TidyMass2: Advancing LC-MS Untargeted Metabolomics Through Metabolite Origin Inference and Metabolic Feature-based Functional Module Analysis

Untargeted metabolomics faces critical challenges in determining metabolite origins and interpreting unannotated features. We present TidyMass2, an enhanced LC-MS computational framework. It introduces: (1) metabolite origin inference tracing compounds to human, microbial, and environmental sources via 11 integrated databases; (2) metabolic feature-based functional module analysis (FFMA) that bypasses annotation bottlenecks using network topology; and (3) an intuitive graphical interface. Applied to human pregnancy data, TidyMass2 increased the proportion of biologically interpretable features from 5.8% to 58.8%, uncovering coordinated metabolic changes across diverse biological modules.