View Single Post
Old 07-23-2015, 10:13 AM   #3
Junior Member
Location: USA

Join Date: Jan 2013
Posts: 4

Thanks cmbetts for the reply.

I agree that "The tophat output says that 17.7% of your reads ~10M have multiple alignments. Those will all get counted as _alignment_no_unique since they can't be assigned to a unique site in the genome." So that explains the 1.8M I calculated.

"Additionally, they'll show up multiple times in the sam/bam file, once for each possible alignment, which is why you can have more overall counts from htseq-count than you have reads." Are you saying the unit of "__alignment_not_unique" is times instead of reads? E.g. One non-unique read mapped 6 locations will be added 6 times in "__alignment_not_unique" by htseq-count but it is counted as 1 by tophat?
chumho is offline   Reply With Quote