Data checks

Every file is checked as soon as it is loaded. The report lists each finding with its row number – and also shows which checks passed.

All checks and the rules behind them in the help →

Check	Why it matters
Every row readable	A single decimal comma from Excel makes a row unreadable. Without a warning, that term would silently be missing from every figure.
Overlap = number of genes	The overlap k/n must match the genes listed. A mismatch points to a truncated or edited file.
P-values in range, adj. P ≥ P	No correction can make a P-value smaller. Violations mean the columns were mixed up or edited.
Benjamini–Hochberg recalculated	MES.Plot recomputes the adjusted P-values from the raw ones. A match confirms the file is complete and unaltered.
Excel date artefacts	Excel turns gene symbols such as MARCH1 or SEPT7 into dates (1-Mar, 7-Sep) – a well-documented problem in published supplementary data.
Duplicate terms and genes	Duplicates distort counts, gene ratios and overlaps.
Species of gene symbols	Mouse-style symbols (Fos instead of FOS) in a human library suggest the wrong library was used.
Odds ratios valid	Terms without a valid odds ratio cannot appear in the volcano plot – you are told which ones.
Number of significant terms	Shows how many terms pass your threshold – so that trends are not presented as findings.

Every mark is traceable

Hover over any dot, node, cell, bar, ribbon or tile to see the underlying values: term, overlap, gene ratio, P, adjusted P and the row in your file. Exported SVG files carry the same information – open one in a browser and hover.

Tested against known answers

MES.Plot ships with automated tests. A small file of invented terms, whose correct result is known by hand, checks every plot type – positions, sizes, colours, edges, cells, intersections, ribbons and groups. A second file with deliberately planted errors checks that the data check finds each one on the right row. The Benjamini–Hochberg implementation is compared with reference values from R's p.adjust.

What no software can check for you

MES.Plot checks the table, not the analysis that produced it. Please report these points in your methods:

Background gene set. Enrichr compares against the whole genome by default, not against the genes expressed in your tissue. This tends to make P-values optimistic.
Input list. Cut-offs for adjusted P and log₂ fold change determine which genes enter the analysis.
Direction. Up- and down-regulated genes analysed together can blur the result; separate analyses are often clearer.
Significance. Terms above your threshold are trends, not findings – say so in the legend.