Your molecules, verifiable.

Submit a structure

Name (optional) Format SMILES XYZ Structure Submit for computation

Assess a structure for bacterial mutagenicity under ICH M7(R2). The expert rule-based structural-alert screen runs here. The guideline also requires a complementary statistical (Q)SAR result, and a documented expert review where the two disagree — enter those below. An assessment missing either is recorded as incomplete and states exactly what is absent.

Compound or impurity identifier Structure (SMILES) Treatment duration — dosing days

Count the days actually dosed, not the calendar span. ICH M7 measures an intermittent regimen by its dosing days: once weekly for two years is 104 dosing days and carries the 20 µg/day limit, not the 10 µg/day that two years of elapsed time would suggest. Leave this blank and the assessment still returns the control action, but no number.

Second methodology — statistical (Q)SAR

Required by ICH M7. We do not supply this; enter the result from your validated statistical model.

Model name Model version Call — not supplied — positive negative equivocal out of applicability domain

Expert review

Required where the methodologies disagree or a result is equivocal. The rationale is recorded verbatim and must be substantive — a reviewer years from now needs the reasoning, not just the conclusion.

Reviewer identifier Conclusion — no review — positive negative equivocal out of applicability domain Rationale

Assess View the alert set

Expert review queue

ICH M7 requires a documented expert review wherever the two methodologies disagree or a result is equivocal. Until that review exists the assessment has no defensible class and the pack reports it as incomplete. Everything waiting on a reviewer appears below, with what each methodology actually said, so you can see what you are being asked to judge before opening it.

Your review is recorded under your identity, taken from your sign-in. It cannot be filed on someone else's behalf. The reviewed assessment becomes a new record and the original is kept, marked superseded — a conclusion reached on a date is never edited.

A control strategy is a list, not a structure. Paste every actual and potential impurity below and receive one signed assessment pack covering the set. The pack hashes the ordered list, so a compound removed after issue is detectable — per-record signatures alone cannot show an omission.

Batch identifier Treatment duration — dosing days

Applies to every impurity in the batch, since a control strategy is normally one drug product on one clinical programme. Count the days actually dosed, not the calendar span.

Impurity list

The third field is your statistical model's call and the fourth your expert model's call. Both are optional per line and both are per-compound on purpose: a single call applied across a batch would be a scientific error, not a convenience. Lines beginning # are ignored.

Expert (Q)SAR identity — optional, applies to the whole batch

If you run your own expert system — Derek Nexus, Leadscope Expert Alerts, or similar — name it here and give its call as a fourth field on each line. Your result then replaces the QENEX structural-alert screen for that compound rather than being added alongside it, so the assessment stays a two-methodology one. Leave this blank and the QENEX screen is used.

Expert model name Expert model version

Statistical model identity — applies to the whole batch

The model is the same across a submission even though its calls differ, so it is named once here. If any line carries a call, this must be filled in — a call no named model stands behind cannot be recorded against a regulatory assessment.

Model name Model version

Assess the list