GSEA: 23,273 ranked genes to pathways
Upload the full differential-expression results and Yila verifies the ranking column and direction first, builds the ranked list from all genes, runs GSEA against GO Biological Process and KEGG gene sets, and delivers NES-ranked plots, enrichment curves, and complete result tables — positive and negative enrichment shown separately.
Checks complete: 23,273 genes, no duplicates, no missing values, log2FC direction confirmed (positive = up in Dex). GSEA uses the ranked list of all genes instead of a threshold-filtered subset — that is exactly what sets it apart from over-representation analysis.
Rank by log2FC: it carries both direction and magnitude — positive means upregulated in Dex-treated cells, negative means downregulated. The ranked list is sorted high to low with unique genes; the first and last 10 rows were shown for review.
GSEA complete: NES-ranked plots for KEGG and GO BP are split by direction — red for positive enrichment at the Dex-up end (RIBOSOME, MINERAL ABSORPTION, …), blue for negative enrichment (SPLICEOSOME, UBIQUITIN MEDIATED PROTEOLYSIS, …), with NES labeled per pathway.

The full tables ship alongside: ES, NES, and adjusted P-values for 2,012 GO BP and 303 KEGG pathways, with PNG, editable SVG, and PDF figures all generated and QA-checked.

The curve climbs when the pathway’s genes cluster near the top of the ranked list; the peak is the maximum enrichment score; the black ticks mark where member genes fall in the list; and the genes before the peak (for positive enrichment) are the leading edge — the core drivers of the signal.
The boundaries to hold: an enrichment trend is not causal proof of pathway “activation”; GO and KEGG share the same ranked list, so agreement between them is not independent validation; and significance depends on gene-set versions and multiple-testing conventions — key claims still need literature or experiments.
Yila task replay complete.
Start with this method