Annotate ouput file from Deseq2

Question

[BM]

I am trying to annotate the results output file from Desq2 so it contains gene names and symbols. The RNA-seq count file I have used comes from Dexseq and contains ensembl transcript ID:

ENSMUSG00000000001:001
ENSMUSG00000000001:002
ENSMUSG00000000001:003

etc.

I have tried various methods to annotate the results.

1. downloaded annotation from Biomart.

> library(DESeq2)

counts = read.delim("3mTA2.txt", header=T, row.names=1)
sample <- read.delim("~/sample.txt")
count.data.set <- DESeqDataSetFromMatrix(countData=counts, colData=sample,design= ~ genotype)
dds<-DESeq(count.data.set)
res <- results(dds)
annotation <- read.delim("mouse.annt.txt") # load annotation file from Biomart
res$EnsemblID <- row.names(res)
res <- merge(res, annotation, by = 'EnsemblID', all.x = TRUE)

It adds column to the output file but values are blank.

2. Also used AnnotationDbi

library("AnnotationDbi")
library("org.Mmu.eg.db")
res$symbol <- mapIds(org.Mmu.eg.db,
+                      keys=row.names(res),
+                      column="SYMBOL",
+                      keytype="ENSEMBL",
+                      multiVals="first")
Error in .testForValidKeys(x, keys, keytype) :
None of the keys entered are valid keys for 'ENSEMBL'. Please use the keys method to see a listing of valid arguments.

Any suggestions?

Answer 1

[James Ashmore]

Those IDs you have listed are Ensembl gene IDs, not transcript IDs. I'm also not sure why they have the ':001' string after them? If you try the BioMart id conversion tool you can see that if you remove this last part and convert the ID to a gene name you get a result e.g. ENSMUSG00000000001 = GNAI3. This Ensembl tutorial may help you discern between the different IDs.

[BM]

ENSMUSG00000000001:001; ENSMUSG00000000001:002 - these refer to the the different exons of the gene.

So the question I suppose is how to combine or merge the different exon counts for the same gene into one count for the gene?

Can this be done in Dexseq or Deseq2?

[Devon Ryan]

You don't want to do that, since doing so will double count a number of things. Just run either htseq-count or featureCounts (this is much faster) and directly get gene level metrics.

[BM]

The initial analysis was performed elsewhere. So I only have the Dexseq count file with ensemble ids of all the different exons of a gene. How can i use this file to proceed - either by annotating exons ids into a gene or using the file in Deseq2 and then annotate ?

[Devon Ryan]

That's unfortunate, particularly if you don't have the BAM or fastq files. Indeed, the best you can do is just remove the :E??? from the names, sum over the results and use that. Note that the results will then be approximate. You could do that with awk.