Possibly functional de novo genes might have evolved in the LTEE

Still haven’t tested them for whether they confer any adaptive benefits. But there have been multiple instances of de novo genes (which are transcribed, and in one case also translated), that have emerged, gone to fixation, and persisted in an actively transcribed state above the level expected by pervasive transcription.

Will certainly be interesting to see whether any of them are adaptive. That would mean we have directly observed de novo gene formation, not “just” inferred the process from comparative genomics and phylogenetics.

Edit: I should add that it’s also inferred through comparative genomics and phylogenetics here, but with the added advantage that scientists have frozen samples of the ancestors in the experiment which we of course lack for species in the wild, so they’ve been able to piece together the chronological order in which the causative mutations that gave rise to these genes actually occurred and it happened under controlled conditions in an experiment.


Rather surprised they didn’t test for selection since they have time-series allele frequency data. A visual inspection of Fig. 9 seems to show rather rapid increases in frequency that appears inconsistent with genetic drift alone and might indicate positive selection. Using neutral simulations, they could’ve tested whether these observed increases were consistent with drift (e.g., the observed rate of increase falls outside of the 95% confidence interval of the neutral simulations). Detecting positive selection would obviously indicate that these de novo genes are functionally significant.

1 Like

Agreed. One confounding factor I can see is they could simply be caught up in selective sweeps, as most of the strongly beneficial mutations are also known to have fixed early in the experiment.

They have to knock out expression and do competition assays. Do they fare better than competitors that don’t express the locus? Possibly also replace the expressed locus with just some random piece of nonfunctional DNA to determine to what extend expression alone affects fitness as the expression level might not be optimized to any putative function.

Out of curiosity, how hard would that be to do? Is it something you could knock out in an afternoon?

Welcome to Instant Rimshot :slight_smile:

That, I suspect, is in the followup paper/preprint.

It’s not hard, though Rumraket brings up a good point that it wouldn’t be able to distinguish between these de novo genes being targets of selection themselves or if they are hitchhiking on other targets.

That makes sense.

Not hard for you.

Just wanted to make the point that you and the other people in this community have an incredible amount of knowledge and expertise, and that I am thankful for your collective patience and kindness in trying to communicate these sophisticated ideas to lay people, and help me in particular to understand them.

Merry Christmas, and a happy new year, everyone

In particular, thank you to:


For your time and expertise.


This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.