DESEQ2 · EDGER · LIMMA · SEURAT
Find the
story.
Drop in a differential expression results file. Foldline writes a plain-English summary of what changed, checks whether the results can be trusted, finds the gene families and pathways involved, and lets you compare two experiments side by side.
Made for scientists who need to understand the output — not just store it.
and a p-value or adjusted p-value.
Demo: a simulated time course — mouse macrophages 2 hours and 24 hours after LPS (a bacterial signal), each compared with untreated cells.
01 · Summary
A written readout
How many genes moved, the biggest and most confident changes, and what they have in common — in sentences.
02 · Quality
Can you trust it?
P-value distribution, up/down balance, low-count noise and normalization checks, each with a verdict.
03 · Biology
Families & pathways
Gene-family enrichment computed instantly in your browser, plus GO, KEGG and Reactome through g:Profiler.
04 · Compare
Two experiments
Load a second contrast to see what is shared, what is unique and what flips direction.
Which columns are which?
Foldline couldn't recognize every column automatically. Pick them below — you need a gene column, a fold change, and at least one p-value column.
The readout
Insights
Automatic checks on your results. Click a card's gene names to inspect them.
Notes about your file
Volcano plot
Each dot is a gene. Right means higher in treatment, left means lower; higher up means more confident. Genes in the upper corners changed a lot and reliably.
MA plot
Fold change against how strongly each gene is expressed overall. A healthy result is centered on zero at every expression level; a wide funnel at the left means low-expression genes are noisy.
Your file has no mean-expression column (baseMean, AveExpr or logCPM), so an MA plot can't be drawn.
Quality checks
These don't prove an experiment is right, but they catch the most common reasons results mislead.
Raw p-value distribution
Ideally flat, with a spike at the left from real changes. A hump in the middle or a pile-up at 1 points to a modelling or low-count problem.
Fold-change distribution
Should be centered on zero. A shifted center can mean a normalization problem rather than biology.
How much do the cutoffs matter?
Significant genes at each combination of cutoffs. If the count collapses between neighbouring cells, your conclusions depend heavily on an arbitrary line. Click a cell to apply it.
Gene families
Which named gene families are over-represented among your up- and down-regulated genes, compared with every gene you tested. Calculated instantly in your browser using standard gene nomenclature (for example, every Cxcl and Ccl gene is a chemokine).
Pathways · GO, KEGG, Reactome
Runs a full enrichment analysis through g:Profiler (University of Tartu), using the genes you tested as the background. Only gene names are sent.
Results appear here.
Compare with a second experiment
Load another results file — a different time point, dose, cell type or replicate study — to see which changes are shared, which are unique, and which reverse.
How they compare
Overlap
Significant genes in each experiment, at your current cutoffs.
Agreement
Fold change, A against B
Genes on the diagonal changed the same amount in both. Top-left and bottom-right are genes that flipped direction.
Watchlist
Paste genes you care about — markers, your favorite pathway, a reviewer's question. They are ringed in lime on every plot and summarized here. Saved in this browser.
| FAMILY |
|---|
IMPACT = |log2 FC| × −log10 adjusted p. It ranks genes that are both large and confident.