filtering GSEA results by NES assumes positives are in one file and negatives are in another

[risserlin]

Using FGSEA to analyze data and create "fake" GSEA results files but putting all results into one file (with both positive and negative in the same file to reduce amount of files that I create).
Run EM with "NES (GSEA) only option --> positive" and it shows both positive and negative results.

If you then divide the file, one positive with +ve NES values and one negative with all the -ve NES values, then it works.

Problem - if you put the negative file in the positive spot in the input panel and the positive file in the negative spot in the input panel EM creates the wrong network and assumes the negative file are actually positive!

[mikekucera]

What is the format of the header row of your files? It looks like EM is assuming that GSEA results have 11 columns.

[risserlin]

here is the header row for fGSEA results that I have created -
name description GS details SIZE ES NES pval padj FWER Rank at Max leading edge genes
(truthfully I am not sure which file the above is referring to but I have a generic method for creating fGSEA results and they all look like this)

[mikekucera]

Hi Ruth,
What is the actual problem? I just tried creating a network from a GSEA data set and I swapped the positive and negative files and it appears to still have worked.