Which aligner allow more then 3 mismatches?

[hannat]

same as topic

[adamdeluca]

based on your cutoff of three, I would guess you are using bowtie in -v mode, look at the -n mode

List of alignment tools:

http://seqanswers.com/forums/showthread.php?t=43

[hannat]

Bowtie -n mode, allow n(0-3) mismatch in first L(28bp) base seed, what about the rest of the reads beside the seed, how can i increase the mismatch?

[drio]

bfast and bwa. There are no specific parameters in bfast for setting the number
of mismatches. Bfast will perform local alignment to any read that aligns to the
reference genome. As for bwa, you have the -n parameter in bwa aln.

[adamdeluca]

Bowtie -n mode, allow n(0-3) mismatch in first L(28bp) base seed, what about the rest of the reads beside the seed, how can i increase the mismatch?

-e will set the sum of the quality of mismatched bases (higher = more mismatches)

http://bowtie-bio.sourceforge.net/manual.shtml#the--n-alignment-mode

[Gianza]

I don't know you might be interested, but Mosaik allows every threshold you want; then the only discrimination become that reads must align to only one place. (-m unique)

Hope might help

[john_mu]

http://biogibbs.stanford.edu/~jiangh/SeqMap/

SeqMap allows up to 5 mismatches/deletions

But it is slower than the other aligners if you want to take advantage of this feature.

[lh3]

Bfast, mosaik and bowtie are not guaranteed to find 4-mismatch hits. They find a fraction of them, determined by read lengths.

[zee]

Novoalign is able to handle this task with Illumina and SOLiD reads. The aligner can be downloaded for free at www.novocraft.com

[sklages]

cross_match / phaster though not especially designed for short-reads ..

[mrawlins]

There have been lots of comments on how to use other programs or parameter sets to get around this, but I thought I'd comment on why some programs limit to 3 mismatches.

The Burrows-Wheeler algorithm (used by bowtie and BWA) is really fast, but doesn't handle mismatches very well. There are at least two workarounds used, one is the seed-extend method (so it only really has to handle mismatches in the seed) and the other is a backtracking heuristic. The backtracking heuristic isn't robust enough to work for more than 3 mismatches without losing all of the speed gains for using Burrows-Wheeler.

Other programs use different algorithms, which will be slower for fewer mismatches, but able to handle 4+ mismatches without losing lots of speed or taking up tons more memory. That's why it's useful to switch to a program that doesn't map using Burrows-Wheeler when you want a large amount of mismatch tolerance within the first 20-30 nucleotides.

[rwenang]

You can try batman-aligner, it allows you to map up to 15-mismatches in the whole read.

http://code.google.com/p/batman-aligner/

Although based on my experience, if you allow more than 8 mismatch in the alignment, usually the non-bwt-based tools (maq,rmap) will be faster.