A colleague of mine has pre-processed 3' QuantSeq data using the pipeline from Lexogen (https://www.lexogen.com/wp-content/uploads/2018/10/015UG108V0201-QuantSeq-Data-Analysis-Pipeline.pdf), and since we are pretty naive on 3´RNAseq the results look weird to us for two main reasons:

First, using STAR alignment parameters from the Lexogen pipeline, we have around 40-50% of reads that are uniquely mapped and the other 50-60% of reads are mapped to multiple loci. The proportion of reads mapped to multiple loci seem very high. So we are wondering if this is something expected with 3' QuantSeq?

Second, when we look where are the reads distributed using RSeQC, it seems there are only 15-20% of all the reads which are located on a 3'UTR region. This is quite counter intuitive has the 3´RNAseq only sequence the firt 50bp from the 3´ends...We even have an equal proportion of the reads (15-20%) that aligns in the 5´UTR, which technically makes no sense.

Does anyone has experience with this, and could help us out?