I used RNA STAR to map my reads for a stranded-RNAseq library. Within RNA STAR, i turned on the option to include gene counts and so I included an hg38 GTF file from UCSC table browser. RNA STAR said that of my 20 million total reads, 86% uniquely map (so 17 million reads uniquely mapping), however, when I take the sum of all the gene counts that RNA STAR outputs, it only adds up to 5 million reads. My question is how come the total sum of all the gene counts doesn't add up to 17 million reads and why is it only 5 million? Considering that I'm using RNA STAR to do both the alignment and gene counting I thought they should be concordant.