Publication of MolAI

"MolAI: A Deep Learning Framework for Data-Driven Molecular Descriptor Generation and Advanced Drug Discovery Applications" by Sayyed Jalil Mahdizadeh and Leif A. Eriksson, published in the Journal of Chemical Information and Modelling, presents a highly generalized and computationally efficient deep learning model for descriptor generation, de novo molecular generation and molecular interpolation.

MolAI trained on a vast dataset of 221 million unique compounds to generate highly quality molecular descriptors using an autoencoder Neural Machine Translation approach, achieving over 99.8% accuracy in regenerating input molecules from their latent space representations. This robust framework enhances drug discovery by enabling precise predictions of molecular properties, including protonation states at neutral pH, significantly improves ligand-based virtual screening and ADMET (absorption, distribution, metabolism, excretion, and toxicity) profiling, offering a transformative tool for pharmaceutical research.