Anveshar
Anveshar is a reproducible research workbench for rare cancers, built to run from inside Claude Science.
Core trick
Pulls Open Targets, DepMap, and cBioPortal data per rare cancer, then fine-tunes an ESM-2 head on ClinVar-labeled variants to score driver pathogenicity.
Description
Anveshar is a reproducible research workbench for rare cancers, built to run from inside Claude Science. Rare cancers are individually rare but collectively common, and each is underserved: too few patients for a dedicated trial or model. Anveshar catalogs 504 rare cancers (743 conditions), each with driver genes, potential causes, and cited cross-condition therapies scored by confidence, and presents them in an interactive Atlas Explorer that links every cancer to its genes, the analyses you can run, and the runnable foundation models. Its pipeline is reproducible and provenance tracked, pulling live data from Open Targets, DepMap, cBioPortal, ClinicalTrials, and PubMed. On top of retrieval it runs and trains sequence models on rare tumors: ESM-2 scores driver variants zero shot (pathogenic VHL R167W reads as damaging; the SF3B1 K700E change of function hotspot correctly reads as tolerated), and a trained ESM-2 head lifts missense pathogenicity from AUC 0.912 to 0.983 on 588 real ClinVar-labeled variants across twelve driver genes. Two findings are reproducible from public data: rectal neuroendocrine tumors are molecularly distinct from colorectal cancer, and in uveal melanoma an integrated multi-omic class predicts survival where a BAP1 mutation call does not. Because Anveshar makes existing knowledge and foundation models transferable, cited, and reproducible for the cancers that lack their own resources, it provides a personalized platform for tailored investigation of rare cancers.
digvijayky/anveshar★ 00 forksJupyter NotebookNOASSERTIONCreated Jul 13, 2026 · last push Jul 14, 2026
README excerpt
# Anveshar A method for rare cancer and rare disease research. Given a rare cancer, and optionally a patient's molecular profile, Anveshar identifies the cancer's molecular dependencies and retrieves therapies that have been studied against the same dependencies in other conditions, with a citation and an evidence tier for each. Where no established therapy fits, it lists testable hypotheses, kept separate from established options. Rare cancers usually lack dedicated clinical trials. Oncology increasingly assigns therapies by molecular marker rather than tissue of origin, for example pembrolizumab for MSI-high tumors, larotrectinib for NTRK fusions, and dabrafenib plus trametinib for BRAF V600E solid tumors. Anveshar makes that cross-condition reasoning reproducible and traceable to sources. Live pages: https://claude.ai/code/artifact/21dd72b9-0696-4402-a72a-9a365b3094d6 (a landing page that links the atlas, the analyses, the multi-modal analysis, a pipeline run, and example reports). ## What it does Given a rare cancer, and optionally patient variants and expression: 1. Characterize the cancer from the literature (PubMed, preprints). 2. Extract molecular targets from the disease biology and, when provided, from the patient's variants and expression. 3. Retrieve therapies studied against the same target or pathway in other conditions, referencing tissue-agnostic approvals (MSI-H, NTRK, BRAF V600E, RET, HER2) where they exist. 4. Match open trials, including basket and t