PNNL

Abstract

Artificial intelligence (AI) methods, including conventional machine learning and deep learning, are greatly improving automation and performance in virtually every area of modern society, from applications and services to scientific research. In mass spectrometry (MS), AI has the potential to enhance the interpretation of complex biological and environmental data which incorporate separations such as chromatography and ion mobility spectrometry (IMS). In particular, the IMS gas-phase separation technique is being increasingly used to distinguish structurally similar molecules and isomeric species in different omics studies (e.g., proteomics, metabolomics, lipidomics, glycomics, etc.), where subtle structural differences can correspond to significant biological variation. The integration of AI with analytical techniques represents a transformative approach that addresses the growing complexity of omics data analysis and the need for automated, high-throughput solutions in modern analytical measurements. The aim of this special issue is to highlight new developments in IMS-MS omics research that leverage AI methods at any point in the experimental or computational workflow. The contributions demonstrate AI-based advancements in various applications related to IMS-MS. Kartowikromo et al. present a review of collision cross section (CCS) measurement and prediction methods, comparing experimental and theoretical approaches and discussing their limitations and future potential1. Kumari and Causon introduce CCSfind, a tool that enables chemically informed database building from LC-IM-MS data, helping to unify IMS and MS measurements into SQL-accessible entries for faster data retrieval and interpretation2. Further contributions include a study by Stepanovic et al., which applies data science to understand modifier effects in differential mobility spectrometry–mass spectrometry (DMS-MS), offering valuable insights into signal suppression mechanisms in metabolomics and peptidomics3. Ross et al. explore deep learning-guided analysis in LipidOz4 to characterize fungal lipid unsaturation via ozone-induced dissociation, showcasing the utility of automated workflows and the retraining of AI models to improve accuracy for non-mammalian lipidomes5. Collectively, the research presented in this special issue exemplifies the expanding role of AI in enabling deeper insights and broader applications of IMS-MS in omics research. Although the scope of this issue is intentionally focused, the selected contributions clearly demonstrate that experts in the ion mobility community are actively integrating AI into their analytical workflows. These studies mark an exciting step forward, stimulating further innovation, and fostering interdisciplinary collaborations at the evolving intersection of AI and MS, even more broadly and towards AI-powered MS and multiomics integration6.