Also want to specifically call out our rising-star Early Career Researcher (who isn't on bluesky/twitter!) Harry Wright, who nearly single-handedly conquered the All of Us data, in particular the WGS VDS format: more from him in the near future...

@anna123.bsky.social @timfrayling.bsky.social

This was a really fun and exciting project to work on, and hopefully shows we're not near to exhausting population-scale WGS data. A great collaboration led by @mnweedon.bsky.social @drarwood.bsky.social and @carolinefwright.bsky.social , with @chundru.bsky.social and @rnbeaumont.bsky.social

We next identify novel rare non-coding single variants not highlighted by imputation-based GWAS. For example, we identified two rare variants upstream of IGF2BP2 with large effects on WHRadjBMI, but notably different effects on adiposity traits

We also identify another example of a gene-phenotype association only identified by aggregation of rare non-coding variants via WGS: 5-prime UTR variants in FGF18 were associated with up to 6cm difference in height, but we observed no coding association (possibly due to strong coding constraint)

For height, given the common-variant saturation by Yengo et. al 2022, we further estimated heritability for ultra-rare variants (MAC>10), and found this co-localisation pattern holds, but find some remaining evidence of population stratification (or assortative mating etc) that we couldn't resolve

We then assessed the heritability of rare variants (MAF>0.01%) using RHE-mc regression in UKB. We found a notable contribution for height (~11%), which almost entirely co-localised with common variants, but little to no evidence of rare-variant heritability for BMI and WHRadjBMI

Based on our discovery-replication framework, 97% of our rare-variant findings co-localised with previously reported common variation.

@chundru.bsky.social @drarwood.bsky.social @anna123.bsky.social @exeter.ac.uk

I believe our work demonstrates that the non-coding genome is rife for biological discovery: the exome doesn't have it all! A great collaborative work at Exeter with lots of my amazing colleagues @mnweedon.bsky.social @carolinefwright.bsky.social @rnbeaumont.bsky.social @timfrayling.bsky.social

And finally, we demonstrate that our findings were enriched for regulatory regions active in liver (where many proteins are manufactured) and blood (where the proteins were measured), suggesting they represent truly tissue-specific regulatory effects

Rare non-coding variants had similar-magnitude effects to coding variants, but were more balanced across the frequency spectrum