QSAR / QSPR Services

European Union guidance on REACH emphasises the use of methods other than animal testing to generate the data required for REACH registration phase. (Q)SAR models are expected to play a significant role in toxicity prediction for hazard and risk assessment required for REACH Registration phase. Intertek Informatics and Computational Chemistry Group offers complete molecular informatics and QSAR services to support our clients REACH compliance process.

Molecular Informatics and QSAR Services

All non-testing methods are based on the similarity principle – a hypothesis that structurally similar compounds should exhibit similar properties. Non-testing methods using this principle include:

• Grouping Approaches and Read Across
• (Quantitative) Structure Activity Relationships – (Q)SARs

Grouping Approaches and Read Across

If a group of compounds is structurally similar then it should be possible to estimate the properties for an unknown from the compounds whose properties are known. To employ this approach, a procedure for deciding whether compounds are similar is required.

The Intertek Informatics and Computational Chemistry Group have a number of methods available for determining molecular similarity. These include similarity searching using a similarity score, hierarchical clustering, k-means clustering or the generation of self-organising maps (SOMs). All of these methods can use either molecular keys (such as the Symyx ISIS set of 933 keys) or sets of molecular descriptors. For large sets of descriptors, some data reduction may be necessary using a technique such as Principal Component Analysis (PCA).

(Quantitative) Structure Activity Relationships – (Q)SARs

(Q)SAR models can be used to predict in a qualitative or quantitative manner the physico-chemical, biological (e.g. toxicological) and environmental fate properties of compounds from a knowledge of their chemical structure. The parameters used in these models, molecular descriptors, are readily and accurately generated using computational chemistry techniques. These parameters more often than not include physical properties (such as logP and aqueous solubility) but can also make use of any of the properties that can be calculated using the computational chemistry techniques available within the group. In addition to generating the component descriptors, we are able to run models that have been approved and validated by the EU