Low mapping rate of TSS-data (5'-seq)

[JonB]

Hi,

We have performed transcription start site sequencing (TSS, 5'-sequencing) using TAP treatment (tap+ and tap- of totalRNA) and Illumina 100bp SE sequencing. I have quality trimmed the reads with Trimmomatic and mapped to the genome using TopHat2. For the tap+ reads, about 50% of the trimmed reads map to the genome, and for tap- only about 20% map.

Have anyone experienced similar results? What could be the reasons for the low overall mapping rate, and why is the tap- data lower than tap+? I don't fully understand what is the effects of the tap-treatments, and what should be the expected differences between the two libraries. We are working on a non-model organism with a quite poorly assembled genome, but with regular mRNA-seq we usually have 80% mapping rate.

[Michael.Ante]

Hi JonB,
are you using any other treatments to the total RNA? If not, maybe the rRNA is messing up your samples. Try to map the trimmed reads first to the rRNA sequences and continue with the unmapped reads with your TopHat2 alignment.
Additionally, check if the tap+ and tap- mappings have major differences in junction detection. TopHat2 can be picky if it doesn't find any junctions.
Cheers,
Michael

[JonB]

Thanks!
No we didn't deplete rRNAs, so definitely rRNAs represent a major fraction of the reads. But do you think this can explain the differences in mapping rate between tap+ and tap-? And should one expect differences between the two samples?

Thanks for the advice about the junctions, I will look into that.

[Michael.Ante]

AFAIK, tap treatment removes the cap from the RNA. Thus, all mRNA molecules in your tap+ sample are equally available to 5' ligation and other methods.
Therefore, it should make a difference using tap in 5'-Seq assays.