Drug Discovery Technology Advances Highlighted in Webinar Series

A Flash Talk webinar series presented recent advances in drug discovery technologies, covering chemoproteomic platforms, AI-driven imaging analysis, and ADME optimization strategies for compound development.

Chemoproteomic platforms are transforming the ability to map the druggable proteome and accelerate target discovery. Chemoprotemic platforms enable proteome-wide profiling of ligandable residues and provide integrated insights into target identification, target engagement, and selectivity profiling within a single assay in native biological settings.

Breakthroughs in screening technologies have led to an expansion of the induced-proximity toolkit, providing ligands for both proteostatic machinery and notoriously challenging POIs (proteins of interest). The discovery of novel covalent recruiters for multiple E3 ligases expands upon the existing options for induced proximity-based modalities, such as targeted protein degraders. Leveraging cysteines within intrinsically disordered proteins, such as transcription factors, provides avenues to target an otherwise largely undruggable sector of the proteome.

Emerging modalities such as TRACER (transcriptional regulation via active control of epigenetic reprogramming) extend the concept of induced proximity to the level of gene expression control. By inducing proximity between a transcription factor and an epigenetic regulator, e.g., a co-repressor complex, TRACERs halt transcriptional activity before the synthesis of a disease driving protein is initiated.

Deep learning methods now allow for a higher degree of automation in HCS (high-content screening). Inspired by innovations in natural image classification and autonomous driving, AI methods have significantly advanced cellular image interpretation.

Generative AI approaches, including ISL (in silico labeling) and PCD (profile-conditioned diffusion), further extend these capabilities. ISL predicts fluorescent images from cost-effective BF (brightfield) microscopy to support large-scale phenotypic similarity searches, while PCD generates synthetic microscopy images from bioactivity profiles, significantly improving hit identification without physical experiments.

Optimizing small molecules in drug discovery is a complex, multifactorial problem that requires distilling vast in silico, in vitro, and in vivo data while balancing multiple properties to guide compound progression. Mechanistic ADME modeling, physicochemical property-based design, machine learning, and PBPK modeling can be integrated in projects to prioritize compounds more effectively. Together, mechanistic ADME and MPO support earlier decisions from limited data, improve efficiency by reducing unnecessary synthesis and animal PK, and prioritize compounds objectively by linking properties directly to PK/PD outcomes.