Pasar al contenido principal

Development of a predictive Artificial Intelligence framework to accelerate chemical compound discovery

PTQ2023-012950

•
Nombre agencia financiadora Agencia Estatal de Investigación
Acrónimo agencia financiadora AEI
Programa Programa Estatal para Desarrollar, Atraer y Retener Talento
Subprograma Subprograma Estatal de Incorporación
Convocatoria Torres Quevedo
Año convocatoria 2023
Plan Estatal 2021-2023 Plan Estatal de Investigación Científica y Técnica y de Innovación
Centro beneficiario AITENEA BIOTECH SL

Resultados relacionados

Resultados totales (Incluyendo duplicados): 2
Encontrada(s) 1 página(s)

DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

MolOrgGPT: De Novo Generation via Large Language Models and Reinforcement Learning

  • Varas Pardo, Pablo
  • Toledano, Oscar
  • Marcos-Ayuso, Guillermo
  • Quesada, David
  • Campillo, Nuria E
We present a general framework for the de novo design of small molecules with desirable chemical properties, developed to aid the creation of novel chemical entities with potential therapeutic use. The system is built upon a foundational Large Language Model trained on a large comprehensive chemical database capable of generating structurally diverse and synthetically accessible compounds. It is then fine-tuned through reinforcement learning to enhance its capacity to generate molecules tailored to specific biological targets. As a case study, we apply this framework to design molecules targeting key proteins involved in Alzheimer's disease. The generated compounds underwent molecular docking studies to assess their binding affinities and prioritize candidates with optimal predicted interactions. The top-ranked molecules were further analyzed based on their binding modes and key molecular interactions with the target proteins. The results suggest that our generative model produces viable, drug-like molecules with favorable interactions, underscoring its potential as a valuable tool in early stage drug discovery., This work was funded by grants PID2022-137331OB-C33 (N.E.C.) and MICIU/AEI/10.13039/501100011033, FEDER and EUandproject TED2021-129970B-C21 (N.E.C.) financed by MCIN/AEI and European Union “NextGenerationEU”/ PRTR, IND2023/BMD-27452 (P.V.P.) financed by CAM, and PTQ2023-012950 (G.M.-A.) financed by MCIU/AEI, Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

PockLigGPT: Pocket-Sequence-Conditioned Molecular Generation with GPTs and RL

  • Varas Pardo, Pablo
  • Marcos-Ayuso, Guillermo
  • Ulzurrun, Eugenia
  • Ríos Insua, David
  • Campillo, Nuria E.
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., This work has been supported by grant PID2022-137331OB-C33 (to NEC), which was funded by MICIU/AEI/10.13039/501100011033, FEDER, and EU; IND2023/BMD-27452 (PV), financed by CAM; and PTQ2023-012950 (GMA), financed by MICIU/AEI. The authors acknowledge the use of computational resources provided by the CESGA (Centro de Supercomputación de Galicia). This article is dedicated to the memory of Carlos Requena Triguero (1992–2026), whose enthusiasm for science, dedication, and exceptional personal qualities left a lasting impression on all who had the privilege of knowing and working with him., Peer reviewed