Scaffold-Conditioned Preference Triplets for Controllable Molecular Optimization with Large Language Models

TL;DR

This paper introduces SCPT, a pipeline for scaffold-conditioned preference triplets that enables controllable, scaffold-preserving molecular optimization using large language models, improving success rates and property gains.

Contribution

The paper presents a novel scaffold-conditioned preference triplet construction method for training LLMs to perform controlled molecular edits that preserve scaffolds.

Findings

SCPT improves optimization success and property gains.

Models trained on limited supervision generalize well to multi-property tasks.

SCPT allows systematic control over similarity and property trade-offs.

Abstract

Molecular property optimization is central to drug discovery, yet many deep learning methods rely on black-box scoring and offer limited control over scaffold preservation, often producing unstable or biologically implausible edits. While large language models (LLMs) are promising molecular generators, optimization remains constrained by the lack of chemistry-grounded preference supervision and principled data curation. We introduce \textbf{Scaffold-Conditioned Preference Triplets (SCPT)}, a pipeline that constructs similarity-constrained triplets $\langle\text{scaffold}, \text{better}, \text{worse}\rangle$ via scaffold alignment and chemistry-driven filters for validity, synthesizability, and meaningful property gains. Using these preferences, we align a pretrained molecular LLM as a conditional editor, enabling property-improving edits that retain the scaffold. Across single- and multi-objective benchmarks, SCPT improves optimization success and property gains while maintaining higher scaffold similarity than competitive baselines. Compared with representative non-LLM molecular optimization methods, SCPT-trained LLMs are better suited to scaffold-constrained and multi-objective optimization. In addition, models trained on single-property and two-property supervision generalize effectively to three-property tasks, indicating promising extrapolative generalization under limited higher-order supervision. SCPT also provides controllable data-construction knobs that yield a predictable similarity-gain frontier, enabling systematic adaptation to diverse optimization regimes.