Troubleshooting

Troubleshooting

Symptom	Cause	Remedy
Every sample is in its own group	The design file was not applied. A common cause is a design saved with classic Mac line endings, which older versions read as one line.	Current versions read all line endings. Check that sample names in the design match the matrix headers exactly.
An extra "sample" named after the ID column	A numeric identifier column (Entrez IDs) was read as data	Current versions detect identifier columns. Place identifiers in the first column with a header such as EntrezID.
Genes appear as numbers or Ensembl IDs	No annotation applied	Choose the species and click Use built-in annotation, or load an annotation file.
A known gene is missing after annotation	The identifier is absent from the annotation release, or was filtered out	Download unmapped IDs; check filtering on the review step.
voom is not offered	The input is not raw counts	voom requires counts. Use the moderated t for normalized input.
Covariate option is absent	The design has one factor only	Add the covariate as a column in the design file.
"Covariate is confounded" message	The covariate separates the groups exactly	The effects cannot be separated. Redesign, or analyze without the covariate and state the confounding.
Few or no DE genes despite visible differences	Low replication; FDR unattainable	See Statistical power.
Enrichment finds no sets	Wrong species library, or identifiers not decoded to symbols	Use the library matching the species; apply annotation first.
Cross-dataset match count is near zero	Different identifier types or species	Set the species of the added dataset; for other species load a Custom ID map.
Session loads with a version warning	The session was saved by another version	Results may differ where defaults changed; see CHANGELOG.md.
A plot looks crowded at poster size	Very long sample or group names	Labels are shortened automatically; shorter names in the design file give cleaner figures.
Self-test reports a failure	A numerical difference in this browser	Download the evidence file and open a GitHub issue with it.