Data-efficient molecular and nanoparticle design via fine-tuned large language models
7 affiliations
- 1Department of Biomedical Engineering
- 2Duke University
- 3Current address: Department of Chemistry, Biochemistry, and Physics,
- 4Florida Southern College
- 5Duke Cancer Institute
- 6Duke University School of Medicine
- 7Center for Quantitative Biodesign
Abstract
Designing molecules with prescribed properties is central to pharmaceutical and materials innovation, but vast chemical design spaces make development lengthy and costly. Generative artificial intelligence can accelerate molecular discovery, but its deployment in many real-world domains remains limited by the need for large task-specific training datasets. To enable generative design when property-labeled data are scarce, we present a data-efficient foundation-model adaptation strategy combining complementary positive and negative property prompts with parameter-efficient fine-tuning. Generated molecules are then ranked by molecular graph-kernel Gaussian process regression for structure-aware prioritization. In a controlled computational setting, the platform generated candidates with novel scaffolds satisfying multiple design objectives, including predicted target binding, synthesizability, and drug-likeness. Although only a few hundred training molecules per target met all three criteria, the model generated thousands of new molecules matching the desired profiles. For prospective validation, we applied the model to nanoparticle design, a domain of high clinical potential but limited excipient diversity. Fine-tuning on only 1,521 labeled drug-excipient pairs - including just 139 nanoparticle-forming examples identified new nanoparticle-forming excipients for the anticancer drug sorafenib. The model proposed lithocholic acid and the sorafenib metabolite M6, and experiments confirmed that both form stable sorafenib nanoparticles with high encapsulation efficiency and increased in vitro potency. These findings show that foundation-model fine-tuning with complementary prompts and graph-based prioritization enables data-efficient molecular design in low-data chemical domains.
Related
- Target-Specific De Novo Drug Design via Fine-Tuned Language Models and Molecular Docking, Molecular Dynamics Simulation Validation
- Mechanism-oriented design of magnetic COF adsorbents for efficient extraction of phenoxy acid herbicides: Insights from DFT calculations and molecular dynamics simulations
- Synthesis, Biological Evaluation, Molecular Docking and DFT Studies of Electronically Tuned Benzylidene-Based Schiff Bases
- Development of Versatile Nanoparticle System for Efficient Drug Delivery into Pulmonary Endothelium.
- Elucidating molecular regulatory rules of nanoparticle biological identity via standardized deep proteomics
BibTeX
@misc{xiang2026dataefficient,
title = {{Data-efficient molecular and nanoparticle design via fine-tuned large language models}},
author = {Yan Xiang and Zilu Zhang and Rebeca T. Stiepel and Lauren A. Onweller and Zachary Fralish and Daniel Reker},
howpublished = {ChemRxiv},
year = {2026},
month = oct,
doi = {10.26434/chemrxiv.15010760/v1},
url = {https://chemrxiv.org/doi/full/10.26434/chemrxiv.15010760/v1},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - Data-efficient molecular and nanoparticle design via fine-tuned large language models
AU - Yan Xiang
AU - Zilu Zhang
AU - Rebeca T. Stiepel
AU - Lauren A. Onweller
AU - Zachary Fralish
AU - Daniel Reker
PY - 2026
DA - 2026/10/21
PB - ChemRxiv
DO - 10.26434/chemrxiv.15010760/v1
UR - https://chemrxiv.org/doi/full/10.26434/chemrxiv.15010760/v1
AB - Designing molecules with prescribed properties is central to pharmaceutical and materials innovation, but vast chemical design spaces make development lengthy and costly. Generative artificial intelligence can accelerate molecular discovery, but its deployment in many real-world domains remains limited by the need for large task-specific training datasets. To enable generative design when property-labeled data are scarce, we present a data-efficient foundation-model adaptation strategy combining complementary positive and negative property prompts with parameter-efficient fine-tuning. Generated molecules are then ranked by molecular graph-kernel Gaussian process regression for structure-aware prioritization. In a controlled computational setting, the platform generated candidates with novel scaffolds satisfying multiple design objectives, including predicted target binding, synthesizability, and drug-likeness. Although only a few hundred training molecules per target met all three criteria, the model generated thousands of new molecules matching the desired profiles. For prospective validation, we applied the model to nanoparticle design, a domain of high clinical potential but limited excipient diversity. Fine-tuning on only 1,521 labeled drug-excipient pairs - including just 139 nanoparticle-forming examples identified new nanoparticle-forming excipients for the anticancer drug sorafenib. The model proposed lithocholic acid and the sorafenib metabolite M6, and experiments confirmed that both form stable sorafenib nanoparticles with high encapsulation efficiency and increased in vitro potency. These findings show that foundation-model fine-tuning with complementary prompts and graph-based prioritization enables data-efficient molecular design in low-data chemical domains.
ER -