Variational Autoencoder (VAE)-based molecular SMILES string generator

This project uses a Variational Autoencoder (VAE) to generate SMILES strings for novel compound generation. The VAE model is trained on a dataset of existing chemical compounds and can generate new, valid SMILES strings, which may represent potentially new and useful chemical entities.

I implemented a Variational Autoencoder (VAE) model for molecular generation. The objective was to generate novel molecular structures based on a given dataset of molecular representations in SMILES format.

SMILES sequence reconstruction using LSTM Autoencoder & Variational Autoencoder

Variational Autoencoder (VAE) for molecular SMILES generation trained on the MOSES dataset — part of the MolVista AI project for AI-powered drug discovery.

Project steps : Use Variational Autoencoders (VAEs) to generate novel molecules from a dataset of SMILES notation, validate their adherence to chemical rules and assess their similarity to existing molecules using the Tanimoto method.

This project implements a Variational Autoencoder (VAE) for generating valid drug-like molecules using the ZINC dataset. It leverages Relational Graph Convolutional Networks (R-GCN) for the encoder and a dense network for the decoder, capable of transforming SMILES strings into molecular graphs and generating new molecules from the latent space.