-
Notifications
You must be signed in to change notification settings - Fork 0
FAQ
Is my data uploaded? No. All computation happens in the browser. See Privacy and architecture.
Are the results the same as R? For the implemented methods, yes, to the precision stated in VALIDATION.md. The self-test lets you confirm this in your own browser.
Why is DESeq2 not included? DESeq2 is an iterative estimation procedure (negative-binomial GLMs, dispersion shrinkage, outlier handling). A browser version that did not reproduce Bioconductor's results exactly would be misleading. The application exports a DESeq2 script for the current contrast instead.
Should I use voom or the moderated t? For raw counts, voom. The moderated t on log₂-CPM is appropriate for normalized input and gives similar results when library sizes are similar.
Why do ORA, GSEA and FRY give different results? They test different hypotheses. See Gene-set testing.
Can I analyze a paired design? Yes. Add the subject identifier to the design file and select it as a covariate.
Can I compare mouse and human data? Yes. See Cross-dataset comparison and Orthologs and homology families.
Does it work offline?
The offline build (benchside-offline.html) needs no network connection.
How do I cite it? See Citing.
Start
Data preparation
Statistics
- Differential expression
- Covariates and pairing
- Multiple contrasts
- Statistical power
- Gene-set testing
- Discovery screen
- Patterns and clustering
- Heatmaps
- Co-expression
Comparing datasets
Output
Background