AI Insight
Researchers developed FRIGID, an AI system that uses diffusion language models to identify unknown molecular structures from mass spectrometry data. The system trains on hundreds of millions of unlabeled molecular structures and uses an innovative approach that generates molecular fingerprints as intermediate representations, then refines predictions by identifying and correcting spectrum-inconsistent fragments. FRIGID achieved over 18% Top-1 accuracy on the MassSpecGym benchmark and tripled the accuracy of leading methods on NPLIB1, with performance improving predictably as more computational resources are applied during inference.
Why it matters
This advancement could significantly accelerate drug discovery, natural product identification, and other scientific workflows that depend on identifying unknown chemical compounds. The system's ability to scale performance with additional compute time offers a clear path for continued improvement in automated molecular structure determination.
⚠️ Preprint – Noch nicht peer-reviewed
Dieser Artikel wurde noch nicht von unabhängigen Experten begutachtet. Die Ergebnisse sind vorläufig und sollten mit Vorsicht interpretiert werden.
-cross
Abstract: Tandem mass spectrometry is prominent in scientific discovery workflows for identifying unknown small molecules, yet high-throughput structural elucidation remains challenging. While recent autoregressive and graph diffusion models have shown promise in de novo elucidation, performance remains limited by poor scalability during both training and inference time. In this work, we present FRIGID, a framework with a novel diffusion language model that generates molecular structures conditioned on mass spectra via intermediate fingerprint representations and determined chemical formulae, training at the scale of hundreds of millions of unlabeled structures. We then demonstrate how forward fragmentation models enable inference-time scaling by identifying spectrum-inconsistent fragments and refining them through targeted remasking and denoising. While FRIGID already achieves strong performance with its diffusion base, inference-time scaling significantly improves its accuracy, surpassing 18% Top-1 accuracy on the challenging MassSpecGym benchmark and tripling the Top-1 accuracy of the leading methods on NPLIB1. Further empirical analyses show that FRIGID exhibits log-linear performance scaling with increasing inference-time compute, opening a promising new direction for continued improvements in de novo structural elucidation. FRIGID code is publicly available at https://github.com/coleygroup/FRIGID.