Mikhail Tikhonov

33 posts

Mikhail Tikhonov

Mikhail Tikhonov

@mikhtikh

Theoretical physicist thinking about ecology and evolution. Of microbes, and spheres in vacuum.

Katılım Şubat 2017
89 Takip Edilen330 Takipçiler
Mikhail Tikhonov
Mikhail Tikhonov@mikhtikh·
Excited to share the fruit of a collaboration with the @SeppeKuehnLab and Madhav Mani, just out in Nature. Driven by the powerhouse duo of @kiseokmicro and Siqi Liu. Predictable functional regimes in minimally processed soil + a simple model recapitulates functional dynamics!
Kiseok (Keith) Lee | kiseokmicro.bsky.social@kiseokmicro

Published in Nature today! Here, we sought to systematically ask how natural community's metabolism changes with the environment. A simple consumer-resource model can predict N cycle metabolism and, more importantly, the mechanism behind its change. nature.com/articles/s4158…

English
0
0
5
191
Mikhail Tikhonov
Mikhail Tikhonov@mikhtikh·
New since preprint: Prompted by reviewers (thanks!), we added an independent dataset and show the same patterns of bias are present - and can be corrected. The effects are usually small... except when they aren't. So the issue seems general, but thankfully so is the solution! 2/2
English
0
0
3
327
Mikhail Tikhonov
Mikhail Tikhonov@mikhtikh·
And if it’s surprising, is it useful for making *predictions*? We present a framework to rigorously define emergent simplicity of prediction, and show it in data. Two experiments, 4 of 5 properties: simple models become more predictive at high diversity. The magic is real! 2/3
English
1
1
5
1.7K
Mikhail Tikhonov
Mikhail Tikhonov@mikhtikh·
Very excited about this work. "Emergent simplicity" in microbial systems is an old hope, but the term means different things to different people. As a result, the field never quite agreed: is it real? If it’s real, is it surprising? @MoranJacobT arxiv.org/abs/2403.19372 1/3
English
2
16
51
7.5K
Mikhail Tikhonov
Mikhail Tikhonov@mikhtikh·
Here, we propose that a generic outcome of life is the formation of competing ecosystems. And eco dynamics may yield biosignatures independent of specific metabolism: specifically, an emergent stratification of chemical resources in order of decreasing energy content. 2/2
English
0
2
6
2.8K
Mikhail Tikhonov
Mikhail Tikhonov@mikhtikh·
Preprint! If you had a soil sample from Mars, what would you look for as evidence of life? Many proposals focus on specific compounds, but extraterrestrial life need not resemble ours. And no known signature comes from self-replication alone.1/2 @eltanin4 arxiv.org/abs/2403.18614
English
1
7
32
8.3K
Mikhail Tikhonov
Mikhail Tikhonov@mikhtikh·
Excited to share the fruit of a collaboration with the fantastic @SeppeKuehnLab and Madhav Mani, driven by the powerhouse duo of @kiseokmicro and Siqi Liu. Predictable functional phases in minimally processed soil + a simple model recapitulates functional dynamics!
English
0
2
5
1.5K
Mikhail Tikhonov
Mikhail Tikhonov@mikhtikh·
We use models with known "ground truth" to show that simple methods can recover the correct structure surprisingly well -- in fact, often better than more expressive models. And present a multi-group Metropolis algorithm that improves on EQO performance. 2/2
English
0
0
0
377
Mikhail Tikhonov
Mikhail Tikhonov@mikhtikh·
Our new preprint: How can methods as simple as the @oxcordero group's EQO (based on a linear regression) work for identifying relevant microbial groups? Reason: finding the right *variables* is different from finding the right model! 1/2 doi.org/10.1101/2024.0…
English
2
7
19
2.5K
Gavin Sherlock
Gavin Sherlock@gsherloc·
@mikhtikh Looks super interesting. Is there a downloadable tool for correcting data?
English
1
0
0
62
Mikhail Tikhonov
Mikhail Tikhonov@mikhtikh·
Just put out a fun new paper with @RS_McGee! In bulk assays where strains/mutants are identified by barcodes, those barcodes are not strictly inert - can differ by e.g. susceptibility to amplification bias. Imagine you wanted to be SUPER careful & correct for that.
English
1
2
9
2.1K
Mikhail Tikhonov
Mikhail Tikhonov@mikhtikh·
One way could be to label every mutant with 10 barcodes… But you don't have to! At least in one popular context (bulk fitness measurement) you can correct for biases with no changes to protocol, even retroactively. Thanks to awesome collaborators @PetrovADmitri @GrantKinsler 2/2
English
1
1
3
1.2K
Mikhail Tikhonov
Mikhail Tikhonov@mikhtikh·
A fantastic collaboration and what a pleasure to be spending my sabbatical with @SeppeKuehnLab at @UChicago. Now we just need to understand *why* this works!
English
0
2
13
2K
Mikhail Tikhonov
Mikhail Tikhonov@mikhtikh·
@mcsymbiont @wc_ratcliff @PetrovADmitri I'm so into this thread! But even in physics the definitions aren't clear-cut. E.g. Newton's laws "combine definitions, observations from nature, partly intuitive concepts and some unexamined assumptions about space and time". What will be the relativity / qmech of biology?.. :)
English
1
0
3
0
Will Ratcliff
Will Ratcliff@wc_ratcliff·
This is true for most widely-used definitions in biology, whether it's species, organism, fitness, etc. And when we *do* ask our colleagues what they mean by it, there's a decent p() that all hell will break loose (remember the 'is fitness an a trait' twitter meltdown?).
Rosie Redfield@RosieRedfield

@vscooper Snark: The assumption being that we all agree on the same rigorous definition of 'population'? Truth: Population geneticists seem to operate under an implicit agreement not to ask their colleagues what they mean by 'population'.

English
2
5
24
0
Mikhail Tikhonov
Mikhail Tikhonov@mikhtikh·
In conclusion: bacteria could quite plausibly learn and adapt to environmental correlations on physiological timescales, and not just evolutionary. 4/4
English
0
0
2
0
Mikhail Tikhonov
Mikhail Tikhonov@mikhtikh·
As long as the three ingredients are there, a variety of implementations can work, including some that resemble two-component signal transduction systems. 3/4
English
1
0
2
0

Keşfet

@SeppeKuehnLab @kiseokmicro @RS_McGee @GrantKinsler @PetrovADmitri @asanchez_lab @MoranJacobT @eltanin4