/tools

This repository serves as a comprehensive resource for machine learning in drug discovery, offering curated datasets, benchmarks, and educational materials. It focuses on enhancing the understanding and application of cheminformatics in predicting molecular properties and interactions.

MoleculeSTM is a multi-modal model designed for text-based editing and retrieval of molecular structures. It provides tools for molecular property prediction and includes datasets for training and evaluation, making it a valuable resource in drug discovery and molecular design.

The Open Free Energy toolkit (`openfe`) is a Python package designed for executing alchemical free energy calculations, primarily aimed at enhancing molecular simulations in drug discovery. It provides robust tools for planning and executing these calculations, contributing to the field of computational chemistry.

The MISATO repository offers a machine learning dataset of protein-ligand complexes designed for structure-based drug discovery. It includes molecular dynamics simulations and quantum mechanics data, facilitating the training of AI models for predicting binding affinities and other molecular properties.

MoleculeKit is a molecule manipulation library that facilitates the handling and visualization of molecular structures. It is particularly useful for tasks in drug discovery and molecular dynamics simulations.

LiGAN is a deep generative model designed for structure-based drug discovery, specifically generating 3D molecular structures that are predicted to bind to target proteins. It utilizes atomic density grids and is built on frameworks like PyTorch and MolGrid, making it a powerful tool for molecular design and optimization.

MolScore is an automated scoring function that facilitates and standardizes the evaluation of generative models for de novo molecular design. It allows users to implement multi-parameter objectives for drug design, benchmark generative models, and evaluate generated molecules using various metrics.

MolTrans is a tool designed for predicting drug target interactions using a transformer-based model. It addresses challenges in molecular representation learning and provides datasets for training and evaluation.

This repository provides a hands-on tutorial on using graph neural networks (GNNs) for various applications in life sciences, including predicting properties of small and macro-molecules, and drug discovery. It includes practical sessions on training GNN models for molecular property prediction and binding affinity prediction for protein-ligand pairs.

RDKit.js is a cheminformatics and molecule rendering toolbelt for JavaScript, enabling users to visualize and manipulate molecular structures. It leverages the RDKit library's functionalities, making it suitable for applications in drug discovery and molecular analysis.

DrugEx is a deep learning toolkit designed for scaffold-constrained drug design using graph transformer-based reinforcement learning. It allows users to generate novel drug molecules by optimizing multiple objectives, making it a valuable tool in the field of drug discovery.

PyAutoFEP is an automated workflow for Free Energy Perturbation (FEP) calculations using GROMACS, aimed at estimating the Relative Free Energies of Binding (RFEB) of small molecules to macromolecular targets. It integrates enhanced sampling methods and provides automation for various steps in the FEP process, making it accessible for both experts and non-experts.

Fragmenstein is a tool that merges and links compounds by stitching them together based on atomic overlap, allowing for the generation of new molecular conformers. It also places follow-up molecules in relation to parent compounds, facilitating molecular design and optimization in drug discovery.

YANK is an open and extensible Python framework that facilitates GPU-accelerated alchemical free energy calculations. It is particularly useful for researchers in drug discovery and molecular dynamics simulations, allowing for accurate predictions of molecular interactions and properties.

Deep-Drug-Coder is a generative neural network designed for de novo drug design, utilizing a conditional recurrent neural network to generate SMILES strings based on specified molecular properties. It aims to facilitate the generation of molecules that meet desired bioactivity criteria, making it a valuable tool in the field of drug discovery.

DrugCLIP is a tool designed for contrastive protein-molecule representation learning aimed at enhancing virtual screening processes in drug discovery. It includes datasets and methodologies for training models that predict interactions between proteins and small molecules.

FlowMol is a flow matching model that facilitates the generation of 3D small molecules from scratch. It employs advanced machine learning techniques to create novel molecular structures, making it a valuable tool for drug discovery and molecular design.

Dockstring is a Python package designed for easy molecular docking, allowing users to dock molecules from SMILES strings. It includes a highly-curated dataset and realistic benchmark tasks to aid in drug discovery efforts.

TankBind is an open-source tool designed for predicting the binding structures and affinities of drugs to proteins using a trigonometry-aware neural network. It supports high-throughput virtual screening and provides scripts for dataset construction and evaluation.

ChemML is a machine learning and informatics program suite that facilitates the analysis and modeling of chemical and materials data. It provides tools for predicting molecular properties and supports various applications in drug discovery and materials informatics.

DrugOOD is a dataset curator and benchmark tool designed for AI-aided drug discovery, focusing on generating datasets for ligand and structure-based affinity prediction. It supports various noise levels and domain annotations, making it a valuable resource for researchers in molecular property prediction.

DrugEx is an open-source software library designed for the de novo design of small molecules using deep learning generative models within a multi-objective reinforcement learning framework. It provides various generator architectures and scoring tools to facilitate the generation and optimization of drug-like compounds.

DeeplyTough is a tool designed for learning structural comparisons of protein binding sites using deep learning techniques. It provides built-in support for several benchmark datasets and allows users to evaluate custom datasets, making it a valuable resource for drug discovery and protein design.

MaSIF-neosurf is a computational tool designed for surface-based protein design, specifically targeting protein-ligand interactions. It employs deep learning techniques to analyze and generate molecular surfaces, facilitating the design of proteins that can interact with small molecules.