We and others have noticed two subtle problems with some ORA tools. The first problem we call the “Background error”, and involves genes without annotated categories being discarded erroneously. The second problem we call the “FDR error” because some tools incorrectly apply false discovery rate correction of p-values. For more information, refer to our README. This tool is released under an MIT Licence and comes without any warranty.

How to use the tool

To get started, you will need to upload two .txt files containing human gene symbols; one for the foreground genes and one for the background. Currently, only human genes are supported. The foreground list contains genes that were identified by some ’omics analysis as being interesting. The background list contains all the genes that were detected robustly in the omics assay. Here “robustly” means that the detection was good enough that it could have been part of the foreground list. Example foreground and background files are provided.

After uploading the lists, select the gene set library you want to test. We have sourced some popular options from MSigDB v2023.2. Then select the error you would like to characterise; the options are “Background error”, “FDR error” and “Both errors”.

Then click on the “Data” tab which will summarise the gene lists that have been uploaded. It is a good idea to check these values to ensure that the list was uploaded in its entirety. Then click on the “Comparative Analysis” tab which will kick off the enrichment analysis. This part could take up to 30 seconds to complete, so be patient and wait for the spinning circle in the top-right corner to stop. In this time, avoid clicking away or reloading the page. The table will eventually appear. By default, the results are sorted to show sets that have divergent FDR values at the top. Then click on the “Charts” tab to view some plots.

The “Download Report” button can then be used to generate a full report for provenance. Again it can take 30-60 seconds to generate so be patient and don’t click away until the file has been downloaded.

Foreground

              Background
              

              No. genes in foregorund
              

              No. genes in background
              

            
Comparison of original (x) and fixed (y) analysis - top 20 pathways with divergent FDR values.