How Our Agent Surfaces Repurposing Candidates in ~3 Weeks
A representative end-to-end run of a PathoscribeAI repurposing pilot, using public data so the methodology is fully transparent. Names and targets are illustrative; the workflow is exactly what we run for partners.
Define the target (Day 1–3)
We start from a single indication — say, an inflammatory disease with poor standard-of-care. The agent assembles the relevant disease biology: associated genes, pathways, and known targets from public sources such as Open Targets, ChEMBL, and the literature.
Screen the existing drug space (Week 1–2)
The agent scores approved and shelved compounds for mechanistic fit against the disease's targets and pathways, using bioactivity data from ChEMBL and drug–disease evidence from repoDB, then ranks them by strength of supporting evidence.
Filter on safety & feasibility (Week 2)
Candidates are passed through our ADMET and toxicity models — benchmarked on the public TDC suite — to flag liabilities early, so the shortlist is biased toward compounds that are realistically advanceable.
Deliver the shortlist (Week 3)
The output is a ranked report of roughly five repurposing candidates, each with a mechanistic rationale, supporting citations, and known safety flags — reviewed by our scientists and walked through with your team.
Want to see this run on a target you care about? Scope a pilot or download the one-pager.