Virtual Screening Models for Predicting Activities in Various Therapeutic Areas

摘要

A very important aim for drug research is to identify the possible side effects of drug candidates.A straightforward way to do it would be to predict those therapeutic areas for which the compounds might be effective,directly from the molecular structures.However there are relatively few software tools that could do this task.The aim of the work was develop virtual screening models for therapeutic areas,in order to help the process of new lead design in high throughput fashion.The virtual screening was done by the means of support vector machines using a cluster of personal computers.Simple atom type descriptors were used to characterize the molecules.