Thread by @ksamocha

Now up is Lily Wang (working with @TalkowskiLab + me) on identifying regions within genes that are specifically intolerant to missense mutations. #ASHG22

Lily Wang #ASHG22: Scalability of functional assays remains a challenge -- still a space for in silico predictors to help variant interpretation. This is particularly important for missense variants.

Lily Wang #ASHG22: Phenotype-agnostic approach to identify potentially pathogenic variants is looking for genes/regions intolerant to missense mutations (constrained). For missense, regions may be more critical since we know pathogenic missense variants can cluster in genes.

Lily Wang #ASHG22: Regional missense constraint (RMC): regional differences in missense observed / expected along a transcript.

New and improved methods (bigger data! base-pair resolution! better model!) identify ~3.6k transcripts with evidence of RMC.

Lily Wang #ASHG22: Notes that sample size impacts our power to identify regions. This is obviously a problem that will be solved as new data (gnomAD v4) becomes available.

Lily Wang #ASHG22: While de novo missense variants seen in controls have a distribution of missense constraint matching the distribution seen in the genome -- de novo missense mutations in neurodev disorder patients are strongly enriched in missense depleted regions.

Lily Wang #ASHG22: For example 4-5x enrichment of de novo variants in genes and regions with <40% observed / expected missense variants in gnomAD v2.

Lily Wang #ASHG22: Also integrated RMC methods into a variant deleterious score (MPC), which also incorporate Polyphen-2 and an amino acid deleterious score.

Lily Wang #ASHG22: De novo missense variants with MPC > 3 have an odds ratio of ~10 when comparing neurodevelopmental disorder patient rates to control rates.

MPC also outperforms REVEL, CADD.

Lily Wang #ASHG22: Scores and methods will be on the gnomAD browser soon!

We have plenty of future directions and are excited to expand this approach to the ~700k samples in gnomAD v4.