Fig. 1

Overview of UmetaFlow. The user can clone UmetaFlow (Snakemake or Jupyter notebook version) from github and follow the step-by-step guide to set it up. a The pre-processing step is a set of algorithms that transforms the raw data to a table of metabolic features. One of the most important algorithms of this step is the one for feature detection, that detects mass traces, deconvolutes them and assembles single isotopic mass traces to metabolite features. Map alignment corrects for RT shifts and feature linking connects corresponding features across individual runs. b Right after, an optional step for re-quantification of features with missing values can be selected. c The generated feature files (re-quantified or not), together with the mzML files, are used as inputs to the SIRIUS executable for formula and structural predictions. d The clustered feature files and mzML files are introduced to the GNPSexport algorithm to generate all the files necessary for FBMN/IIMN. e, f The final output of UmetaFlow is a feature matrix and a GraphML network file with MS² library matches, and formula and structural prediction annotations