex chromatograms into their constituent signals: joss.theoj.org/papers/10.21...

I am cautiously optimistic that my lab will have an open postdoc position in 2024. If you are potentially interested, please reach out. Or if you know somebody who might be interested, please connect us. My email is on our lab website sklab.science

Beyond putting together documentation and tutorials, we put together a brief preprint on the Chemrxiv
outlining the major goals and performance of the software. We would love to know how it works on your chromatographic data! t.co/8YXu7YZ3MS

Unlike extant chromatographic processing software, hplc-py allows the user to constrain specific parameters of the fit distribution. This allows one to deconvolve completely overlapping signals like the following:

This programmatic interface i) automatically detects peaks in a chromatogram, ii) fits a plausible mixture model to reconstruct the observed signal, and iii) computes and returns properties of each analyte. With hplc-py, you can go from raw data to this in ≤10 lines of Python.

Over the past two years, I've used HPLC to measure carbon turnover in microbes. The most painful part was quantifying the resulting chromatograms. Today, I (with Jonas Cremer) released hplc-py: A python tool that makes quantification easy. cremerlab.github.io/hplc-py

If you’re interested in comprehensive guidance for improving your communication skills I cannot recommend the book “Trees, Maps, and Theorems” enough, www.principiae.be/X0100.php. 2/5