The MetaQSAR project consists in the development of a computational platform that enables the automatic generation of QSAR models for the prediction of any molecular property of interest, with a minimal intervention of the user. To achieve this, to following aims were raised:
⦁ Develop a computational flux that integrates the construction, validation and application of QSAR models, following that good practices of computation, recommended by the Organization for Economic Cooperation and Development (OECD) and the European Chemicals Agency (ECHA).
⦁ Implement a metaheuristic for optimization and search, that allows to explore the calculated set of parameters to obtain the best models (with suitable predictive power) in the shortest time possible
⦁ Integrate automatic learning methods in the platform, such as artificial neuronal networks, support vector machines, tree classification and variants, k-NN as well as statistical methods for multiple regressions, linear discriminant analysis and partial least squares,