@inproceedings{dey-etal-2025-large,
title = "Large Language Models for Controllable Multi-property Multi-objective Molecule Optimization",
author = "Dey, Vishal and
Hu, Xiao and
Ning, Xia",
editor = "Christodoulopoulos, Christos and
Chakraborty, Tanmoy and
Rose, Carolyn and
Peng, Violet",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2025",
month = nov,
year = "2025",
address = "Suzhou, China",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.findings-emnlp.1145/",
doi = "10.18653/v1/2025.findings-emnlp.1145",
pages = "20996--21023",
ISBN = "979-8-89176-335-7",
abstract = "In real-world drug design, molecule optimization requires selectively improving multiple molecular properties up to pharmaceutically relevant levels, while maintaining others that already meet such criteria. However, existing computational approaches and instruction-tuned LLMs fail to capture such nuanced property-specific objectives, limiting their practical applicability. To address this, we introduce C-MuMOInstruct, the first instruction-tuning dataset focused on multi-property optimization with explicit, property-specific objectives. Leveraging C-MuMOInstruct, we develop \textbf{GeLLM{\ensuremath{^4}}O-C}s, a series of instruction-tuned LLMs that can perform targeted property-specific optimization. Our experiments across 5 in-distribution and 5 out-of-distribution tasks show that \textbf{GeLLM{\ensuremath{^4}}O-C}s consistently outperform strong baselines, achieving up to 126{\%} higher success rate. Notably, \textbf{GeLLM{\ensuremath{^4}}O-C}s exhibit impressive 0-shot generalization to novel optimization tasks and unseen instructions. This offers a step toward a foundational LLM to support realistic, diverse optimizations with property-specific objectives. C-MuMOInstruct and code are accessible through \url{https://github.com/ninglab/GeLLMO-C}."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="dey-etal-2025-large">
<titleInfo>
<title>Large Language Models for Controllable Multi-property Multi-objective Molecule Optimization</title>
</titleInfo>
<name type="personal">
<namePart type="given">Vishal</namePart>
<namePart type="family">Dey</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiao</namePart>
<namePart type="family">Hu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xia</namePart>
<namePart type="family">Ning</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2025</title>
</titleInfo>
<name type="personal">
<namePart type="given">Christos</namePart>
<namePart type="family">Christodoulopoulos</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tanmoy</namePart>
<namePart type="family">Chakraborty</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Carolyn</namePart>
<namePart type="family">Rose</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Violet</namePart>
<namePart type="family">Peng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Suzhou, China</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-335-7</identifier>
</relatedItem>
<abstract>In real-world drug design, molecule optimization requires selectively improving multiple molecular properties up to pharmaceutically relevant levels, while maintaining others that already meet such criteria. However, existing computational approaches and instruction-tuned LLMs fail to capture such nuanced property-specific objectives, limiting their practical applicability. To address this, we introduce C-MuMOInstruct, the first instruction-tuning dataset focused on multi-property optimization with explicit, property-specific objectives. Leveraging C-MuMOInstruct, we develop GeLLM\ensuremath⁴O-Cs, a series of instruction-tuned LLMs that can perform targeted property-specific optimization. Our experiments across 5 in-distribution and 5 out-of-distribution tasks show that GeLLM\ensuremath⁴O-Cs consistently outperform strong baselines, achieving up to 126% higher success rate. Notably, GeLLM\ensuremath⁴O-Cs exhibit impressive 0-shot generalization to novel optimization tasks and unseen instructions. This offers a step toward a foundational LLM to support realistic, diverse optimizations with property-specific objectives. C-MuMOInstruct and code are accessible through https://github.com/ninglab/GeLLMO-C.</abstract>
<identifier type="citekey">dey-etal-2025-large</identifier>
<identifier type="doi">10.18653/v1/2025.findings-emnlp.1145</identifier>
<location>
<url>https://aclanthology.org/2025.findings-emnlp.1145/</url>
</location>
<part>
<date>2025-11</date>
<extent unit="page">
<start>20996</start>
<end>21023</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Large Language Models for Controllable Multi-property Multi-objective Molecule Optimization
%A Dey, Vishal
%A Hu, Xiao
%A Ning, Xia
%Y Christodoulopoulos, Christos
%Y Chakraborty, Tanmoy
%Y Rose, Carolyn
%Y Peng, Violet
%S Findings of the Association for Computational Linguistics: EMNLP 2025
%D 2025
%8 November
%I Association for Computational Linguistics
%C Suzhou, China
%@ 979-8-89176-335-7
%F dey-etal-2025-large
%X In real-world drug design, molecule optimization requires selectively improving multiple molecular properties up to pharmaceutically relevant levels, while maintaining others that already meet such criteria. However, existing computational approaches and instruction-tuned LLMs fail to capture such nuanced property-specific objectives, limiting their practical applicability. To address this, we introduce C-MuMOInstruct, the first instruction-tuning dataset focused on multi-property optimization with explicit, property-specific objectives. Leveraging C-MuMOInstruct, we develop GeLLM\ensuremath⁴O-Cs, a series of instruction-tuned LLMs that can perform targeted property-specific optimization. Our experiments across 5 in-distribution and 5 out-of-distribution tasks show that GeLLM\ensuremath⁴O-Cs consistently outperform strong baselines, achieving up to 126% higher success rate. Notably, GeLLM\ensuremath⁴O-Cs exhibit impressive 0-shot generalization to novel optimization tasks and unseen instructions. This offers a step toward a foundational LLM to support realistic, diverse optimizations with property-specific objectives. C-MuMOInstruct and code are accessible through https://github.com/ninglab/GeLLMO-C.
%R 10.18653/v1/2025.findings-emnlp.1145
%U https://aclanthology.org/2025.findings-emnlp.1145/
%U https://doi.org/10.18653/v1/2025.findings-emnlp.1145
%P 20996-21023
Markdown (Informal)
[Large Language Models for Controllable Multi-property Multi-objective Molecule Optimization](https://aclanthology.org/2025.findings-emnlp.1145/) (Dey et al., Findings 2025)
ACL