RNASeq QC and STAR results confused

Question

[shafiharafi]

I have RNA seq samples approximately 80 million coverage (Total RNA). I ran FastQC on them and I saw they have too many duplicated reads and when I used STAR for alignment on hg38 their uniquely mapped reads are about ~28%. The overall alignment score is ~98% with reads mapped to multiple loci being ~79%. I am planning to do aberrant splicing analysis on the downstream and I am worried if the low uniquely mapped reads count would affect the expression analysis, is this alignment good or bad. Is the sample quality bad?

[1769mkc]

please show the plots or data table, may be you can share your multiqc report.

[lieven.sterck]

My bet would be that, since it is total RNA sample, it might not have been rRNA depleted. That will certainly result in the numbers you mention.

So, check if this is the case. If so it's not a huge problem. The sample is in essence not bad just cluttered with lots of reads that are not valuable for your analysis goals.

Answer 1

[Istvan Albert]

I want to note that the duplication rate you see in QC is not directly related to the unique mapping rates.

Duplication rates measure sequence identity. Unique mapping relates to how many locations sequences align.

It is possible that non depleted rRNA leads to both, but just not necessarily so.

As lieven.sterck mentions, investigate where the reads align, you can do that by eye as well in IGV.