EdgeR Error, 0 gene counts for 1 sample

Question

[nikelle.petrillo]

Hi everyone,

I have performed de novo transcriptome assembly through Trinity. I am now trying to perform DE in EdgeR. However, I have one sample that has 0 reads for every gene. Does anyone know why this would be? This seems to be tripping up EdgeR.

Any help would be much appreciated! Thank you, Nikelle

[Devon Ryan]

How are you generating counts for the samples? When you map the reads from that sample back to the assembly, what sort of alignment rate do you get?

[nikelle.petrillo]

Hi Devon,

I used bowtie in Trinity's pipeline to align. Actually, looking back, there was a 0% alignment success rate for this one particular sample while all the other samples averaged a 77% alignment success rate. Do you think this could be causing the problem in EdgeR? Seems like it must be...

I re ran the alignment for that one sample however I am still getting 0% alignment success rate. Any idea why this is happening to only one of my samples?

Thank you for the help, Nikelle

[Devon Ryan]

Yes, the 0% alignment rate is the cause of the problem. Have you tried adapter trimming and local alignment? I don't know if rsem can handle locally aligned reads, but at least that'd allow you to see if there's some sort of adapter or other oddness attached to the reads that's preventing alignment.

[nikelle.petrillo]

I'm sorry but I am pretty new to RNA seq, what do you mean by local alignment?

I trimmed my reads according to a quality score of 30. Do you know how I would go about finding the adapter sequences?

[Devon Ryan]

Local vs. global (the default for bowtie/bowtie2) is described here. For bowtie2, you can just add the --local option. For finding adapter sequences you could use FastQC, though honestly I'd recommend just retrimming the problematic sample with "Trim Galore!", which is highly likely to default to trimming off the adapter sequences you have.