From raw data to longitudinal insight: an open-source workflow for untargeted metabolomics of ulvan polysaccharide fermentation

Fermentation systems are useful tools to assess the impact of dietary treatments on the gut microbiota in vitro. When combined with untargeted metabolomics, they can reveal temporal and treatment-related metabolic dynamics but require reproducible data processing and robust statistical approaches. Following best practices in nutrimetabolomics [2], this study presents an integrated workflow for data processing, annotation, and longitudinal analysis applied to [3], was subjected to in vitro digestion (INFOGEST 2.0) and colonic fermentation [1], sampling at 0-12-24-48 h. LC-QToF data were acquired in randomized order, with blanks, pooled QC, and external standards ensuring reproducibility. Data processing in MZmine 3 included smoothing, peak detection, alignment, and gap filling. Annotation with SIRIUS 5.8.6 (CSI:FingerID, ZODIAC, CANOPUS, COSMIC) and data filtering retained features with RSD < 20%, and high variance (top 60%). The final dataset (943 ESI and 485 ESI features) included 173 features with MSI 2a confidence. Dynamic responses were modeled using repeated-measures ASCA (RM-ASCA) in RStudio. PC1 (66-72% variance) described the global fermentation trajectory, showing decreased peptides and bile acids and a rise in fatty acid derivatives, while PC2-PC3 captured treatment-specific effects, distinguishing ulvan from control samples. Two cyclic proline dipeptides, cyclo-(Pro-Pro) and cyclo-(Pro-Met), showed opposite time trends and were tentatively linked to Lactobacillus and Bacteroides via MicrobeMASST, suggesting selective stimulation of beneficial microbial groups. Additional ulvan-specific sulfur-containing metabolites, not previously described, may represent markers. This study, building on a previous targeted metabolomic analysis, integrates open-source tools with advanced longitudinal modeling, providing a workflow for untargeted metabolomics data interpretation of ulvan fermentation by human microbiota.