When assessing pharmacological and toxicological profile of chemical structures, several approaches are possible, including comprehensive scientific literature review, data migration from analogue compounds or through preclinical and clinical studies. Nevertheless, the need to generate data by means of other methods than animal tests is widely acknowledged and has led to an overall development of computational chemistry. Computational predictions from valid (Q)SARs methodologies enable to predict different endpoints such as physicochemical properties, pharmacokinetic and toxicological endpoints, or ecotoxicity and environmental features.
Computational toxicology is widely recognized, and its use is actively promoted since it allows limiting experimentation on animals and at the same time saving time and resources. In addition, computational methods allow the easy and immediate application of QSAR models to the study of new structures.
Relation structure activity quantitative methods (QSAR) are computational models that predict the properties of a substance based on its molecular structure. According to the OECD, the development of an unambiguous algorithm, a defined applicability domain and an appropriate fit, robustness and predictivity of the QSAR system are desirable properties that must be met by the models used. The inclusion of a mechanistic interpretation of the QSAR result is also desirable, when possible.
QSAR expert “rule-based” systems are built from the extensive review and assessment of the plausible mechanisms and scientific data available. The scientific review enables the identification of toxicities associated with specific structural groups, termed toxicophores. From this assessment, an alert based expert system is extracted which lead to predict specific properties in the compound under evaluation. Considering that this alert and predictions system is based on the existing scientific data, QSAR results will be as good as the available data-set.
Statistical systems are based on empirical data which allows to assign a probability value based on likelihood of occurrence of each specific parameter. These programs are based on internal structural group comparison, physiochemical properties, the adjacent groups and chemical environment.
In Azierta we commonly use both statistical and expert-based QSAR models to perform consistent toxicological assessments as well as environmental evaluations. Our toxicology team have a broad experience on QSAR analysis of pharmaceutical impurities, cosmetics or biocides, with the support of our consultant partner Ledscope for statistical QSAR assessments. If you want to learn more details, visit https://azierta.eu/toxicological-experts/?lang=en and contact us.