Unexpected quality controls for RNA-seq
Updated: Feb 25, 2019
It is amazing how many things can go wrong during sequencing. In my experience, samples mix-ups and contaminations are some of the most common. And then of course there are those mysterious ones, for which no one knows what went wrong. Here are few RNA-seq control measures you might not have thought of, but which has helped us in the past to identify sources of faulty results:
Antisense transcription. Sure you check the rate of antisense transcription, but did you know that the gene expression may be globally influenced even when the differences in antisense transcription rate are at about 1%? Make sure you antisense counts are comparable across samples, especially when antisense transcripts appear as a separate module in your co-expression networks.
SNPs call on RNA-seq. I bet you think, its unreliable, and probably better avoided. But especially when you work with RNA-seq from related individuals, checking for mendelian errors can bring you a piece of mind that the samples are indeed assigned to correct families.
These are probably the two measures that most surprised me by how accurately they detected any inconsistencies. Good luck!
