Have you ever seen perfect (100%) mapping rate (RNASeQC) on human RNASeq (hg38) using STAR?

Question

[mark.ebbert]

Hi,

I've been out of the RNA-Seq world for a while, so I'm trying to get up to speed on the latest expectations, protocols, QC, etc.

We've run four test samples (human blood) for our specific protocol (hemoglobin AND rRNA depletion, obviously no poly-A selection). I used STAR to align to the hg38 genome (annotated using gencode v25). I then ran FastQC, RNASeQC, and RSeQC (via QuaCRS) and am digging through the various QC metrics. I was immediately surprised to find the mapping rate is 1 (100%) for all four samples, according to both RSeqC and RNASeQC. The "Mapped Unique Rate" ranges from 70%-80% (according to RNASeQC).

It seems very strange that every read maps to the human genome, even from blood, but this is the first time I've run a blood RNASeq. Based on various google searches, I've only found posts concerned about low mapping rates.

I really appreciate any thoughts!

[h.mon]

Could you tell us which STAR version and your exact command-line?

Answer 1

[h.mon]

Search the STAR manual for --outSAMunmapped, and see an explanation here.

edit: check the final.log files, it should state the number of unmapped reads.

[mark.ebbert]

uggh, you're right. I was looking for much more complicated answers than that.