In a new preprint led by @TheNikhilMilind, we explored a fascinating paradox: For many traits the number of duplications or loss-of-function (LoF) mutations is correlated with phenotype. Curiously, for most traits, the AVERAGE direction of LoFs and Dups is the SAME. Why? - ThreadSky

jkpritch.bsky.social • 104 days ago

In a new preprint led by @TheNikhilMilind, we explored a fascinating paradox:
For many traits the number of duplications or loss-of-function (LoF) mutations is correlated with phenotype. Curiously, for most traits, the AVERAGE direction of LoFs and Dups is the SAME. Why?

Comments

triggerloop.bsky.social•103 days ago

📌

jrossibarra.bsky.social•104 days ago

I don't recall if he talks about direction of effect, but this is still quite similar to Birchler's gene dosage/balance stuff? https://pubmed.ncbi.nlm.nih.gov/22908297/

nikhilmilind.dev•104 days ago

Yes! Birchler's discussion about genes involved in complexes requiring certain stiochiometries is one of the mechanisms we are considering for the non-monotone relationships between genes and traits.

jrossibarra.bsky.social•103 days ago

Although I've wondered how much of his dosage stuff can just be summarized as "incomplete dominance". His results on triploidy and fitness for example is easily explained by models of deleterious alleles that are partially recessive.

vscooper.micropopbio.org•104 days ago

Fascinating, now I wonder if this would hold in bacteria as well. Feasible to test experimentally and also important.

jkpritch.bsky.social•104 days ago

This seems surprising because, at the level of individual genes, we usually expect LoFs and duplications to have opposite direction effects.
In a new preprint we explore why this happens:
@nikhilmilind.dev @courtsmithrun.bsky.social Julie Zhu @jeffspence.bsky.social

https://www.medrxiv.org/content/10.1101/2024.11.11.24317065v1

pjie2.bsky.social•104 days ago

That was my naive instinct, but after a bit more thought I second-guessed myself. If the effect on the trait is indirect, why should the effect of mis-dosage be directional?

Take fitness for example - genes will be selected to be near optimum dosage, so deviation in either direction lowers fitness.

pjie2.bsky.social•104 days ago

Then let's say you have a gene whose dosage is somehow relevant to neuronal development. Perturbation in either direction may screw up neuronal development and cause learning difficulties.

pjie2.bsky.social•104 days ago

So after putting some deeper thought into it, I'd say my new biological prior is that only *small effect selectively neutral mutations* affecting gene expression are more likely to have directional effect on traits (increasing expression pushes trait one way, decreasing pushes it the other way)

pjie2.bsky.social•104 days ago

Large effect mutations such as frank gain or loss of gene copies are more likely to both perturb the trait in the same direction, because the effect on the trait reflects gross dysfunction of some pathway that was operating near a local optimum.

jeffspence.bsky.social•104 days ago

Absolutely! Check out the last couple of sections of the results — we propose models very similar to what you’re describing. We also propose a model reconciling why GWAS variants seem to have symmetric effects on traits but LoFs and dups affect traits in the same direction on average.

jkpritch.bsky.social•104 days ago

Our biological prior is that deleting a gene should have the opposite effect of duplicating the same gene as in the examples below.
But somehow, aggregating this effect across a bunch of genes, results in a SAME-direction effect on average, genome-wide. Why?

jkpritch.bsky.social•104 days ago

To model this, we introduce the gene dosage response curve (GDRC), which we define as the continuous relationship between gene dosage and average trait value. We propose two models that may explain the directional genome-wide effects. Both include a kind of directional buffering.

jkpritch.bsky.social•104 days ago

First we wanted to know whether non-monotone genes, like in Model 2, even exist. Surprisingly, we found around 40% of gene-trait pairs have a non-monotonic relationship. That is, both deletion and duplication of these genes have the same effect on the trait!
Some top hits:

jkpritch.bsky.social•104 days ago

We hypothesize that genes with non-monotone effects likely affect the complex trait through multiple pathways. We explored one such gene in detail. TWAS/MR-type methods, which assume a linear relationship between expression and trait, might not pick up on such genes.

jkpritch.bsky.social•104 days ago

We found that both models contribute to the genome-wide effect. One essential aspect of this is that many traits are perturbed much more in one direction than the other. We call this phenomenon “trait buffering”, as the curves are all buffered against one trait direction.

johndmcenany.bsky.social•103 days ago

Super cool! Do you have a sense of whether non-monotonicity in GDRCs depends on trait polygenicity? E.g. height vs a “trivial” trait like expression of an RNA, which might have a simpler linear relationship to gene dosage. Is there a continuum between these worlds?

jkpritch.bsky.social•102 days ago

Yes, excellent point. We agree with this logic... we think the non-monotonicity probably (can) arise when multiple distinct pathways connect gene to trait. So it stands to reason that more complicated traits may be more likely to have this. Further testing is in progress... cc: @nikhilmilind.dev

nikhilmilind.dev•102 days ago

Yes, this type of pleiotropy at the pathway level might lead to non-monotone GDRCs. However, our monotonicity estimates are surprisingly consistent across the complex traits we tested (Fig 3), so polygenicity might not directly translate to the number/contribution of non-monotone genes for a trait.

tdrivas.bsky.social•88 days ago

This is super cool stuff, Jonathan! Two questions that might be ignorant of details of your methods. 1. How did you consider dups that affect multiple genes? 2. There is evidence that duplications can result in gene disruption and LOF, can this be accounted for?: https://pmc.ncbi.nlm.nih.gov/articles/PMC4320257/

jkpritch.bsky.social•88 days ago

Thank you! Great questions: (1) yes, we worried about this too: many duplications span multiple genes and so the analysis accounts for this using methods similar to LD modeling of SNPs. The specific gene examples presented in figures show genes with minimal "LD"...

jkpritch.bsky.social•88 days ago

(2) We tried to only consider duplications that span entire genes to hopefully minimize this type of effect. We're doing some follow-up analysis to get a better handle on type of thing but don't have the results yet.
cc: @nikhilmilind.dev

tdrivas.bsky.social•87 days ago

Very cool! Yeah I don’t think it will be possible to completely account for the potentially disruptive effects of duplications without something like long read to fully characterize them, and that’s sadly not gonna be easy to come by at scale, yet

jonhuang.bsky.social•104 days ago

📌

germline.bsky.social•103 days ago

📌

Comments

Posting Rules

Reply