[ICLR 2024] Domain-Agnostic Molecular Generation with Chemical Feedback

Molecular Generation by Fast Assembly of SMILES Fragments

Lo-Hi Splitter for Modern Splits of Molecular Datasets

The backend services of the GenUI framework. The backend provides the REST API used for molecular generation, QSAR modelling and chemical space visualization.

GenUI frontend application. It provides a GUI to the GenUI REST API web services.

Official code for the publication "HyFactor: Hydrogen-count labelled graph-based defactorization Autoencoder".

Official code for "Flatten Graphs as Sequences: Transformers are scalable graph generators" (NeurIPS 2025)

Lightweight toolkit for de novo molecular generation: SMILES & SELFIES tokenizers, CharRNN / MolGPT / VAE models, training, sampling, and MOSES-style metrics.

Deep Learning And Applied Artificial Intelligence Project 2019/2020 - Molecular Synthesis & Reconstruction

Stage-based evaluation pipeline for generative molecular design: filters, retrosynthesis checks, docking, pose validation, reports, CLI/TUI.

A curated list of awesome Molecular Modeling And Drug Discovery 🔥

A repository to build and deploy GenUI docker images. GenUI is a collection of web services for interactive molecular generation, QSAR modelling and chemical space visualization.

[NeurIPS 2024] Official impl. of "Hybrid Generative AI for De Novo Design of Co-Crystals with Enhanced Tabletability". An automated screening pipeline leveraging hybridization of generative AI and evolutionary optimization for pharmaceutical co-crystal design.

Code for the paper "FAME: Fragment-based Conditional Molecular Generation for Phenotypic Drug Discovery", published on SDM 2022.

mirror of Open Molecule Generator from https://git.code.sf.net/p/openmg/code

Official implementation of "Navigating the Fragrance space Via Graph Generative Models And Predicting Odors". This repo is for generative models for designing odor molecules

🧬 Generative model for de novo drug design using Graph Attention Networks (GAT) and an Autoregressive VAE. Trained on EGFR bioactivity data to generate novel, synthetically accessible molecules optimized for drug-likeness (QED) and SA Score.

SMILES-based chemical language models (LSTM & Transformer) in PyTorch/🤗 Transformers for de novo drug design — beam search generation, perplexity-based molecule ranking, and bioactivity-conditioned fine-tuning.

A context-aware computational framework for de novo design of topoisomerase inhibitors conditioned on transcription–replication conflict landscapes, integrating molecular generation, structural dynamics, and genomic bias modeling to identify disease-selective candidates.

Extensions of DiGress for molecular graph generation with logical constraints, focusing on Lipinski’s Rule of Five.