Refine Drugs, Don't Complete Them: Uniform-Source Discrete Flows for Fragment-Based Drug Discovery

2025-09-30

AI Summary

This work addresses key challenges in drug discovery—including de novo small-molecule generation, fragment-constrained design, and multi-objective lead optimization—by proposing InVirtuoGen, the first discrete-flow-based generative model for molecular design. Methodologically, it introduces a uniform source distribution to shift from sequence completion to structural refinement, decoupling sampling steps from sequence-length dependency; adopts fragmented SMILES representation and mask-agnostic, full-position denoising training; and employs proximal policy optimization (PPO) fine-tuning tailored to discrete flows. On standard benchmarks, InVirtuoGen achieves significantly higher top-10 AUC than state-of-the-art methods. In lead optimization tasks, it generates molecules with superior molecular docking scores. The model and implementation are fully open-sourced, ensuring reproducibility.

Technology Category

• Search and Optimization: Mixed Discrete/Continuous Search
• Machine Learning: Deep Generative Models & Autoencoders
• Natural Language Processing: Generation

Application Category

• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

We introduce InVirtuoGen, a discrete flow generative model for fragmented SMILES for de novo and fragment-constrained generation, and target-property/lead optimization of small molecules. The model learns to transform a uniform source over all possible tokens into the data distribution. Unlike masked models, its training loss accounts for predictions on all sequence positions at every denoising step, shifting the generation paradigm from completion to refinement, and decoupling the number of sampling steps from the sequence length. For extit{de novo} generation, InVirtuoGen achieves a stronger quality-diversity pareto frontier than prior fragment-based models and competitive performance on fragment-constrained tasks. For property and lead optimization, we propose a hybrid scheme that combines a genetic algorithm with a Proximal Property Optimization fine-tuning strategy adapted to discrete flows. Our approach sets a new state-of-the-art on the Practical Molecular Optimization benchmark, measured by top-10 AUC across tasks, and yields higher docking scores in lead optimization than previous baselines. InVirtuoGen thus establishes a versatile generative foundation for drug discovery, from early hit finding to multi-objective lead optimization. We further contribute to open science by releasing pretrained checkpoints and code, making our results fully reproduciblefootnote{https://github.com/invirtuolabs/InVirtuoGen_results}.

Problem

• Generating novel molecules for drug discovery using fragment-based approaches
• Optimizing molecular properties and lead compounds through hybrid algorithms
• Decoupling sampling steps from sequence length in molecular generation

Innovation

• Discrete flow generative model for fragmented SMILES
• Uniform source transformation into data distribution
• Hybrid genetic algorithm with property optimization