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

• URL: http://arxiv.org/abs/2604.12350v1
• Date: Tue, 14 Apr 2026 06:38:07 GMT
• Title: Scaffold-Conditioned Preference Triplets for Controllable Molecular Optimization with Large Language Models
• Abstract summary: We introduce textbfScaffold-Conditioned Preference Triplets ( SCPT), a pipeline that constructs similarity-constrained triplets $langletextscaffold, textbetter, textworserangle$ via scaffold alignment and chemistry-driven filters for validity, synthesizability, and meaningful property gains.<n>Compared with representative non-LLM molecular optimization methods, SCPT-trained LLMs are better suited to scaffold-constrained and multi-objective optimization.
• Score: 32.68069719676566
• License: http://creativecommons.org/licenses/by/4.0/
• 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.

Related papers

• APO: Unsupervised Atomic Policy Optimization for 3D Structure Prediction of Atomic Systems [64.56252173115246]
  We propose (Atomic Policy Optimization) a fully unsupervised alignment framework that eliminates the need for ground-truth reference structures.<n>APO adapts group-relative policy optimization to 3D atomic environments, utilizing a novel dual-reward mechanism.<n>Our results suggest that intrinsic physical consistency can serve as a superior guide for alignment compared to noisy, supervised coordinate matching.
  arXiv  Detail & Related papers  (2026-07-30T17:21:58Z)
• Be Your Own Teacher: Steering Protein Language Models via Unsupervised Reward Optimization [85.4863852505905]
  We introduce unsupervised reward optimization of protein language models (PLMs)<n>Our key insight is that task-agnostic rewards, which combine intrinsic model uncertainty with semantic consistency informed by protein representation models, exhibit strong correlation with controllability measures across base models and temperature regimes.<n>Our framework provides a scalable pathway toward controllable biomolecular design in settings where labeled preferences or experimental feedback are scarce or unavailable.
  arXiv  Detail & Related papers  (2026-06-17T11:42:01Z)
• PVminerLLM2: Improving Structured Extraction of Patient Voice via Preference Optimization [0.19582799051022884]
  Patient-generated text contains critical information on patients' lived experiences, social context, and care engagement.<n>Prior work introduced the PV-Miner benchmark and PVMinerLLM models for structured extraction.<n>We present PVminerLLM2, an improved set of LLMs for structured patient voice extraction.
  arXiv  Detail & Related papers  (2026-06-15T00:18:47Z)
• Reasoning Matters: Mitigate Hallucination in Multimodal Large Reasoning Models via Reasoning-Conditioned Preference Optimization [59.20570719781289]
  Multimodal Large Reasoning Models introduce the reasoning paradigm, demonstrating strong capabilities on complex vision-language tasks.<n>Existing training-based methods typically mitigate hallucinations through response-level direct preference optimization (DPO), where the Chain-of-Thought (CoT) and the final answer are treated as a monolithic output and optimized jointly.<n>We derive Reasoning-Conditioned Direct Preference Optimization (RC-DPO), which models the CoT as a condition for answer generation and contrasts the preference for the same preferred answer under different CoT conditions, promoting answer-supportive reasoning chain alignment.
  arXiv  Detail & Related papers  (2026-05-27T03:27:23Z)
• C-MORAL: Controllable Multi-Objective Molecular Optimization with Reinforcement Alignment for LLMs [16.006924868560795]
  C-Moral is a reinforcement learning framework for controllable multi-objective molecular optimization.<n>Experiments on the C-MuMOInstruct benchmark show that C-Moral consistently outperforms state-of-the-art models.
  arXiv  Detail & Related papers  (2026-04-24T23:11:44Z)
• MAESTRO: Meta-learning Adaptive Estimation of Scalarization Trade-offs for Reward Optimization [56.074760766965085]
  Group-Relative Policy Optimization has emerged as an efficient paradigm for aligning Large Language Models (LLMs)<n>We propose MAESTRO, which treats reward scalarization as a dynamic latent policy, leveraging the model's terminal hidden states as a semantic bottleneck.<n>We formulate this as a contextual bandit problem within a bi-level optimization framework, where a lightweight Conductor network co-evolves with the policy by utilizing group-relative advantages as a meta-reward signal.
  arXiv  Detail & Related papers  (2026-01-12T05:02:48Z)
• Bayesian Optimization for Molecules Should Be Pareto-Aware [6.877358955271652]
  Multi-objective Bayesian optimization (MOBO) provides a principled framework for navigating trade-offs in molecular design.<n>We benchmark a simple MOBO strategy against a simple fixed-weight scalarized baseline using Expected Improvement (EI)<n>Our results show that even strong deterministic instantiations can underperform in low-data regimes.
  arXiv  Detail & Related papers  (2025-07-18T07:12:19Z)
• Mitigating Hallucination Through Theory-Consistent Symmetric Multimodal Preference Optimization [69.05600758833471]
  Direct Preference Optimization (DPO) has emerged as an effective approach for mitigating hallucination in Multimodal Large Language Models (MLLMs)<n>We propose a Symmetric Multimodal Preference Optimization (SymMPO) which conducts symmetric preference learning with direct preference supervision (i.e., response pairs)<n>In addition to conventional ordinal preference learning, SymMPO introduces a preference margin consistency loss to quantitatively regulate the preference gap between symmetric preference pairs.
  arXiv  Detail & Related papers  (2025-06-13T12:29:15Z)
• Large Language Models for Controllable Multi-property Multi-objective Molecule Optimization [2.152507712409726]
  We introduce C-MuMOInstruct, the first instruction-tuning dataset focused on multi-property optimization with explicit, property-specific objectives.<n>We develop GeLLMO-Cs, a series of instruction-tuned LLMs that can perform targeted property-specific optimization.<n>Our experiments show that GeLLMO-Cs consistently outperform strong baselines, achieving up to 126% higher success rate.
  arXiv  Detail & Related papers  (2025-05-29T20:29:14Z)
• Latent Preference Coding: Aligning Large Language Models via Discrete Latent Codes [54.93980123979578]
  We introduce Latent Preference Coding (LPC), a novel framework that models the implicit factors as well as their combinations behind holistic preferences.<n>LPC seamlessly integrates with various offline alignment algorithms, automatically inferring the underlying factors and their importance from data.
  arXiv  Detail & Related papers  (2025-05-08T06:59:06Z)
• ScaffoldGPT: A Scaffold-based GPT Model for Drug Optimization [3.240904428766923]
  We introduce ScaffoldGPT, a Generative Pretrained Transformer (GPT) for drug optimization based on molecular scaffolds.<n>A three-stage drug optimization approach integrates pretraining, finetuning, and decoding optimization.<n>We demonstrate via a comprehensive evaluation on COVID and cancer benchmarks that ScaffoldGPT outperforms the competing baselines in drug optimization benchmarks.
  arXiv  Detail & Related papers  (2025-02-09T10:36:33Z)
• Multi-Reference Preference Optimization for Large Language Models [56.84730239046117]
  We introduce a novel closed-form formulation for direct preference optimization using multiple reference models. The resulting algorithm, Multi-Reference Preference Optimization (MRPO), leverages broader prior knowledge from diverse reference models. Our experiments demonstrate that LLMs finetuned with MRPO generalize better in various preference data, regardless of data scarcity or abundance.
  arXiv  Detail & Related papers  (2024-05-26T00:29:04Z)
• DecompOpt: Controllable and Decomposed Diffusion Models for Structure-based Molecular Optimization [49.85944390503957]
  DecompOpt is a structure-based molecular optimization method based on a controllable and diffusion model. We show that DecompOpt can efficiently generate molecules with improved properties than strong de novo baselines.
  arXiv  Detail & Related papers  (2024-03-07T02:53:40Z)
• Beyond One-Preference-Fits-All Alignment: Multi-Objective Direct Preference Optimization [76.09576643028362]
  We present Multi-Objective Direct Preference Optimization (MODPO) for multiple alignment objectives. MODPO folds language modeling directly into reward modeling, training language models as implicit collective reward models. It theoretically yields the same optimal solutions as MORLHF but is practically more stable and efficient.
  arXiv  Detail & Related papers  (2023-10-05T17:35:26Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.