I'm excited to share our most theoretical work yet, led by 3rd year PhD student Paul Torrillo. We revisit the old observation that dN/dS depends on the distance between compared sequences. We arrive at a radically different conclusion than current prevailing theory. www.biorxiv.org/content/10.1... - ThreadSky

contaminatedsci.bsky.social • 533 days ago

I'm excited to share our most theoretical work yet, led by 3rd year PhD student Paul Torrillo.

We revisit the old observation that dN/dS depends on the distance between compared sequences.

We arrive at a radically different conclusion than current prevailing theory.
https://www.biorxiv.org/content/10.1101/2023.09.14.557751v1

Comments

acritschristoph.bsky.social•533 days ago

Any luck you'd talk about this at MVIF?

1 question: Imagine env changes and residue at pos 100 is selected. Then env changes back, but pos 100 is not reverted, instead pos 102 is (another path back to orig phenotype). Then did this phenotypic reversion result in higher dN/dS not lower?

acritschristoph.bsky.social•533 days ago

put another way, do you have to quantify/distinguish between "perfect" reversions (at the exact AA level, seem to decrease dN/dS) and "imperfect" reversions (mutating other residues in the same gene to revert that gene's phenotype, would increase dN/dS further)?

contaminatedsci.bsky.social•533 days ago

Our model is considering largely stop codons, which would usually need perfect reversion to be restored. Other types of reversions would be less hidden from dN/dS on long times scales.

acritschristoph.bsky.social•533 days ago

Oh, didn't have stops on my mind...

BTW this is a super interesting idea, you could explore some fun things with it I think:
"The number of invertible promoters in a given genome approximates a lower bound on the number of fluctuating selective pressures that these genomes frequently experience"

contaminatedsci.bsky.social•533 days ago

Stops happen at a shocking frequency in within-person evolution, even in genes that have can be shut off via invertible promoters. That observation was a large part of the inspiration for this work.

erikvannimwegen.bsky.social•533 days ago

What's your interpretation of alpha in this model? In the 'standard' model the interpretation is 'fraction of nonsynomous mutations that are neutral'. But if dN/dS is driven by sweeping reversions, that no longer applies. I have some ideas but would like to hear how you think of this.

contaminatedsci.bsky.social•533 days ago

In this model, alpha still represents the fraction of N muts that are neutral. To be clear, we still include a lot of deleterious N muts, just lower s -- such that they are purged more quickly by selection and reversion

Of course, alpha could reflect a mixture of neutral and directional selection.

contaminatedsci.bsky.social•533 days ago

Thanks for your comment @erikvannimwegen.bsky.social , curious if this fits with your thinking.

erikvannimwegen.bsky.social•533 days ago

hmm.. yeah well.. I guess your model is not radical enough for me, i.e. still to close to the decaying nonneut mutations. I totally agree that the Rocha model of a well-mixed pop with purifying selection weeding out non-neutral N muts so slowly that it take until dS ~ 10^(-3) make little sense 1/n

erikvannimwegen.bsky.social•533 days ago

But I believe that by the time you get to dS ~ 10^(-4) these SNPs were mostly introduced by recombination. So I do not believe one can use molecular clock to pull out a 'time' from dS at those distances. But maybe this is not a crucial problem. I have to think about it more. A technical thing 2/n

erikvannimwegen.bsky.social•533 days ago

that confuses me is that in (7) you seem to suggest ALL non-neut N muts come from these loci with fluctuating selection, but they are still all assumed to decay away to WT so that at large times you only have the fraction alpha of neutral N muts left. This doesn't fit my intuition. I would think 3/n

0ana.bsky.social•533 days ago

Happy to see this out!

pcr-guy.bsky.social•533 days ago

love to see it

contaminatedsci.bsky.social•533 days ago

When bacterial genomes were first sequenced, low levels of dN/dS predominated, suggesting purifying selection. But as more closely related organisms started being sequenced in the early 2000s, high dN/dS (~1) values were observed genomewide.

contaminatedsci.bsky.social•533 days ago

Explanation came from Rocha, Maynard Smith (posthumous), et al. They proposed that purifying selection takes many generations to realize-- such that closely related cells carry newly emerged N that will eventually be purged. The implication is clear: don't trust dN/dS from closely related organisms.

contaminatedsci.bsky.social•533 days ago

Here, we outline the problems with the prevailing theory and propose a model which comes to the rather opposite conclusion: don't use dN/dS from distant bacteria to give intuition on adaptive dynamics.

The paper is short-- so I encourage you to take a look for yourself. And give us your feedback!

contaminatedsci.bsky.social•533 days ago

This work was enabled by the availability of data from the human microbiome project and carefully computed values of dN/dS from @nanditagarud.bsky.social @benjaminhgood.bsky.social @shoestrapped.bsky.social 🙏

Comments

Posting Rules

Reply