Good morning to everyone, I'm trying to use the function goana to run a GO enrichment analysis after performing the workflow RnaSeqGeneEdgeRQL. I obtained a list of DE genes with the function glmQLFTest (on the left), and shortly after I decided to use glmTreat as well (on the right), to obtain a more limited dataset. I picked a threshold of logFC > 1.5.
Here is the result
I performed a further GO enrichment analysis on the DE genes.
I used the result of the function glmTreat at first, but what I obtained is an error:
data frame with 0 columns and 0 rows No DE genes
(is the dataset too small to perform a statistically significant GO enrichment analysis?). This just works fine while I try to compare other conditions from which a higher number of genes is obtained.
I also tried to make the function glmTreat work like glmQLFTest, like suggested in the pipeline. I set
tr <- glmTreat(fit, contrast=B.LvsP, lfc=0)
And I obtain the same number of DE genes I did for glmQLFTest. Everything works fine, till I try to use goana again on this specific object, where I obtain a list of GO terms showing p values of 1 in the columns. I guess the test for some reason doesn't work...
How can I solve this problem and perform a GO enrichment analysis? Did I do anything wrong with the two functions? Thanks in advance!
Could it be that the problem is that you're dealing with an unannotated organism as you seem to allude to in your other post? If so the question should probably be 'how to get edgeR::goana to work with an organism not represented in whatever database edgeR uses?'
Hello, thanks for your answer! Sadly, it doesn't seem to be the case. In the past, I used the function goana (with a mild change in the code) to acquire an external given database, not online ones, and it worked perfectly... The problem seems to be something different since, as I explained in the topic I'm unable to use an object generated from glmTreat (set with lfc=0) as input for the function goana and I keep getting those pvalue = 1. The reason of doing this is that, apparently, goana only accepts tr objects generated from glmTreat, but I'm not interested in the typical glmTreat outputs (too tiny datasets, goana won't find any difference between DE genes). So, I tried to use, as suggested in the workflow itself, glmQLFTest as equivalent of it, by setting glmTreat lfc=0. I can't figure out why this didn't work.
Cross-posted on Bioconductor ( https://support.bioconductor.org/p/129909/ ; https://support.bioconductor.org/p/129910/ )