Anti-breast cancer drug screening based on Neural Networks and QSAR model

Breast cancer is one of the most lethal cancers, estrogen receptor α Subtype (ERα) is an important target. The compounds that able to fight ERα active may be candidates for treatment of breast cancer. The drug discovery process is a very large and complex process that often requires one selected from a large number of compounds. This paper considers the independence, coupling, and relevance of bioactivity descriptors, selects the 15 most potentially valuable bioactivity descriptors from 729 bioactivity descriptors. An optimized back propagation neural network is used for ERα, The pharmacokinetics and safety of 15 selected bioactivity descriptors were verified by gradient lifting algorithm. The results showed that these 15 biological activity descriptors could not only fit well with the nonlinear relationship of ERα activity can also accurately predict its pharmacokinetic characteristics and safety, with an average accuracy of 89.92~94.80%. Therefore, these biological activity descriptors have great medical research value.