PyTDC (Claude Skill)

Claude skill that drives PyTDC, the Python client for Therapeutics Data Commons — a curated benchmark suite of drug-discovery ML datasets spanning ADMET prediction, drug-target interaction, drug-drug interaction, drug-response prediction, molecular generation, and retrosynthesis.

   
Type Claude Skill
Supplier K-Dense Inc. (community OSS); PyTDC by Harvard mims-harvard/TDC
Availability GA — actively maintained 2025–2026
Pricing Free / OSS skill (MIT collection); PyTDC itself is MIT-licensed; TDC datasets follow per-dataset licenses
Capabilities Read/Write — Claude executes PyTDC via Python/Bash to load datasets, run benchmarks, and call generation oracles

How to install

  • Also packaged in the SciAgent-Skills collection (jaechang-hits (community OSS, CC BY 4.0)): clone jaechang-hits/SciAgent-Skills and run /plugin install sciagent-skills in Claude Code (or copy skills/structural-biology-drug-discovery/pytdc-therapeutics-data-commons into ~/.claude/skills/).
  • Claude Code / Claude.ai — Skills CLI (recommended): 
    npx skills add K-Dense-AI/scientific-agent-skills

    Installs the K-Dense collection; enable the pytdc skill when prompted (also works in Cursor/Codex via the Agent Skills spec; requires Node ≥ 18).

  • Claude Code / Claude Desktop — manual clone: 
    git clone https://github.com/K-Dense-AI/scientific-agent-skills
cp -r scientific-agent-skills/skills/pytdc ~/.claude/skills/
pip install pytdc

Project-scoped alternative: copy into .claude/skills/ instead of ~/.claude/skills/.

What it does

SKILL.md with recipes for:

  • Single-instance prediction (single_pred) — ADMET, toxicity, quantum properties, paratope, epitope
  • Multi-instance prediction (multi_pred) — drug-target interaction (DTI), drug-drug interaction (DDI), GDA, drug response, PPI
  • Generation (generation) — de-novo molecular generation, retrosynthesis, reaction yield, paired generation
  • Bundled helper scripts: load_and_split_data.py, benchmark_evaluation.py, molecular_generation.py
  • Reference docs: datasets.md, oracles.md (17+ molecule-generation oracles incl. QED, SA, DRD2, GSK3B, JNK3), utilities.md
  • Standard data splits (random, scaffold, cold-start), leaderboard metrics, and TDC benchmark suites (ADMET_Group, DTI_DG_Group, Drug_Response_Group)

Primary use cases: Benchmarking ML models for drug discovery, ADMET property prediction, drug-target interaction screening, molecular generation with property oracles, retrosynthesis evaluation.

Notes

Pairs with the deepchem, medchem, datamol, rdkit-skill, and molfeat entries — PyTDC supplies labelled benchmark splits while those skills supply featurizers and models. Some TDC datasets auto-download on first use (a few GB total across the full suite); allow disk space and network access. Skill is documentation plus Python recipes — Claude calls PyTDC locally via Bash/Python.

Sources

Installed this tool?

Share feedback — install path, OS, errors, workarounds. The form opens with this tool pre-selected and a link back to this page.