Hi.

I have use the piano package to create list of Gene Ontology groups ordered from top differentially expressed (up regulated groups) to down differentially expressed (down regulated groups) including all the groups between both extremes.

I have used the function consensusHeatmap, so I have a consensus of some different GSA methods.Just write the next paragraph to make clear the output in case of not knowing the output of this function.

Now I have for each Gene Ontology group up to 5 different p-values from this consensus; Distinct directional (up and down), Mixed directional (up and down) and Non-directional. The shows whether the GO group is being expressed in one way, the second shows whether if within a GO group there are groups genes that are being expressed up and other down, and for the last one whether a GO group is being expressed differentially.

What I want to do is to mix four different lists  (each one has all 5 p-values), to see which GO groups are being expressed at the top and bottom of all four lists, and also which GO groups are not being showing up at the extremes. I'm probably also going to be interested in checking the genes that are being expressed up and down in the top and bottom GO groups of the list in the case of Mixed directional groups.

I have checked the geneSelector package, I've read about the GetStabilityOverlap function, but I'm not sure if this gives me all I want.

So ¿Any hint about this?

Thanks for your help.

Hi,

You could also take a look at the consensus function from the R package relations. This is what piano is using to establish which gene sets are ranked high by most GSA methods. Maybe you could do something similar, but with your four different lists?

Probably you already know, but since you mentioned checking the genes of different GO groups, I just wanted to mention that you can use the geneSetSummary function in piano for this. (Note also that gene members of a gene set are required to be present in your gene-level data, so typically you will get a smaller number of genes for a given GO-term than if you check the genes from the GO database.)

Good luck with your analysis!