Avash | Geometric Deep Learning Methods with Deep Chemistry for Drug Pair Interactions

Geometric Deep Learning Methods with Deep Chemistry for Drug Pair Interactions

Project photo

Project links

Skills

About this project

Obtained the Simplified molecular-input line-entry system(SMILES) component of each drug, performed feature extraction using autoencoders, and built a multi-class classifier that can predict the interaction class from a range of 86 different interaction classes. Obtained accuracy of 80% (bound to increase further with better computational resources)