Illuminating the chemically tractable human proteome through data-driven exploration of covalent fragment chemical space

Dr William McCarthy (Crick Institute) Astbury 11B, 2-3 pm, Thursday 5th November

Chemoproteomic profiling of covalent fragments has dramatically accelerated the discovery of ligandable binding sites across the human proteome. However, the limited specificity and selectivity of initial hits often hinders their utility as tools to evaluate the functional consequences of ligand binding. To address this challenge, we generated large-scale chemoproteomic datasets to systematically investigate how two key dimensions of covalent fragment chemical space, (i) electrophile chemistry, and (ii) fragment diversity, influence proteome-wide binding behaviour.

First, we integrated chemoproteomic profiling of enantiomerically pure pairs of cysteine-targeting electrophilic fragments (enantiopairs) with machine learning (ML) to guide the design of fragment libraries optimised for proteome-wide selectivity. This ML-guided library evolution yielded a second-generation enantiopair library markedly enriched in selective and stereospecific interactions relative to its first-generation predecessor. In a complementary approach, we generated proteome-wide binding profiles for a library of diverse cysteine-targeting electrophiles, identifying several under-utilised electrophiles suitable for fragment-based discovery. Together, these studies establish a data-driven framework for covalent fragment library design, integrating electrophile chemistry and fragment diversity to deliver higher-quality initial hits.

Beyond informing library design, these experiments revealed distinct electrophile and stereoselectivity preferences at hundreds of cysteine sites across the human proteome. Furthermore, they delivered hundreds of high-quality ligand:cysteine interactions, representing a significant expansion of the ligandable proteome.