Bowtie: number of alignments

Nick · 03-04-2011, 10:17 AM

I'm using bowtie in repetitive regions and need all possible alignments of these reads. My question is about the backtracking mentioned in the original bowtie paper, http://genomebiology.com/2009/10/3/R25. If I understand correctly the algorithm limits the number of backtracking steps to 125. That means that you get all valid alignments for 125 strings? Is this still the case? Is this irrelevant if you are only allowing 0-1 mismatches?

paulr · 03-04-2011, 11:58 AM

Assuming you're using the --all option:

In -n mode (quality-aware) you can specify --tryhard to get all (?) the alignments. Not sure if it guarantees all of them, but the manual claims maximum sensitivity with that.

In -v mode (ignore quals) it'll always report all the alignments (implied by the manual).

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Cuffdiff Gene Expression. ideal minimum number of alignments	SEQond	Bioinformatics	4	05-08-2012 08:12 AM
bowtie- paired end - no alignments	madsaan	Bioinformatics	1	06-27-2011 02:24 PM
Decoding colorspace alignments by Bowtie	rlegendre	Bioinformatics	0	02-09-2011 06:03 AM
Help with Bowtie, only unique alignments	khb	General	1	12-16-2010 01:35 AM
Bowtie got different alignments for fliped paired-end	狗熊趴趴	Bioinformatics	0	06-26-2010 08:39 AM

03-04-2011, 10:17 AM	#1
Nick Member Location: Philadelphia, PA Join Date: Jun 2009 Posts: 16	Bowtie: number of alignments I'm using bowtie in repetitive regions and need all possible alignments of these reads. My question is about the backtracking mentioned in the original bowtie paper, http://genomebiology.com/2009/10/3/R25. If I understand correctly the algorithm limits the number of backtracking steps to 125. That means that you get all valid alignments for 125 strings? Is this still the case? Is this irrelevant if you are only allowing 0-1 mismatches?

03-04-2011, 11:58 AM	#2
paulr Junior Member Location: USA Join Date: Mar 2011 Posts: 2	Assuming you're using the --all option: In -n mode (quality-aware) you can specify --tryhard to get all (?) the alignments. Not sure if it guarantees all of them, but the manual claims maximum sensitivity with that. In -v mode (ignore quals) it'll always report all the alignments (implied by the manual).