Loss of strandedness in the Illumina Strandedness Total RNA protocol

Question

[Anjan]

Hi, My lab has used the Illumina Stranded Total RNA prep to sequence a set of samples. Refer to this article on how strandedness is maintained by this protocol. Paired-end reads were aligned to the human genome reference using the STAR aligner. Before performing a count of reads using the featureCounts command in Subread, I decided to test the strandedness of my dataset using RSeQC. Surprisingly, RSeQC informed me that the dataset is paired-end, but non-stranded. Have you ever dealt with a dataset from the Illumina Stranded Total RNA prep protocol that has lost its strandedness? If so, would you advise proceeding with feature counts and differential gene expression? Many thanks in advance, Anjan

[ATpoint]

No, I've never seen this before other than people used different kits than anticipated, or things were done wrong in the lab. Use an independent validation as suggested, check the BAM files on the IGV, and then decide whether you simply use as unstranded, or want to discard the experiment due to questionable quality.

[Anjan]

I checked the BAM files on IGV. Notably, for a given gene, the sashimi plot displays junction reads in both orientations! That should be a strong indicator of loss of strandedness, no?

Answer 1

[Gordon Smyth]

Running featureCounts would confirm for certain whether the RNA-seq is stranded or not, and also whether it appears usable or not. See: RNA-seq analysis