Genotoxicity & Impurities
In silico (Q)SAR assessment for mutagenic impurities
ICH M7 allows the mutagenic potential of an impurity to be assessed computationally, using (quantitative) structure–activity relationship — (Q)SAR — models. We run two complementary methodologies for each structure and provide the expert review that turns model outputs into a defensible conclusion.
- Guideline basis ICH M7(R2), section 6 and Note 2
- Typical inputs Impurity structures (SMILES or files)
- You receive (Q)SAR report with predictions, expert review and ICH M7 class
Overview
About In Silico QSAR Assessment
For each impurity, ICH M7 expects predictions from one expert rule-based and one statistical-based methodology. When both are negative and there are no concerns on review, the impurity can be treated as non-mutagenic (Class 5). Positive, conflicting or out-of-domain results need expert analysis — and that analysis is what reviewers scrutinise most closely.
Our reports present each prediction, the alerts triggered, applicability-domain information and an expert assessment that considers mechanism, analogues and available experimental data.
Use cases
When this service helps
ICH M7 assessment of new impurities
Classify impurities identified in route development or stability studies.
Dossier and DMF sections
Provide the (Q)SAR evidence expected in Module 3.
Regulatory queries
Respond when an authority asks for model outputs or expert review.
Re-assessment with updated models
Refresh older predictions with current software versions.
Scope
What the service covers
- Structure checking and standardisation
- Expert rule-based (Q)SAR prediction
- Statistical-based (Q)SAR prediction
- Applicability-domain evaluation
- Expert review of positive, equivocal, conflicting and out-of-domain results
- Analogue and literature search where helpful
- ICH M7 classification and recommended next steps
Who does what
Predictions are run with commercially licensed or validated (Q)SAR systems; the systems and versions are reported in every assessment. Where an Ames test is recommended, it is performed by a laboratory you appoint.
What if the two models disagree?
Conflicting predictions are resolved by expert review: examining the alert, its chemical context, mechanism, analogues with Ames data and model domain. The conclusion and its reasoning are documented. If uncertainty remains, the impurity is treated as potentially mutagenic until data resolve it.
Methodology
How we work
A defined sequence, agreed with you at the start, so you always know what happens next and what we need from you.
- Receive structuresWe check and standardise each structure with you.
- Run two methodologiesExpert rule-based and statistical predictions are generated.
- Review the resultsA toxicologist evaluates alerts, domain and supporting evidence.
- Classify and reportEach impurity is assigned an ICH M7 class with its rationale.
Deliverables
What you receive
(Q)SAR report
Predictions, alerts and expert review for every structure.
Classification table
ICH M7 class and proposed acceptable intake for each impurity.
Next-step recommendations
Where testing, purge arguments or control are needed.
Technical basis
Guidelines we work to
- ICH M7(R2), section 6Requires two complementary (Q)SAR methodologies and allows expert review to support or refute predictions.
- ICH M7 Note 2Describes the purpose and scope of expert review.
- OECD (Q)SAR validation principlesThe reference criteria for model validity, including a defined applicability domain.
What we need from you
- Structures as SMILES, MOL/SDF files or clear drawings
- Impurity names and origin (process or degradation)
- Any existing Ames or other genotoxicity data
- Maximum daily dose and treatment duration of the product
Please share confidential information only after an NDA is in place — we are happy to sign one first.
Market and jurisdiction considerations
ICH M7 is implemented by ICH members including the EMA, US FDA, PMDA, Health Canada and CDSCO, and is widely referenced by other authorities. Some agencies may ask for the software version and model details to be stated explicitly — our reports include them.
FAQs
In Silico QSAR Assessment: frequently asked questions
How many structures can you assess at once?
We routinely assess single structures and larger sets. For large sets we agree batches and a delivery schedule.
Is a (Q)SAR prediction enough on its own?
Two negative predictions with a supporting expert review are generally sufficient to conclude no mutagenic concern under ICH M7. Positive results lead to control at a TTC-based or compound-specific limit, or to testing.
Can you review predictions run by someone else?
Yes. We can provide expert review of existing model outputs, provided the full outputs are available.
Talk to us about In Silico QSAR Assessment
Share your product, target market and timeline. We will review your requirement and propose a defined scope.
- OfficeFlat no 8-9-3/23/1, Subhash Chandra Bose Colony, Dathunagar, Kanchanbagh, Hyderabad, Telangana 500058, India
Open in Google Maps - Phone+91 81219 33381
+91 99894 74590 - Emailbd@globaxa.in
- Office hoursMonday to Friday, 10:00–19:00 IST