Aug 2026· Journal of Chemical Information and Modeling· 0 citations· 88 references
TL;DR
PockLigGPT achieves competitive docking-oriented performance under a standardized evaluation protocol while maintaining chemical plausibility, favorable physicochemical profiles, and Lipinski-based drug-likeness.
Abstract
De novo drug design aims to generate molecules targeting specific protein pockets while retaining chemical plausibility and drug-like properties. Recent 3D structure-based generative methods explicitly model pocket-ligand geometry, but this does not always translate into chemically realistic or practically usable candidate molecules. Molecular language models provide a complementary sequence-based alternative. However, it remains unclear whether sequence-based pocket information can effectively guide ligand generation, whether multistage training improves pocket-specific generation, and whether docking-guided reinforcement learning can be integrated into a practical generation pipeline. We introduce PockLigGPT, a GPT-based framework for pocket-sequence-conditioned molecular generation. Rather than producing fixed 3D coordinates, PockLigGPT formulates ligand design as a sequence-generation problem conditioned on the amino acid composition of the protein pocket. The model is trained in four stages: large-scale chemical pretraining based on ZINC20; bioactivity-oriented adaptation based on ChEMBL; pocket-sequence-conditioned fine-tuning using binding-pocket amino acid sequences paired with ligands; and, finally, pocket-specific docking-guided reinforcement learning using AutoDock Vina-based rewards. PockLigGPT achieves competitive docking-oriented performance under a standardized evaluation protocol while maintaining chemical plausibility, favorable physicochemical profiles, and Lipinski-based drug-likeness. Docking studies on Alzheimer’s disease-associated targets and token-level analyses further support the utility of PockLigGPT for de novo drug design.
Drug discovery and development is time-consuming and resource-intensive, motivating computational approaches such as diffusion models for de novo drug design. Many such models follow the structure-based drug design (SBDD) paradigm, generating molecules to fit a target binding pocket. However, existing diffusion-based SBDD methods typically couple pocket and ligand representation learning, model interactions only at the atom level, and prioritize binding affinity over other developability properties. Here, we introduce conDitar-dev, a conditional diffusion-based SBDD framework for generating ligands with strong binding affinities and favorable ADMET properties. It consists of three modules: msPRL, a pretrained multi-scale pocket representation learning module; conDitar, a pocket-conditioned diffusion model guided by msPRL representations; and paOPT, a generation-time method for optimizing ligand developability. On a newly curated benchmark of human disease targets, conDitar outperforms state-of-the-art SBDD baselines, achieving an average binding score of -8.85 kcal/mol. Across five ADMET properties, conDitar-dev improves performance by up to 73% over conDitar. To further validate the abilities of conDitar-dev to generate developable molecules, we have applied it to two validated druggable targets: programmed death-ligand 1 (PD-L1) and colony-stimulating factor 1 receptor (CSF1R) proteins. Top-ranked generatively designed molecules and their analogs have been experimentally synthesized and biologically tested. Two molecules generated directly by conDitar-dev for PD-L1 exhibited SPR-derived $K_D$ values of 3.49 and 3.75 $\mu$M, respectively. Hit expansion based on conDitar-dev-designed molecules identified selective CSF1R inhibitors with IC$_{50}$ values as low as 200 nM, while also uncovering opportunities for drug repositioning.
Ruoxi Gao, Jiangweizhi Peng, Ziqi Chen et al.· 0 citations
A clear pattern is revealed in LLM spatial capabilities: while they still lag behind state-of-the-art approaches, they are promising and can handle multiple spatial constraints simultaneously, enabling scaling to heterogeneous setups.
Thomas MacDougall, Maksim Kuznetsov, Roman Schutski et al.· 0 citations
Leveraging large language models (LLMs) for molecular generation has shown remarkable potential in chemical and drug design. Current methods primarily rely on supervised training or fine-tuning with limited datasets, which are insufficient to capture complex molecular design objectives. While some approaches attempt to guide generation toward specific goals, they often lack direct optimization mechanisms, making it difficult to align generated molecules with desired properties. To tackle these challenges, we propose \textbf{LLMol}, a principled reinforcement learning framework that directly incorporates verifiable rewards for targeted molecule generation. The key insight is to formulate molecular design as a goal-conditioned sequence prediction task, where verifiable rewards serve as explicit supervision to drive generation toward desired objectives. LLMol follows a two-stage training paradigm combining supervised learning and reinforcement learning. In the first stage, large language models are supervised fine-tuned to capture chemical syntax and molecular distributions. In the second stage, we introduce Reinforcement Learning with Verifiable Rewards (RLVR), which directly integrates property-based reward signals to guide molecular generation toward task-specific objectives. To address the high variance and instability common in discrete sequence optimization, we adopt Group Relative Policy Optimization (GRPO), a stable on-policy algorithm that smooths reward signals and improves training robustness. This framework enables LLMol to effectively handle a range of molecular design tasks, including single-property targeting (e.g., penalized logP, QED) and structure-constrained optimization. Experimental results demonstrate that LLMol consistently outperforms existing methods, achieving higher success rates and improved efficiency across diverse molecular benchmarks.
Small-molecule drug discovery relies on iterative molecular optimization, where chemists repeatedly modify candidate compounds to balance multiple competing properties such as efficacy, toxicity, and solubility. Recent advances in generative AI (GenAI) have shown promise in accelerating this process by automatically proposing new molecular structures or targeted modifications. However, existing GenAI-based molecular design tools remain poorly aligned with experts'real-world workflows. Specifically, they offer limited support for specifying structure-level modification intents on molecules, provide insufficient transparency into model-generated modifications, and lack integrated support for downstream property evaluation with external computational tools. To address these challenges, we introduce MolecularCanvas, an interactive system that enables users to iteratively construct an optimization context by integrating high-level goals, structure-level annotations, property constraints, and reference-based preferences. This context guides the generation of candidate molecules across diverse molecular structures. MolecularCanvas further enhances transparency by providing evidence for AI-generated suggestions and streamlines molecular evaluation by integrating commonly used computational tools for property assessment into a unified interface. Finally, a user study with 12 participants demonstrates the usefulness and effectiveness of MolecularCanvas in helping users optimize candidate molecules.
Haoyu Dong, Rui Sheng, Shuhao Zhang et al.· 0 citations
PGFS++ is introduced, a synthesis-aware reinforcement learning framework for input-specific molecular improvement that improves target properties while preserving high output diversity, and experiments show that PGFS++ improves target properties while preserving high output diversity.
Boqiao Zhang, Godbless James, S. Gottipati et al.· 0 citations
The resulting model, HydrAffinity, is an interaction-free, dynamic sparse model that uses pre-trained encoders and MoE for parameter-efficient learning and outperforms all interaction-free methods and matches state-of-the-art interaction-based methods on CASF-2016.