Chemical Proteomics-based Target Prioritization through a Residue Agnostic Ligandability Assessment Platform

The landscape of drug discovery is undergoing a transformative phase with the influx of structural biology and omics data. Identifying optimal drug targets amid this data surge presents a multifaceted challenge. Covalent inhibitors, once undervalued, now hold substantial promise, especially targeted covalent inhibitors (TCIs), effectively engaging 'undruggable' proteins and overcoming resistance mechanisms. Existing ML software can proficiently model covalent ligands but lack comprehensive utility across large chemoproteomics sites. Challenges persist in predicting and assessing cryptic ligandable sites and sites beyond cysteine, demanding advanced computational tools. As cysteine-ligandable proteins represent only ~20% of the quantifiable proteome, there is a requirement for ligandability mapping of other nucleophilic amino acids. This study introduces a pioneering computational pipeline leveraging an AI-based ligandable predictor for meticulous evaluation of chemical proteomics-based reactive sites. The pipeline offers a scalable framework to assess covalent ligandability on a large scale, filter out improbable hits and systematically evaluate potential drug targets. Our work addresses covalent drug design challenges through a pipeline that fills crucial gaps in predicting cryptic ligandable and covalent sites in addition to cysteines to foster more efficient drug discovery methodologies.