Expert: many popular genomics benchmarks are deeply flawed — publish their limits

anshulkundaje · x · 2026-09-03

Anshul Kundaje says he can cite many examples of highly flawed popular benchmarks in variant interpretation and genomics, plus issues with collating disparate datasets for model evaluation. Establishing and clearly exposing the validity and limits of every benchmark matters, and he hopes such information will be explicitly provided.

Related event: Stanford Professor Says Most Public Benchmarks Have Systematic Flaws(3 posts)→

Original post →

More from Research

Research channel →