1 Summary of Inputs

featureMSEA requires three inputs: a feature table from your LC-MS experiment, an MS1 metabolite database for annotation, and a metabolite set database for enrichment testing.

1.1 Feature Table

The feature table is the primary input and can be provided as either an .rda file (containing a data frame) or a .csv file. Each row represents one metabolic feature detected in your LC-MS experiment. The following columns are required:

Column	Type	Description
variable_id	character	Unique identifier for each metabolic feature
mz	numeric	Measured mass-to-charge ratio (m/z)
rt	numeric	Retention time in seconds
condition	numeric	Phenotype-associated ranking statistic — use the absolute value of an effect-size measure such as signal-to-noise ratio (|SNR|), fold-change (|log₂FC|), or correlation coefficient (|r|)
polarity	character	Ion mode: "positive" or "negative"
mean_intensity	numeric	Average feature intensity across all samples

Note on condition: featureMSEA ranks features in descending order by the condition value before running enrichment, such that features with larger values are ranked higher and contribute more strongly to the enrichment score. The condition column should contain the absolute value of a phenotype-associated statistic, such as |SNR|, |log₂FC|, or |correlation coefficient|. Because magnitude-based ranking is used, featureMSEA identifies metabolite sets that are collectively dysregulated, without distinguishing between directionally up- or down-regulated signals.

1.2 Reference Databases

Two reference databases are required.

MS1 metabolite database — used for accurate-mass-based feature annotation (m/z matching). Supported options:

• KEGG
• HMDB

Metabolite set database — defines the sets of metabolites tested for enrichment. Supported options:

Database	Description
KEGG	Metabolic pathways from the Kyoto Encyclopedia of Genes and Genomes
PathBank	Small molecule pathways with detailed biochemical context
WikiPathways	Community-curated biological pathways
Reactome	Manually curated human biological reactions and pathways
iMetPD	Integrated metabolite pathway database compiled from multiple sources

1.3 Uploading Data

In the Data Upload panel, upload your feature table (.rda or .csv). The MS1 metabolite database and metabolite set database are not uploaded — select them from the dropdown options provided on the website.

Figure 1.1: Data upload panel: upload the feature table and select the reference databases

If you have already completed a previous featureMSEA run, you can upload the saved fmsea_result.rda directly. This skips annotation and enrichment entirely and loads the prior results straight into the visualization panel.

Figure 1.2: Uploading a previously saved result to resume from the visualization step