bowtie2 local vs. end to end mode

[1500]

Hi,

I've made some bowtie2 test to find the best alignment parameters on artificial miRNA sequences. After a few unsatisfactory results, I've tried mapping my reference fasta data to itself.
Is there any explanation why with "end-to-end" mode I've got 100% alignment and with "local" mode only 96%? Shouldn't it be also 100%? These are the same sequences in aligning data and in reference.

Thanks for any response,
Marta

[carmeyeii]

this is interesting...

[raphael123]

In my case, a pair of read align slearly both on chromosome 3 on end to end.
By setting local I have the pair split between 2 chromosmes, I have a better match at a wrong place ..
I feel like I don t get the interest of local alignment except if you don t make a good trimming of you sequences.

My guess for your question is : With local alignment you have several hits when miRNA are close and you finally consider those reads as ambiguous ...