Primary Target Prediction of Bioactive Molecules from Chemical Structure

Abstract

AbstractThere are various tools for computational target prediction of bioactive molecules from a chemical structure in a machine-readable material but these tools can’t distinguish a primary target from other targets. Also, due to the complex nature of bioactive molecules, there has not been a method to predict a target and or a primary target from a chemical structure in a non-digital material (for example printed or hand-written documents) yet. In this study, an attempt to simplify primary target prediction from a chemical structure was resulted in developing an innovative method based on the minimum structure which can be used in both formats of non-digital and machine-readable materials. A minimum structure does not represent a real molecule or a real association of functional groups, but is a part of a molecular structure which is necessary to ensure the primary target prediction of bioactive molecules. Structurally related bioactive molecules with the minimum structure were considered as neighbor molecules of the query molecule. The known primary target of the neighbor molecule is used as a reference for predicting the primary target of the neighbor molecule with an unknown primary target. In results, we confirmed the usefulness of our proposed method for primary target prediction in 548 drugs and pesticides involved in four primary targets by eight minimum structures.