Anyone experiencing the anomaly of having a stranded library prep for RNAs-seq look as though it is unstranded as greater concentrations of virus are introduced into cells?

Question

[tbInformatics]

I have a set of cell lines and I am interested in looking at bulk RNAseq differential expression as more virus is exposed to the cell lines. There are 3 groups: groupA = control with almost no virus, groupB = infected with a mid-range concentration of virus, and groupC = infected with high concentration of virus.

All samples and groups were run using mRNA poly-A selections and a stranded Illumina library prep was used to generate the final library pool. All were run on the same flow cell in the same lane.

After alignment, I used RSeQC infer_experiment.py to do a sanity check of the library strandedness since some downstream gene counting softwares need this information. The groups all look strange in terms of strandedness:

• group A looks to be nearly full 2nd strand synthesis (~94% 2nd to ~6% 1st; this is also true of all the biological replicates in this group), which is expected
• group B is mostly 2nd strand synthesis but definitely has a lot more than expected of 1st strand synthesis coming from a stranded library prep (~70% 2nd to ~30% 1st; this is also true of al the biological replicates in this group)
• group C is pretty much a 50-50 split of 1st:2nd strand synthesis, indicating it is unstranded (also true of all biological replicates in this group)

The host cell line is from mice.

Even if I only align the genome to the mouse genome and ignore the viral genome, I see the results above. If I treat the viral genome as a chromosome and align it all together with the mouse genome, the results do not change from above.

Is there anyone who has experienced this with host-viral genome interaction in RNA-seq or any suggestions (open to technical and/or biological hypotheses) as to what may be going on?

Thanks!

[Asaf]

Make sure you don't see a batch effect. Did the same person prepared the libraries? Were they all prepared together and sequenced together? You can definitely add the reverse of the genes as entities and test them for differential expression

[tbInformatics]

Yes, all the libraries were prepared by the same person and sequenced together as one pool all on the same flowcell and same lane.