Screening

Physical and virtual compound libraries for phenotypic and target-based drug discovery.

TNC Repurposing Library

We have assembled a 2,975-compound repurposing library of FDA-approved drugs and pharmacological tool compounds spanning 10 target classes including kinases, GPCRs, ion channels, and epigenetic regulators.

This library is designed for phenotypic screening in iPSC-derived neurons. Because every compound has known clinical pharmacology and safety data, hits can be rapidly advanced into disease-relevant animal models and, in some cases, directly into investigator-initiated clinical trials.

Target class distribution of the TNC repurposing library
Chemical space plot: molecular weight vs logP, colored by BBB penetrance

The library occupies drug-like chemical space (MW 150-670, logP -7 to 7) with strong CNS-drug characteristics. Compounds were sourced from MedChemExpress and WuXi and standardized with predicted BBB penetrance, logP, and target annotations.

The majority of compounds (64%) are in Phase 2 clinical trials, providing a balance between validated pharmacology and freedom to operate for new indications in rare pediatric neurological disorders.

Distribution of library compounds across clinical development phases, from Phase 1 to approved drugs
Clinical development stage of library compounds, from Phase 1 through approved drugs.

Virtual Screening

Small molecule docked in protein binding site

We complement physical screening with computational approaches to explore much larger chemical spaces. Structure-based virtual screening with Schrödinger Glide allows us to dock and score millions of compounds against validated targets from our phenotypic screens.

We are also adopting next-generation methods: GPU-accelerated docking tools like Uni-Dock enable ultra-large virtual screens at over 1,000 ligands per second, while generative diffusion models such as DiffDock perform blind docking without predefined binding sites by sampling ligand poses through learned score functions. Together these advances make it practical to screen billion-compound libraries and dock against novel targets where binding site geometry is uncertain.