A chemoinformatics approach to predicting rate of gene silencing on transfection reagent of RNAi therapeutics

Abstract

RNAi is a specific gene silencing mechanism triggered by siRNA. The application of RNAi for gene diseases requires the development of safe and effective delivery systems. For this reason, development of transfection reagents has been expanding. A transfection reagent is a lipid which introduces siRNA into a cell efficiently. We need to predict adequate structures for development of efficient reagents, because the efficiency of introducing siRNA into a cell and the rate of gene silencing differ depending on the structures of reagents. However, the structure-activity relationship has not been clear yet. We therefore proposed a chemoinformatics approach predicting the rate of gene silencing on transfection reagents from the structure by using a regression model constructed with database of transfection reagents. In this study, descriptors as explanatory variables in construction of the model were calculated with the softwares, DRAGON and CODESSA, and an objective variable was their Luciferase expression in the HeLa cell. By using a genetic algorithm based-partial least squares method, which is one of the variable selection methods, a high predictive model could be constructed with a small number of explanatory variables.