Large Language Model Agent for Modular Task Execution in Drug Discovery
Агент на основе большой языковой модели для модульного выполнения задач в разработке лекарственных средств
2026-02-09
SCID: 54.1/phux9kte
Discuss with AI
large language model agentmodular task executionmolecular generationprotein-ligand structure generationretrieval-augmented generation
Figures from the paper
Abstract (AI)
We present a modular framework powered by large language models (LLMs) that automates and streamlines key tasks across the early stage computational drug discovery pipeline. By combining LLM reasoning with domain-specific tools, the framework performs biomedical data retrieval, literature-grounded question answering via retrieval-augmented generation, molecular generation, multiproperty prediction, property-aware molecular refinement, and 3D protein-ligand structure generation. The agent autonomously retrieves relevant biomolecular information, including FASTA sequences, SMILES representations, and literature, and answers mechanistic questions with improved contextual accuracy compared to standard LLMs. It then generates chemically diverse seed molecules and predicted 75 properties, including ADMET-related and general physicochemical descriptors, which guids iterative molecular refinement. Across two refinement rounds, the number of molecules with QED >0.6 increased from 34 to 55. The number of molecules satisfying empirical drug-likeness filters also rose; for example, compliance with the Ghose filter increased from 32 to 55 within a pool of 100 molecules. The framework also employed Boltz-2 to generate 3D protein-ligand complexes and provide rapid binding affinity estimates for candidate compounds. These results demonstrate that the approach effectively supports molecular screening, prioritization, and structure evaluation. Its modular design enables flexible integration of evolving tools and models, providing a scalable foundation for AI-assisted therapeutic discovery.
Key Findings
1
A modular LLM-based agent integrates biomedical retrieval, literature-grounded question answering, molecular generation, multiproperty prediction, refinement, and 3D complex generation.
2
Across two refinement rounds, molecules with QED >0.6 increased from 34 to 55, and Ghose-filter-compliant molecules increased from 32 to 55 among 100 molecules.
3
Boltz-2-generated protein-ligand complexes and rapid binding-affinity estimates support candidate screening, prioritization, and structural evaluation.
4
The framework retrieves FASTA sequences, SMILES, and literature, while retrieval-augmented answers provide improved contextual accuracy compared with standard LLMs.
5
The system predicts 75 molecular properties, including ADMET-related and physicochemical descriptors, to guide property-aware iterative refinement.
Research Object
early-stage computational drug discovery pipeline, including candidate molecules and protein–ligand complexes
Research Subject
automated molecular generation, multiproperty evaluation, property-aware refinement, and 3D structure-based screening and prioritization
Publication Details
Publication Date
2026-02-09
Journal
Publisher
ISSN
Cited by
12
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest
References available in scid.ai9
Highly accurate protein structure prediction with AlphaFold2021
Accurate structure prediction of biomolecular interactions with AlphaFold 32024
MoleculeNet: a benchmark for molecular machine learning2017
ADMETlab 2.0: an integrated online platform for accurate and comprehensive predictions of ADMET properties2021
The AI Scientist: Towards Fully Automated Open-Ended Scientific Discovery2024
Mordred: a molecular descriptor calculator2018
Autonomous chemical research with large language models2023
Augmenting large language models with chemistry tools2024
Modular large language model agents for multi-task computational materials science2026