I have Illumina paired-end RNA-Seq data (prepared with the TruSeq stranded kit) for human tissue biopsies.
After QC and alignment to the ENSEMBL genome and gtf (GRCh38 rel 84 from ensembl.org) using STAR (alignment perc. of 75% - 90%), I use featureCounts (in R) to count genes.
I get very different rates of successfully assigned fragments, ranging from 20% to about 60%, with the majority about 45%. Is this normal? Or could something be wrong with (some of) my libraries? Do I need to account for the different assignment proportions?