Bowtie2 end-to-end versus local alignment for small RNAseq

Question

[theune]

hello,

I'm working with small RNAseq data from Illumina Trueseq small RNA, my reads have been trimmed and when I align with bowtie2's --end-to-end alignment, I get 96% overall alignment. When I use bowtie2's --local alignment I get only 30% alignment. Using bowtie1 I get less than 1% overall alignment.

Does anyone know what might be leading to this issue? I've read that local alignment or bowtie1 are more accurate for small RNA seq, and want to compare these alignment tools, but no matter what parameters I vary for either bowtie2 local or bowtie 1, I cannot improve my percent alignment.

[GenoMax]

Generally small RNA seq kits use a specific adapter that is ligated to the end of the small RNA. This adapter is present in the sequence and needs to be clipped off the ends of reads leaving 24-26 bp of small RNA (kit should have sequence of adapter and any specific instructions).

Since bowtie v.1x only does ungapped alignments it is not able to map the reads containing adapter. bowtie2 is likely doing spliced/gapped alignments which is why you can get the higher alignment %.

Proper way would be to trim the adapter and then do ungapped alignments with bowtie v.1.x.

[theune]

I've already trimmed 5' and 3' adaptor sequences. I'm running bowtie after trimming & removal of reads less than 17 bp.

[GenoMax]

Standard illumina adapters are different. Have you removed small RNA adapter (which is more than likely present in your data)?

[theune]

Yes, all adaptors are removed. I've inspected the reads after trimming they are all unique and after trimming several mapped to mature miRNAs when blasted on miRbase. Also, after running bowtie2 in end-to-end mode, I looked at where reads aligned in IGV viewer and the whole read is aligned to reference genome.

[xiaoleiusc]

are end-to-end and gapped/spliced alignments incompatible with each other?