bowtie with long reads, is it a problem?...help!

[cp229]

Hi everyone, I am a very new beginner with bowtie. And I have some trouble about reads...When I run bowtie with long reads(the read length is nearly 200), the misaligned percentage is nearly 98% (I tried human, yeast and e_coli...)
The result is very weird and I cannot find the reason for it...Is there anyone who knows how to solve this problem? Please help me


Thanks!

[simonandrews]

Bowtie is, by design, a short read aligner. You can make it work with longer reads, but the default setup will not work well there. The major limitation is that it doesn't allow the insertion of gaps in alignments, so that any indel errors will probably make your alignment fail.

Bowtie also has a cutoff for the sum of the mismatched quality scores. The way it's set up you only need around 3 mismatches for this to reject an alignment. If you are using long reads then you will probably want to increase this value to allow for more flexibility in your alignments.

[cp229]

Thank you very much...as you say, I tried another e_coli sample which is 35 reads, it`s working nicely. I should use another algorithm to solve the long read problem.

[lh3]

I think for reads longer than 100bp, gapped alignment become important. for long reads (>150bp), try ssaha2/mosaik/bwasw.

[cp229]

Thx!
When we use mosaik, we found it failed to align long genome, such as yeast and human, it seems that mosaik only works well with very short genome...Is that true??

[Mark]

Is bowtie2 more effective given that it permits gaps?