DNA-encoded library (DEL) screening generates -omics scale ligand-target interaction datasets within a single experiment, shifting the bottleneck for drug discovery programs from data generation to data utilization. X-Chem’s Chemomics platform integrates scientific expertise in DEL data analysis and medicinal chemistry with computational tools to support navigation within the empirically screened DEL chemical space, as well as the generation and evolution of physics-based and machine learning (ML) models from DEL datasets to identify and optimize tractable chemical matter from both within and outside of this chemical space.
We describe a case study where Chemomics was deployed to discover chemically distinct hit series targeting Brachyury (TBXT), a transcription factor which is a major vulnerability in chordoma, a rare bone cancer with no effective treatment options. X-Chem’s DEL deck comprising >150 billion diverse compounds was screened against the Brachyury DNA-binding domain, yielding a rich dataset of ~73 k compounds with positive enrichment signal. The scientific team worked in tandem to group this dataset into 960 distinct chemical families, gather corresponding negative data, generate a 3D pharmacophore panel using our DEL2PH4 workflow, and train a machine learning model using our HitMiner® workflow.
We then prioritized compound acquisition from three avenues (1) off-DNA resynthesis of exact DNA-encoded structures scored highly by the DEL2PH4 pharmacophore panel, (2) compounds in commercial catalogs scored highly by both the HitMiner® ML model and the DEL2PH4 pharmacophore panel, and (3) rapid parallel synthesis of analogs of enriched library compounds scored highly by the DEL2PH4 pharmacophore panel using multicomponent reactions. 76 compounds have been tested independently by the Chordoma Foundation in an SPR assay using full-length Brachyury protein, among which target-binding signal was detected for 30 compounds, and dose-dependent signal was detected for 15 compounds, with the top hit so far having a Kd = 4.5μM. Dose-dependent binders were identified from every prioritization strategy, demonstrating the power of X-Chem’s Chemomics platform.
