Hello all!
I am having trouble with an experiment done in my lab, and hopefully someone can help...
We have two plants, one is a wild-type, and the other a stable transgenic line with our gene of interest knocked down by RNAi. We performed an RNA-seq experiment to compare these plants, but my issue is that our gene of interest does not appear down-regulated in my RNAi plants, its expression is only slightly lower... We have confirmed the knockdown by RT-qPCR, so I imagine that in the RNA-seq we probably are seeing portions of the RNAi cassette rather than the endogenous mRNA of our gene. Our RNAi construct is made up of the complete CDS of the gene, an intron, and the reverse complementary CDS, producing a hairpin.
I hope someone could give ideas as to how I can solve this problem... It is the only thing keeping us from publishing, since differential expression analysis of interesting genes from the RNA-seq has been confirmed by RT-qPCR.
Is that a strand-specific RNA-seq experiment and is your bioinformatics pipeline strand-specific, what is the pipeline? Is the RNAi polyadenylated?
No, unfortunately not strand-specific. I tried both genome alignment (STAR + Stringtie) and pseudo alignment (Salmon) to quantify expression, then edgeR/limma for differential expression. Theoretically the RNAi is polyadenylated, yes.
Not strand-specific, that is the problem I guess. What you could try is to include the exact sequence of the cassette (so what is really transcriped into RNA) into the reference transcriptome and then quantify again with salmon. Maybe that kind of decoys away some of the reads that can uniquely be mapped to the RNAi and get a bit better differential results. I mean, if you validated it you just need some kind of eye candy to show the reviewer, or you skip that particular gene for DE analysis, mentioning that due to non-strandedness and experimental design you cannot distinguish cDNA and RNAi reads and just leave it be. Show the validation in a main figure, I would be ok with that. Saw that in our own knockdown/out experiments often, so validation was ok, both on qPCR and protein level but RNA-seq was meh.
Thank you for the suggestion! I was thinking of doing that, but was unsure if it was ok... Also thank you for giving me a possible answer for the reviewer, I literally dreamed of them asking me why the RNA-seq didn't show the same as the qPCR, but your explanation makes total sense!