The constituent based approach for hazard assessment of mixtures and UVCB substances described by ECHA requires the grouping of the constituents in blocks based on structural similarity/mode of action in order to identify representative structures for each block. CORINA Symphony with ChemTunes•ToxGPS® offer a unique cheminformatics solution to apply the constituent based approach in a rigorous way. 

If extractables & leachables can be screened for mutagenicity at the point of elucidation, remedial action can be actioned faster and more efficiently than waiting for the toxicological panel to meet. Alternative routes or materials can be considered earlier. If you are a CRO, you can alert your sponsor faster.

If impurities & degradants can be screened for mutagenicity at the point of elucidation, remedial action can be actioned faster and more efficiently than waiting for the toxicological panel to meet. Alternative routes or materials can be considered earlier. If you are a CRO, you can alert your sponsor faster.

If a lead compound and its metabolites can be screened for mutagenicity prior to synthesis using in silico predictions, then the molecular design can be adjusted at an early stage saving considerable time, effort and cost. Once in vivo tests are conducted and the actual metabolites are identified, in silico mutagenicity screening at the point of elucidation can also accelerate the process of drug development. The concept behind screening is to provide an automated reliable first pass, that can be easily evaluated by an analytical chemist. 

The ICH M7 guideline provides a framework for assessing and controlling DNA reactive impurities in a pharmaceutical product. In the absence of adequate experimental data, the results from two complementary QSAR  methodologies, one expert rule-based and the second statistical-based, may be combined to support an initial hazard classification. The unique feature of this system is the application of a weight of evidence (WoE) algorithm to automatically combine expert rule-based and statistical-based models in a final prediction.