Molecular Fingerprints for Robust and Efficient ML-Driven Molecular Generation

2022/11/16 by Ruslan N. Tazhigulov, Tazhigulov, Ruslan N., Joshua Schiller +5 · 1 citation
Computer Science · Materials Science · Biochemistry, Genetics and Molecular Biology · #Computational Drug Discovery Methods #Machine Learning in Materials Science #Chemical Synthesis and Analysis

paper · pdf · doi:10.48550/arxiv.2211.09086

Abstract

We propose a novel molecular fingerprint-based variational autoencoder applied for molecular generation on real-world drug molecules. We define more suitable and pharma-relevant baseline metrics and tests, focusing on the generation of diverse, drug-like, novel small molecules and scaffolds. When we apply these molecular generation metrics to our novel model, we observe a substantial improvement in chemical synthetic accessibility (ΔSAS = -0.83) and in computational efficiency up to 5.9x in comparison to an existing state-of-the-art SMILES-based architecture.

Cited by

• NovoMolGen: Rethinking Molecular Language Model Pretraining

Related

• SMILES Transformer: Pre-trained Molecular Fingerprint for Low Data Drug Discovery
• Phenotypic Profile-Informed Generation of Drug-Like Molecules via Dual-Channel Variational Autoencoders
• A Python library for efficient computation of molecular fingerprints
• Molecular generative model based on conditional variational autoencoder for de novo molecular design
• Deep learning to generate in silico chemical property libraries and candidate molecules for small molecule identification in complex samples