Genome Interpretation in Action: Lessons from CAGI and ClinGen Calibration Efforts
Abstract :
Genome interpretation is essential for identifying pathogenic variants in both research and clinical contexts. However, confidence in variant impact predictors varies, limiting their clinical use. This talk presents two major initiatives addressing these challenges: the Critical Assessment of Genome Interpretation (CAGI) and ClinGen’s calibration of computational tools for variant classification.
CAGI is a community experiment that evaluates methods for predicting the phenotypic effects of genomic variants. Over six editions, CAGI has highlighted both progress and limitations, showing that while top methods outperform chance, individual variant accuracy remains limited. Uncalibrated consensus among tools can lead to overconfidence.
Building on these insights, recent work has calibrated variant impact predictors (VIPs) to align with ACMG/AMP evidence levels. Several tools now support “moderate” or “strong” clinical evidence and are included in ClinGen recommendations. Analysis of Rare Genomes Project cases suggests these calibrated tools do not inflate pathogenic classifications. To support their use, we developed VIPdb, a resource cataloguing over 400 predictors with clinical thresholds and performance data.
Together, these efforts improve the reliability and clinical relevance of computational genome interpretation.
BIO:
Steven Brenner is a Professor at UC Berkeley whose research focuses on computational genomics, including variant interpretation, RNA regulation, protein function prediction, and genomic privacy. He is particularly interested in using genome sequencing for newborn screening and genetic diagnosis. He serves on the boards of the Human Genome Variation Society and the HUGO Pathogenicity Committee. He was founding chair of Berkeley’s Computational Biology graduate program, a founding editor of PLoS Computational Biology, and helped lead key initiatives in GA4GH and the Open Bioinformatics Foundation. His honors include the ISCB Overton Prize, and fellowships from AAAS, ISCB, Sloan, and Searle.