Integrating 730,947 exome sequences with clinical literature improves gene discovery

Area of Science:

• Genomics
• Human Genetics
• Bioinformatics

Background:

• Accurate allele frequency estimation is crucial for genetic discovery, rare disease diagnosis, and population genetics.
• Previous releases of the Genome Aggregation Database (gnomAD) have been instrumental in these areas.

Purpose of the Study:

• To present Genome Aggregation Database version 4 (gnomAD v4), featuring a fivefold increase in exome sequences.
• To enhance the detection of selective constraint and improve disease gene prediction.
• To establish a unified framework for accelerating gene discovery and rare disease diagnosis.

Main Methods:

• Incorporated exome sequences from 730,947 individuals into gnomAD v4.
• Developed a novel loss-of-function (LoF) annotation pipeline with 90% precision for variant classification.
• Integrated gene-disease associations, biological features, and refined constraint metrics within a Bayesian framework.

Main Results:

• Demonstrated increased statistical power for detecting strong selective constraint with larger sample sizes.
• The improved LoF pipeline and inclusion of deleterious missense variants enhance disease gene detection, especially for short or gain-of-function genes.
• Achieved state-of-the-art prediction of gene-disease relevance, highlighting under-characterized genes linked to severe phenotypes.

Conclusions:

• gnomAD v4 significantly advances the scale and accuracy of human genetic variation data.
• The developed methods improve the identification and characterization of disease-associated genes.
• This work provides a powerful framework for accelerating genetic discovery and improving the diagnosis of rare genetic disorders.