Importance of updated benchmark sets for statistically correct AlphaFold applications

Abstract

AbstractAlphaFold2 changed structural biology by providing high-quality structure prediction for all possible proteins. Since that a plethora of applications were built on AlphaFold2, expediting discoveries on virtually all fields related to protein science. In many cases it seems like optimism made scientists forget about data leakage, a serious issue that needs to be addressed when evaluating machine learning methods. Here we provide a rigorous benchmark set that can be used in a broad range of applications built around AlphaFold2/3.