Screen ADMET

Best fit: screen ADMET — a screening goal.

The prompt

Fetch gefitinib, erlotinib, and osimertinib from PubChem, run ADMET-AI on all three, then dock each against CYP3A4 (PDB: 4NY4) with DiffDock to predict CYP binding orientation and inhibition mechanism - load ADMET-AI CYP profiles and DiffDock poses into the sandbox and use matplotlib to generate a comparative CYP inhibition dashboard with 2D depictions and binding-mode thumbnails via py3Dmol, select the best candidate for co-administration with a CYP3A4 substrate, and outline the clinical DDI study design?

What each tool contributes

• admet_ai — 50+ ADMET endpoints (comprehensive panel)
• deeppurpose — 20 deep-learning ADMET endpoints (CYP panel, hERG, BBB, Caco-2, AMES, LD50, DILI) + DTI/DDI predictions

Why these tools pair

Comprehensive ADMET coverage: ADMET-AI provides 50+ endpoints while DeepPurpose adds 20 deep-learning ADMET endpoints (CYP1A2/2C9/2C19/2D6/3A4, hERG, AMES, BBB, Caco-2, lipophilicity, solubility, clearance, half-life, LD50, DILI) plus DTI and DDI predictions

Data flow

Every tool operates on the same input: smiles_array.

Read next

• Which tool? — match a single goal to a tool.
• Workflows — every workflow and recipe.
• How the orchestrator picks tools — the same chain from a plain-language goal.