Multi-objective fluorescent molecule design with a data-physics dual-driven generative framework

2026-01-20

AI Summary

This work addresses the challenges in inverse design of fluorescent small molecules—namely, the vastness of chemical space, low efficiency and poor generalizability of conventional methods, and the high computational cost of quantum mechanical calculations—by introducing LUMOS, a novel framework that pioneers a data–physics dual-driven paradigm. LUMOS enables end-to-end inverse generation from target optical and physicochemical properties to molecular structures by coupling a generator with a multi-scale predictor through a shared latent representation. The framework integrates an attribute-guided diffusion model, a multi-objective evolutionary algorithm, and an efficient TD-DFT/molecular dynamics pipeline to establish a hybrid prediction system that balances speed, accuracy, and physical plausibility. Experimental results demonstrate that LUMOS significantly outperforms existing approaches across multiple benchmarks and successfully generates high-performance fluorescent molecules satisfying complex multi-objective constraints.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Search and Optimization: Mixed Discrete/Continuous Search
• Computer Vision: Diffusion Models for Vision

Application Category

• Economics, Online Markets and Human Computation: Uses of LLMs and GenAI for marketplace design, bidding, and strategic interactions
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Designing fluorescent small molecules with tailored optical and physicochemical properties requires navigating vast, underexplored chemical space while satisfying multiple objectives and constraints. Conventional generate-score-screen approaches become impractical under such realistic design specifications, owing to their low search efficiency, unreliable generalizability of machine-learning prediction, and the prohibitive cost of quantum chemical calculation. Here we present LUMOS, a data-and-physics driven framework for inverse design of fluorescent molecules. LUMOS couples generator and predictor within a shared latent representation, enabling direct specification-to-molecule design and efficient exploration. Moreover, LUMOS combines neural networks with a fast time-dependent density functional theory (TD-DFT) calculation workflow to build a suite of complementary predictors spanning different trade-offs in speed, accuracy, and generalizability, enabling reliable property prediction across diverse scenarios. Finally, LUMOS employs a property-guided diffusion model integrated with multi-objective evolutionary algorithms, enabling de novo design and molecular optimization under multiple objectives and constraints. Across comprehensive benchmarks, LUMOS consistently outperforms baseline models in terms of accuracy, generalizability and physical plausibility for fluorescence property prediction, and demonstrates superior performance in multi-objective scaffold- and fragment-level molecular optimization. Further validation using TD-DFT and molecular dynamics (MD) simulations demonstrates that LUMOS can generate valid fluorophores that meet various target specifications. Overall, these results establish LUMOS as a data-physics dual-driven framework for general fluorophore inverse design.

Problem

• fluorescent molecule design
• multi-objective optimization
• chemical space exploration
• inverse molecular design
• property prediction

Innovation

• data-physics dual-driven
• inverse molecular design
• multi-objective optimization
• property-guided diffusion model
• TD-DFT integration