APPLICATION OF MACHINE LEARNING TO STRUCTURAL MOLECULAR-BIOLOGY

A technique of machine learning, inductive logic programming implement ed in the program GOLEM, has been applied to three problems in structu ral molecular biology. These problems are: the prediction of protein s econdary structure; the identification of rules governing the arrangem ent of beta-sheets strands in the tertiary folding of proteins; and th e modelling of a quantitative structure activity relationship (QSAR) o f a series of drugs. For secondary structure prediction and the QSAR, GOLEM yielded predictions comparable with contemporary approaches incl uding neural networks. Rules for beta-strand arrangement are derived a nd it is planned to contrast their accuracy with those obtained by hum an inspection. In all three studies GOLEM discovered rules that provid ed insight into the stereochemistry of the system. We conclude machine leaning used together with human intervention will provide a powerful tool to discover patterns in biological sequences and structures.