PARM: A Genetic Evolved Algorithm To Predict Bioactivity
Hongming Chen, Jiaju Zhou, Guirong Xie · Journal of Chemical Information and Computer Sciences · 1998
Based on Walters' GERM (Genetic Evolved Receptor Model) algorithm, an improved algorithm PARM (Pseudo Atomic Receptor Model) was put forward. PARM uses a combination of a genetic algorithm and a cross-validation technique to produce an atomic-level pseudoreceptor model, based on a set of known structure−activity relationships. During the genetic process, an artificial interfering method, which is based on a complementary principle of ligand−receptor interaction, was used to accelerate the search speed. The evolved models show a high correlation between intermolecular energy and bioactivity and can predict the bioactivity of an unknown molecule by interpolating in the regression equation of the structure−activity relationship. This algorithm was applied to two systems and produced reasonable results.