I have some SE100 RNA-Seq data I'm exploring for gene expression by mapping it back to the sequence of a single gene using Bowtie2.

Bowtie2 identifies some valid alignments, but when I view the sorted/indexed bam file, the results seem to almost be shifted one nucleotide to the left from the reference sequence. I used both --very-sensitive-local and --very-sensitive and compared the outputs and they both seem to have this shift visible.

Is this a feature of the program itself, or do I have some misunderstanding of the algorithm or output from the program?

I've attached some IGV images, if that helps visualize the question.

IGV Images