Thread by @jakphd

Really impressive talk on noncoding #constraint in #gnomAD genomes from Siwei Chen (@konradjk's lab).

#ASHG21

Chen: Calculated constraint on 1kb windows using Z scores.

How do known non-coding elements (like enhancers) look when viewed through the lens on constraint?

#ASHG21

Chen: When looking at largest constraint Z-scores (top Z-score was 4 on the figures)
- Super enhancers ~3x enriched
- ENCODE cCRE enhancers ~2.25x enriched
- FANTOM enhancers ~1.75x enriched

#ASHG21

Chen: known #GWAS associations enriched for constrained non-coding regions. Predictably, GWAS loci with external replication had the largest enrichment, followed by GWAS loci with internal replication.

#ASHG21

Chen: Are enhancers linked to constrained genes also constrained (@jengreitz's @Nature enhancer paper comes into play here)? Yes.

More interestingly, are there unconstrained genes with constrained enhancers?
#ASHG21

Chen: 40% of unconstrained genes are linked with a constrained enhancers.
Most of these "unconstrained" genes are small and underpowered for LoF intolerance. These "unconstrained" genes enriched for histone proteins as well as secreted proteins in cell signaling.

#ASHG21

[this finding suggests that perhaps additional power might be gained if rare variants in enhancers are included in the burden tests - thought there's a lot of caveats here]

Chen: When examining 76k CNVs, they found those CNVs in severe developmental disorder patients were enriched in constrained noncoding regions

#ASHG21