iiinova — where your digital self keeps evolving

Nikhil Milind

@nikhilmilind.dev

305 followers 454 following 48 posts

PhD Candidate in the Pritchard Lab at Stanford University. Interested in statistical and population genetics. nikhilmilind.dev

Posts Replies Media

Nikhil Milind @nikhilmilind.dev · 27/05/2026

We observed that the average effect of gene deletions (LoF) and gene duplications was positively correlated across traits. We use the GDRC to conceptualize what types of gene dosage architectures may explain this pattern. [3/n]

A graphical abstract describing the paper. The first panel displays the positive correlation between average LoF and duplication effects. THe second panel introduces the concept of a gene dosage response curve (GDRC). The third panel describes trait buffering, which is the buffering of GDRCs against one trait direction. Finally, the fourth panel proposes a simple evolutionary model to explain why GDRCs may be organized this way.

110

Nikhil Milind @nikhilmilind.dev · 06/01/2026

We dug into the association of GWAS variants at the GCK locus with blood glucose levels. There was little eQTL signal in the relevant pancreatic tissue, but strong associations in the incorrect direction with evidence of colocalization in tibial nerve and thyroid tissue. 13/n

110

Nikhil Milind @nikhilmilind.dev · 06/01/2026

The false sign rate drops substantially if we analyze the protein data using the relevant tissue. This suggests that tightly-linked eQTL in irrelevant tissues may show strong TWAS and colocalization signals, even though they are not mechanistically related to the GWAS variant. 12/n

120

Nikhil Milind @nikhilmilind.dev · 06/01/2026

Similar to the proteomics analysis, we found that around 33% of genes had an inconsistent direction-of-effect. 11/n

100

Nikhil Milind @nikhilmilind.dev · 06/01/2026

TWAS methods are known to have high false-positive rates, and evidence of colocalization is often used as a filtering strategy. However, even with extremely high evidence of colocalization, the false sign rate remained at around 10%, and resulted in a large drop in recall. 9/n

120

Nikhil Milind @nikhilmilind.dev · 06/01/2026

This is a much higher false sign rate than rare-variant burden tests using LoF variants or duplications, which tend to correctly assign the direction-of-effect for the same proteomics data. 8/n

120

Nikhil Milind @nikhilmilind.dev · 06/01/2026

Surprisingly, we found that TWAS confidently assigns discordant direction-of-effects for a quarter of genes! 7/n

151

Nikhil Milind @nikhilmilind.dev · 06/01/2026

We designed a simple evaluation using proteomics data from the UK Biobank, since we expect gene expression to have a positive effect on protein expression of the cognate gene. 6/n

130

Nikhil Milind @nikhilmilind.dev · 12/11/2024

In addition, 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.

130

Nikhil Milind @nikhilmilind.dev · 12/11/2024

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

110

Nikhil Milind @nikhilmilind.dev · 12/11/2024

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! Here are examples of top hits:

130

Nikhil Milind @nikhilmilind.dev · 12/11/2024

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.

141

Nikhil Milind @nikhilmilind.dev · 12/11/2024

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

121

Nikhil Milind @nikhilmilind.dev · 12/11/2024

For many traits there is a correlation between the number of duplications or loss-of-function (LoF) mutations someone carries, and their phenotype. Curiously, for most traits, these effects are aligned in the SAME direction. Why?

64725