I am very new with analysis of RNA-seq. I have sent some fastq files (homo sapiens) from EBI to Galaxy. In order to obtain read counts, I aligned them to hg19 using HiSat (default parameters). Then since my reference genome was hg19, I used GTF file (Version 19 (July 2013 freeze, GRCh37) - Ensembl 74, 75) from Gencode to obtain read counts using htseq.

The total number of counts obtained for features is "10347508" which seems to be ok. While I have lost a number of counts about

__no_feature 2362227
__ambiguous 788874
__too_low_aQual 1001993
__not_aligned 2517255
__alignment_not_unique 3866370

Do you think the result is reasonable?

Something confusing is that from total 57820 genes, the counts for each gene up to gene 18356 are mostly non-zero, but counts for each gene from gene 18356 to gene 57820 are mostly zero (a few of them are non-zero).

Why is that?

Do you think I have to change my GTF file? Which version?

Or do you think I have to consider only the first 18356 genes for DE analysis ?

Thanks