Bowtie mapping with more than 3 mismatches?

[rndouglas]

I'm trying to map some sRNA reads to specific loci of interest while allowing up to 3 mismatches using -v 3.

For some of my target loci, this seems to work well. All of the mapped reads have no more than 3 mismatches.

However, when I use the exact same code, changing only the bowtie files for a new locus, I'll get reads with up to 20 mismatches (and the sRNA read is only 24nt long!).

I've attached two screenshots from IGV showing the same sRNA sample mapped with Bowtie against two different loci. All of the nucleotides shown are mismatches.

How I expect it to always look:


How it usually turns out:


The code I'm using is as follows:

bowtie -f -v 3 -S ~/CRM2 ~/sRNA.fa > ~/output.sam

Oddly, I also have the exact same issue (too many mismatches) when I use this code:

bowtie -f -n 0 -l 25 -S ~/CRM2 ~/sRNA.fa > ~/output.sam

Am I doing something wrong? Has anyone else seen this?

[rndouglas]

Cranking up the frustration a notch....

This week I discovered the same situation (up to 20 mismatches on a 24 nt "mapped" read) occurs when I set -v to 0, supposedly not allowing for mismatches.



This happens with ~4-5 loci I'm interested in mapping to, but several other loci of similar length do not show this problem at all.

[Nepher]

Have the same problem

Hi all,

since this thread has no answer and that I came across exact the same issue I wanted to know if someone knows how this type of alignment is possible. Where is the mistake ?
I also have 24 mismatches on sequences that were mapped with NO mismatch option...

If you could help I would be very grateful

Thanks a lot

[rndouglas]

For me, it was an issue with IGV. When I looked more closely IGV was coloring nucleotides as being mismatched, but they really were not mismatched. I never found a workaround/explanation, so I just had to ignore all of the claimed mismatches.

Bowtie was mapping properly with zero mismatches allowed.

[Nepher]

Thanks for your answer.

I came across a solution this morning. After checking lots of stuff I looked again at my reference sequence and I saw that there were special characters at the beginning of the sequence :

>name
??????ATCGTGCTA

I simply erased them and IGV removed the most part of mismatches (it left those from the mapping). I think that the mapping coordinates where simply shifted/reference sequence.

Hope this will help other users with IGV problems