/tools

Caliby is a software tool for protein sequence design based on the Potts model, enabling users to condition designs on structural ensembles. It includes functionalities for sequence scoring and ensemble generation, making it a valuable resource for molecular design in computational biology.

The riff_diff_protflow repository provides an implementation of the RiffDiff pipeline, which is designed for generating and optimizing enzyme structures from theozymes. It utilizes various protein design tools and scripts to create fragment libraries and refine structures, facilitating the design of novel proteins.

This repository provides an implementation of Thompson Sampling for virtual screening of un-enumerated libraries in molecular design. It allows users to efficiently search and score potential molecules based on various scoring functions, facilitating the exploration of chemical space.

The 'AI in chemistry workshop' repository provides educational resources and hands-on sessions on applying AI and machine learning techniques in chemistry, particularly focusing on molecule generation and data exploration for drug discovery.

Particle Guidance is a tool that enhances the diversity and efficiency of sampling in generative models, specifically applied to molecular conformer generation. It reduces the median error in conformer generation, making it a valuable resource for molecular design and optimization.

This tool implements a reinforced genetic algorithm for structure-based drug design, utilizing neural models to enhance the efficiency of molecular optimization. It aims to intelligently explore chemical space to identify potential drug candidates with improved binding affinity.

The smina-docking-benchmark repository provides tools for evaluating molecular docking models and optimizing generated molecules. It includes benchmarks for various models and allows users to generate and assess molecules based on their docking scores.

MCMG is a software tool designed for generating molecules based on specific constraints using advanced machine learning techniques. It allows users to customize tasks for molecular generation, making it suitable for applications in drug discovery and molecular design.

CA RFdiffusion is a repository that provides training and inference code for a protein structure diffusion model. It generates protein backbones through a two-step process involving diffusion and refinement, making it a valuable tool for computational protein design.

The ESMDiff repository provides a framework for generating protein conformations using structure language models. It employs a discrete variational auto-encoder and conditional language modeling to capture conformational distributions, making it a valuable tool for protein design and molecular dynamics.

Chemeleon is a text-guided diffusion model designed for generating crystal structures based on natural language descriptions or specified chemical compositions. It aids in material discovery by allowing users to explore and create crystal structures, making it a valuable tool in computational chemistry.

BO-ICL is a tool that utilizes Bayesian optimization and in-context learning to predict molecular properties and generate new molecular data. It allows users to perform regression with uncertainties and supports inverse design for proposing new molecular structures based on desired properties.

PCMol is a multi-target de novo molecular generator that utilizes AlphaFold's protein embeddings to create relevant molecules for various protein targets. It is designed to aid in drug discovery by generating SMILES representations of potential drug candidates.

Geometry-Deep-Learning-for-Drug-Discovery is a repository that explores the application of geometric deep learning methods in drug discovery and life sciences. It includes functionalities for predicting molecular properties, designing molecules, and provides datasets and benchmarks relevant to molecular machine learning.

FLAG is a framework for generating 3D molecules that bind to target proteins using a fragment-based approach. It employs deep generative models to create realistic molecular structures, enhancing the process of structure-based drug design.

SurfGen is a tool designed for generating 3D molecular structures by learning from topological surfaces and geometric features. It utilizes a dataset for training and provides functionalities for generating molecules, particularly targeting pharmaceutical applications.

DeepFMPO is a tool designed for optimizing drug design through deep reinforcement learning. It allows users to generate and modify molecules based on lead compounds, facilitating the exploration of new drug candidates.

Delete is a tool for deep lead optimization that generates new molecular structures by utilizing a structure-aware network and deleting specific fragments from lead compounds. It is designed to assist in drug discovery by suggesting new ligands that fit within protein pockets.

Frame2seq is a tool for generating and scoring protein sequences using a structured-conditioned masked language model. It allows users to design sequences based on PDB structures and provides functionality for scoring these sequences based on their likelihood.

This repository provides a PyTorch implementation of a variational autoencoder designed for automatic chemical design. It focuses on generating continuous representations of molecules, which can be utilized for drug discovery and molecular optimization.

PGMG is a PyTorch implementation that utilizes a pharmacophore-guided deep learning model to generate bioactive molecules with structural diversity. It allows users to input pharmacophore hypotheses and generates a large number of candidate molecules that meet specified conditions.

This repository contains code for a deep generative model aimed at generating molecular fragments, as presented in the AISTATS 2020 paper. It includes functionalities for data preprocessing, model training, sampling, and postprocessing, making it a useful tool for molecular design.

This repository serves as a companion to the book 'Machine Learning for Drug Discovery', providing code and data for various machine learning techniques applied to drug discovery. It covers topics such as ligand-based screening, generative models for de novo design, and structure-based drug design, making it a valuable resource for researchers in the field.

cG-SchNet is a conditional generative neural network that focuses on the inverse design of 3D molecular structures. It allows users to generate molecules based on specified conditions, leveraging a dataset of small molecules to train the model.