Target Aware De Novo Generation

• 3D-Generative-SBDD: Focuses on structure-based drug design (SBDD) using a 3D generative model to sample molecules for specific protein pockets.
  
  
  
• AlphaDrug: protein target specific de novo molecular generation (generative chemistry) (standalone).
  
  
  
• BindDM: BindDM extracts subcomplex from protein-ligand complex, and utilizes it to enhance the binding-adaptive 3D molecule generation in complex
  
  
  
• cMolGPT: cMolGPT, a Conditional Generative Pre-trained Transformer, aims for de novo molecular design by enforcing target specificity during the generation process, emphasizing the incorporation of target embeddings.
  
  
  
• DESERT: Zero-shot 3D drug design by sketching and generating, as presented at NeurIPS 2022.
  
  
  
• DRAGONFLY:
  
  
  
• DrugGEN:
  
  
  
• HySonLab:
  
  
  
• Ligand_Generation: This project focuses on target-aware variational autoencoders for ligand generation, employing multimodal protein representation learning for structure-based drug design.
  
  
  
• PCMol: A multi-target model for de novo molecule generation, using AlphaFold protein embeddings for thousands of protein targets.
  
  
  
• PETrans:
  
  
  
• Pocket2Mol: Uses equivariant graph neural networks for efficient molecular sampling based on 3D protein pockets.
  
  
  
• PocketFlow: data-and-knowledge driven structure-based molecular generative model
  
  
  
• RELATION: A software for DL-based de novo drug design, focusing on generating molecules based on target protein interactions.
  
  
  
• ResGen: pocket-aware 3D molecular generation model based on parallel multiscale modelling (standalone).
  
  
  
• SCRdkit: assesses the 3D similarity between generated molecules and a reference molecule
  
  
  
• SINGA: SINGA is a Molecular Sampling model with Protein-Ligand Interactions aware Generative Adversarial Network, focusing on generating molecules considering their interactions with proteins.
  
  
• DrugHIVE: Structure-Based Drug Design with a Deep Hierarchical Generative Model
  
  
  

Active-learning

• ChemSpaceAL: an Efficient Active Learning Methodology Applied to Protein-Specific Molecular Generation (generative chemistry) (standalone, 2024).
  
  
  

Benchmarks

• From Theory to Therapy: Reframing SBDD Model Evaluation via Practical Metrics:
  
• CBGBench:
  
  

Diffusion

• DiffDec: A model that uses an end-to-end equivariant diffusion process for optimizing molecular structures through scaffold decoration conditioned on 3D protein pockets.
  
  
  
• DiffSBDD: Structure-based Drug Design with Equivariant Diffusion Models (standalone, 2024).
  
  
  
• MolSnapper: Conditioning Diffusion for Structure Based Drug Design
  
  
  .)
• TargetDiff: 3D Equivariant Diffusion for Target-Aware Molecule Generation and Affinity Prediction (standalone, 2024).
  
  
  
• NucleusDiff:
  
  
  

Docking-based

• FREED: Utilizes explorative experience replay in a generative reinforcement learning setup for drug design.
  
  
  
• OptiMol:
  
  
  
• SampleDock: Navigating Chemical Space By Interfacing Generative Artificial Intelligence and Molecular Docking
  
  
  
  
• Lib-INVENT: Reaction-based Generative Scaffold Decoration for in Silico Library Design
  
  
  
• KGDiff: A model for explainable target-aware molecule generation with knowledge guidance.
  
  
  
• MORLD:
  
  
  
• liGAN: A project for structure-based drug discovery using deep generative models of atomic density grids.
  
  
  
• LS-MolGen: A ligand-and-structure dual-driven deep reinforcement learning method for target-specific molecular generation, integrating docking scores and bioactivity data for molecule optimization.
  
  
  
• PMDM: dual diffusion model enables 3D binding bioactive molecule generation and lead optimization given target pockets
  
  
  
• PocketOptimizer 2.0: Design of small molecule‐binding pockets in proteins (standalone)
  
  
  
• RGA: A reinforcement learning-based genetic algorithm for structure-based drug design, introduced at NeurIPS 2022.
  
  
  
• SBMolGen: integrates a recurrent neural network, a Monte Carlo tree search, and docking simulations
  
  
  
• SECSE:
  
  
  
• DockStream:
  
  
  

Dual-target

• Alx-Fuse:
  
  
  

Fragment-based

• FragGen: Fragment wise 3D Structure-based Molecular Generation
  
  
  .)

Interaction fingerprint constrained

• IFP-RNN: A molecule generative model used interaction fingerprint (docking pose) as constraints.
  
  

Language Model

• Lingo3DMol: A pocket-based 3D molecule generation method that combines language model capabilities with 3D coordinate generation and geometric deep learning.
  
  
  

PROTACs

• PROTACable: Integrative Computational Pipeline of 3-D Modeling and Deep Learning To Automate the De Novo Design of PROTACs (standalone, published 2024).
  
  
  .)

Pharmacophore

• DEVELOP:
  
  
  

Physics-based generation

• SPOTLIGHT:
  
  

Protein-ligand interactions

• DeepICL: 3D molecular generative framework for interaction-guided drug design
  
  
  
• DrugGPS: Focuses on learning subpocket prototypes for generalizable structure-based drug design, introduced at ICML 2023.
  
  
  
• FLAG: A Fragment-based Ligand Generation framework to generate 3D molecules with valid and realistic substructures fragment-by-fragment.
  
  
  
• GraphBP: Implements a method for generating 3D molecules targeting protein binding, presented at ICML 2022.
  
  
  

Reviews

• Structure-based Drug Design Benchmark: Do 3D Methods Really Dominate?: benchmark to evaluate the performance of sixteen models across these different algorithmic foundations by assessing the pharmaceutical properties of the generated molecules and their docking affinities with specified target proteins
  
  
  
• Integrating structure-based approaches in generative molecular design [2023]:
  
• Docking-based generative approaches in the search for new drug candidates [2022]:
  
• Generative Deep Learning for Targeted Compound Design [2021]:
  

Scaffold Hopping

• DeepHop:
  
  
  

Sequence-based

• Bajorath_Gen: