Neural computational prediction of oral drug absorption based on CODES 2D descriptors

Abstract

A neural model based on a numerical molecular representation using CODES program to predict oral absorption of any structure is described. This model predicts both high and low-absorbed compounds with a global accuracy level of 74%. CODES/ANN methodology shows promising utilities not only as a conventional in silico tool in high-throughput screening or improvement of absorption capabilities procedures but also the improvement of in vitro-in vivo correlation could be addressed.