Bowtie2 Local Alignment Mode

[SHeaph]

Hi,

Can someone please advise on the below taken from the Bowtie2 manual?

When the --local option is specified, Bowtie 2 performs local read alignment. In this mode, Bowtie 2 might "trim" or "clip" some read characters from one or both ends of the alignment if doing so maximizes the alignment score.


How does Bowtie2 calculate the number of characters that are trimmed or clipped from the alignment ends?

If for example, I am mapping a 100bp read to a genome by local alignment and if only 20bp's in the middle of the read align, does bowtie2 trim or clip the 40bp's from either side to maximize the alignment score?

Many thanks in advance!

[kleinman]

I haven't used that setting myself, but I imagine Bowtie2 implements some version of the Smith-Waterman algorithm. So it would dynamically search over all possible local alignments and return the maximal-scoring one.

[carmeyeii]

Does anyone know how many bases it is limited to clipping when trying to find a match to the reference?

I have a set of reads that come from transposon insertion junctions and wanted to use some sort of local alignment tool to get the positions at which each read maps in the transposon itself, and also in the sequence where it generally inserts into.

Any insight?

Thanks!
Carmen

[shi]

Dear Carmen,

You may find this paper useful for your project: http://www.ncbi.nlm.nih.gov/pubmed/23533276. In this paper, the Subread aligner was used to aligner transposon reads to the mouse genome.

Cheers,
Wei

[shi]

I haven't used that setting myself, but I imagine Bowtie2 implements some version of the Smith-Waterman algorithm. So it would dynamically search over all possible local alignments and return the maximal-scoring one.

If you are interested in find local alignments, you may try Subread (http://subread.sourceforge.net). It automatically determines if a read should be locally aligned or globally aligned, by using a number of subreads (16mers) extracted from the entire read.

Cheers,
Wei