What happens when you have no or limited experimental data and you need to predict the toxicity or a physico-chemical property of a molecule without testing?
Our team of in silico experts is ready to help...
At Envigo we know that meaningful in silico predictions can only be made by people who understand the chemistry, toxicology and the statistical basis behind the models. Envigo is proud of our team of experts who have many years’ experience in conducting (Q)SARs and read-across and the understanding of the regulatory requirements. We are in frequent communication with academia and software providers in order to improve our service.
Our recommended basic approach for (Q)SAR predictions is to select the most appropriate model from a suite of (Q)SAR models. This means, as part of the assessment we test each model in terms of reliability and confidence in the prediction for the endpoint requested. In addition, the assessment also includes a search for analogue structures which may then be used to extend a model in order to improve the prediction.
Our suite of (Q)SAR models consists of:
- Biovia Discovery Studio (TOPKAT) 4.5 extensible
- OECD Toolbox 3.3
- DEREK Nexus 5.0.1
- MultiCASE Ultra
- VEGA NIC 1.1.1
- US EPA T.E.S.T. 4.1
- US EPA EPI Suite 4.11
As the predictions are made our expert team will assess the quality and confidence of the results. Depending on the endpoint we may recommend using different types of predictive models. For instance, expert rule based systems and statistical based models may complement each other in the case of qualitative endpoints to provide a more confident prediction. As the work continues we may recommend different models which are more appropriate to the molecule or endpoint in question. Finally, we always recommend a thorough review of the predictions in an expert statement to put the predictions into context.
A preferred approach may be to predict the hazard classification of a molecule by read-across. The approach allows biological testing to be waived on the basis of test data from similar molecules in a group. Read-across is possible when there are good data on similar molecules and it can be shown that the toxicological properties of the query molecule can be predicted by comparing between the group members.
Envigo has been providing in silico and read-across services for many years,
- for predicting toxicity of impurities or metabolites in pesticide registrations
- as part of our REACH services
- for cosmetics and pharmaceuticals
- or in the early stage of a client’s product development.
As we move towards the 2018 REACH deadline we fully expect to conduct read-across evaluations and (Q)SARs to classify a large number of molecules.