GO analysis of RNAseq data error

[JustinH]

I am working on GO analysis in arabidopsis and have done the analysis bit and am working on creating the heat map and scatterplot but am getting the error below.

# Heatmap and scatter plot following GO biological process analysis
> for (gs in rownames(go.bp$less)[1:3]) {
+ outname = gsub(" |:|/", "_", substr(gs, 12, 100))
+ geneData(genes = ath.sets[[gs]], exprs = exp.fc, ref = NULL,
+ samp = NULL, outname = outname, txt = TRUE, heatmap = TRUE,
+ limit = 3, scatterplot = TRUE)}

Error in 1:ncol(exprs) : argument of length 0

It appears to me that the error is in my expression data. I am using the "RNA-seq data Pathway and Gene-set Analysis Workflos as my guide so my expression data comes in from edgeR. If you have ideas please help!

[sarvidsson]

What do you get when you type

dim(exp.fc)

?

[hjv]

I am having the same issue trying to analyse significant GO terms for RNAseq data as per "RNA-seq data Pathway and Gene-set Analysis Workflow".

The "exp.fc" variable generated has NULL dimensions, which doesn't seem to be a problem for gage but it is a problem for geneData. Does anyone have any idea of what's wrong here?

[Aditi Verma]

Could you sort the issue out? I have the same problem with 'exp.fc' with null dimensions working fine with gage and with kegg datasets but getting stuck at only geneData. My exp.fc comes from DESeq2 results.

[JustinH]

I did manage to resolve the issue using the manual from gene data and could graph all genes as a heat map from this
#Graphing all genes
geneData(genes = ath.sets[ath.subs$BP], exprs = cpms3, ref = 1:4,
samp = 5:12, outname = "cpms3", txt = TRUE, heatmap = TRUE,
limit = 3, scatterplot = TRUE)

Notice for the gene sets I had to specify my GO gene set to all genes in biological process for it to work and used the "pseudo.counts" from my DGE list for exprs. To graph the actual pathways, I unfortunately couldn't manage to get it to work from the GAGE list and had to graph each pathway individually which i defined as gsX

gs1 = unique(unlist(ath.sets[ath.subs$BP][rownames(go.bp$greater)[1]]))

Then I could use geneData and set the genes to gsX and set the outnames myself

#GO:0006354 DNA-templated transcription, elongation
geneData(genes = gs1, exprs = cpms3, ref = 1:4,
samp = 5:8, outname = "DNA-templated trascription", txt = TRUE, heatmap = TRUE, limit = 3, scatterplot = TRUE)

Hope this helps you I know it took me a while to get it to work