Target-Aware Generation

This group covers generative models that condition molecule design on protein target information. The goal is to generate molecules complementary to a specific protein, whether through sequence translation, pocket encoding, or 3D-aware coordinate generation.

Paper	Year	Approach	Key Idea
Evolutionary Design	2021	GA + RNN decoder	Genetic algorithm on ECFP fingerprints with DNN property predictor
Protein-to-Drug	2021	Seq2seq Transformer	Amino acid sequence to SMILES translation
AlphaDrug	2022	Transformer + MCTS	Monte Carlo tree search with docking rollouts for target-specific design
PrefixMol	2023	GPT + prefix tuning	Learnable prefix embeddings for joint pocket and property conditioning
TamGen	2024	GPT + pocket encoder	Pocket-conditioned generation with VAE refinement and experimental validation
Lingo3DMol	2024	LM + geometric DL	Fragment-based SMILES with 3D coordinates for pocket-conditioned design
BindGPT	2025	GPT	Autoregressive SMILES+XYZ generation with RL-based docking optimization

All Notes

Computational Chemistry

Bar chart comparing AlphaDrug docking scores against known ligands across five protein targets

AlphaDrug: MCTS-Guided Target-Specific Drug Design

AlphaDrug generates drug candidates for specific protein targets by combining an Lmser Transformer (with hierarchical encoder-decoder skip connections) and Monte Carlo tree search guided by docking scores, achieving higher binding affinities than known ligands on 86% of test proteins.

Computational Chemistry

Bar chart showing BindGPT RL achieves best Vina binding scores compared to baselines

BindGPT: GPT for 3D Molecular Design and Docking

BindGPT formulates 3D molecular design as autoregressive text generation over combined SMILES and XYZ tokens, using large-scale pre-training and reinforcement learning to achieve competitive pocket-conditioned molecule generation.

Computational Chemistry

Bar chart showing peak absorption wavelength increasing across evolutionary generations

Evolutionary Molecular Design via Deep Learning + GA

An evolutionary molecular design framework that evolves ECFP fingerprint vectors using a genetic algorithm, reconstructs valid SMILES via an RNN decoder, and evaluates fitness with a DNN property predictor.

Computational Chemistry

Bar chart showing Lingo3DMol achieves best Vina docking scores on DUD-E compared to five baselines

Lingo3DMol: Language Model for 3D Molecule Design

Lingo3DMol introduces FSMILES, a fragment-based SMILES representation with local and global coordinates, to generate drug-like 3D molecules in protein pockets via a transformer language model.

Computational Chemistry

Bar chart showing PrefixMol Vina scores across different conditioning modes: target, property, combined, and scaffold

PrefixMol: Prefix Embeddings for Drug Molecule Design

PrefixMol prepends learnable condition vectors to a GPT transformer for SMILES generation, enabling joint control over binding pocket targeting and chemical properties like QED, SA, and LogP.

Computational Chemistry

Bar chart comparing docking scores of generated vs known ligands for CDK2 and EGFR targets

Protein-to-Drug Molecule Translation via Transformer

Applies the Transformer architecture to generate drug-like molecules conditioned on protein amino acid sequences, treating target-specific de novo drug design as a sequence-to-sequence translation problem.

Computational Chemistry

Diagram showing the TamGen three-stage pipeline from protein pocket encoding through compound generation to experimental testing

TamGen: GPT-Based Target-Aware Drug Design and Generation

Introduces TamGen, a target-aware molecular generation method using a pre-trained GPT-like chemical language model with protein structure conditioning. A Design-Refine-Test pipeline discovers 14 inhibitors against tuberculosis ClpP protease, with IC50 values as low as 1.9 uM.

All Notes#

All Notes