Total RNAseq multiple mapping issue

[wri_ichp]

We are running paired-end total RNA (using blood) on HiSeq and after adapter and quality trimming (Q>30), we use tophat2 for alignment. We are getting about 90% overall mapping, but around 80% of those are mapping to multiple locations.

I am pretty sure this is way too high, but don't know what the issue is. Are there any recommendations for how we can reduce the multi-mapping or ideas on what might have caused it. Thanks!

[Brian Bushnell]

Mapping to a transcriptome will have a high rate of multiple mapping because the same references occur multiple times. If you want unique mapping, map to the genome instead, using a splice-aware aligner.

Also, trimming to Q30 will reduce the length of your reads and bias your data with no real advantage. Q10-12 will give you much better results. Short reads will multi-map more because they contain less information.

[Michael.Ante]

I would rather guess, that you have rRNA contamination in your sample, or highly expressed transcripts originating from a cluster of similar genes. Check your reads with FastQC and if detected, blast your over-represented sequences. You can also use the repeat-masker to analyse your alignment towards rRNA and other elements.