Head-to-head safety comparison

Best fit: you have two or three compounds in the same class and want to know which has the cleaner safety profile — and whether the prediction explains the clinical difference.

The prompt

Compare gefitinib, erlotinib, and osimertinib head-to-head on safety — CYP
inhibition, hERG cardiotoxicity, oral bioavailability, the kinome-wide
off-target panel for each, and any real-world adverse-event signals when
patients are on these EGFR inhibitors. Pull the CYP-mediated pharmacogenomic
guidelines and tell me which compound has the cleanest safety profile, plus
whether the predicted ADMET explains the differential clinical toxicity we
actually see.

What it does

The orchestrator runs each compound through the same safety panel and lines the results up side by side:

1. ADMET-AI predicts CYP inhibition, hERG liability, and oral bioavailability.
2. DeepPurpose maps the kinome-wide off-target panel.
3. FAERS surfaces real-world adverse-event signals with disproportionality analysis.
4. PharmGKB pulls the CYP-mediated pharmacogenomic guidelines.

Watch for

• Prediction versus observation. A predicted off-target is a hypothesis; the value is when it explains a real-world FAERS signal. Ask for that link explicitly, as the prompt does.
• Class matters. Head-to-head comparison is most informative within one mechanism (here, EGFR inhibitors) — cross-class comparisons need more care.
• Signal ≠ causation. FAERS disproportionality flags associations, not proven mechanisms.

Read next

• Safety assessment — the guided version: toxicity prediction and adverse-event analysis step by step.
• Which tool? — solubility, toxicity, and safety alternatives.
• What a passport is — the Safety passport assembles this for a single drug.
• How the orchestrator picks tools — the same chain from a plain-language safety question.