I have some transcriptomic (Whole) sequencing data that I should analyse. I would like to do raw data alignment to a reference genome taking into account the multi mapped reads and uniquely mapped reads. I would like to know the consequences on my transcriptomic analysis results : 1. if I should only focus on uniqueMapping reads (mapped to one loci) and neglect MultiMapping read (mapped to several loci or variants)! 2. or take them all together

So my questions are :

1- Considering unique and multi-aligned reads, wouldn't this be a source of bias in my results interpretation?

2- By hiding the multi-mapped reads, could I lose information on the expression of important genes?

thank you in advance


There are several methods/tools that take into account multi-mappers and deal with them in reasonable ways (e.g. pseudo-alignment methods like salmon and more traditional alignment-based methods like RSEM). If you completely ignore multi-mappers, then yes, you will be losing information, which may or may not be valuable to you.

Both of those methods utilize the expectation-maximization algorithm (or a variant thereof) to determine the relative contributions of multi-mapping reads to the various mapping locations. With salmon, you can forego alignment completely and end up with very accurate gene quantifications in hand in like 20 minutes if you want.

  • $\begingroup$ I followed your advice and I didn't have a big difference between the results of considering single mapped and multimapped reads. Thanks you Jared Andrews $\endgroup$
    – Diango
    May 27 at 13:50

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.