Pathway Enrichment

Gene Set Enrichment Analysis (GSEA) helps you to identify enriched reactions and pathways in your metabolic model using a ranked list file that you upload.

Analysis is done by the gseapy Python package (via the gp.prerank function).

Input

You need to provide a ranked list file containing reaction IDs and associated scores.

Format (comma-separated, .csv):

Reaction      Rank
PMANM         5.881484159
SELCYSLY      5.667135956
SELCYSLY2     5.667135956
FBP           5.548850782
FBP26         5.548850782
PI4PLC        5.376448076
PI45PLC       5.376448076
PIPLC         5.376448076

• Reaction → identifier of a metabolic reaction in the model
• Rank → score indicating importance (e.g., fold change, z-score, differential expression statistic)

---

Parameters

In the modal, you can configure the following analysis options:

Parameter	Description	Default	Notes
Permutations	Number of random permutations to compute statistical significance	1000	Higher values = more stable p-values, but slower
Min Size	Minimum pathway size (number of reactions) to be tested	15	Avoids very small pathways
Max Size	Maximum pathway size (number of reactions) to be tested	500	Avoids overly broad pathways
Metrics	Statistical measures reported	—	FDR q-val = False Discovery Rate FWER p-val = Family-wise Error Rate
p-value cutoff	Significance threshold to filter enriched pathways	0.05 or 0.01	Applied to FDR/FWER values

---

Results

After running GSEA, a ranked table of enriched pathways is displayed:

Each row corresponds to a significantly enriched pathway.

• Clicking on a row will:
  1. Open a reaction–reaction network displaying the subset of reactions responsible for the enrichment.
  2. Highlight these reactions as green nodes in the main canvas, linking enrichment back to the global network view.

Column	Description
pathway	Name of the pathway (from your gene_sets database, e.g. KEGG, MetaCyc, or custom).
es	Enrichment Score → raw measure of how strongly the pathway’s reactions are skewed toward the top or bottom of the ranked list.
nes	Normalized Enrichment Score → ES adjusted for pathway size and permutation background, allowing fair comparison across pathways.
fdr	False Discovery Rate (q-value) → estimated probability that a pathway identified as enriched is a false positive.
fwer	Family-Wise Error Rate p-value → conservative correction accounting for multiple hypothesis testing.
nom	Nominal p-value → unadjusted significance of enrichment for that pathway.
gene %	Proportion of reactions in the pathway that appear in the ranked list.
tag %	Proportion of ranked list contributing to the enrichment score (leading-edge reactions).
lead_genes	The genes / reactions that caused the pathway to be enriched

Important

Only pathways meeting the p-value cutoff are shown in the table.