equivalent of -m -best with bowtie2

[carolW]

Hi,
Is there an equivalent of -m and -best for bowtie2? Does -m for the version lower than 2 remove alignments with the worse scores and leaves the best all-m?

Look forward to your reply,

Carol

[dpryan]

I'll quote from the manual:

Bowtie 2 does away with Bowtie 1's notion of alignment "stratum", and its distinction between "Maq-like" and "end-to-end" modes. In Bowtie 2 all alignments lie along a continuous spectrum of alignment scores where the scoring scheme, similar to Needleman-Wunsch and Smith-Waterman.

For bowtie1, using the -m option should suppress all alignments for a read with more than -m found alignments. Look at the manual if that's unclear, they give a number of examples.

[carolW]

when using bowtie1, how to select alignments with -a --stratum options as quality values (0 or 255) should be ignored?

This refers also to another discussion http://seqanswers.com/forums/showthread.php?t=43596 on how to discriminate the reads mapping to repetitive regions from those that they have not mapped to repetitive regions and mapq cutoff was suggested to be used for this purpose.

[dpryan]

when using bowtie1, how to select alignments with -a --stratum options as quality values (0 or 255) should be ignored?

Sorry, that's not a coherent sentence. Do you mean, "Given a particular -a and --stratum setting in bowtie1, how should one determine which alignments to keep/ignore?"

[carolW]

yes. That's right. Sorry for confusion.

Also I think the sam file is in the same order as the reads and the reads in the sam file are numbered (incremental from 0). When using -a, does the multiple alignments for each read are on the consecutive lines and how to recognise the multiple aligments of each read, start line, end line for each read?

[dpryan]

It's been years since I've used that version of bowtie, so I can't say I recall exactly. I presume that one can just use the read name (in which case, it's just simple scripting in python/perl/etc.), but perhaps bowtie1 does something weird in these cases.

[bommel]

I'll quote from the manual:
Quote:
Bowtie 2 does away with Bowtie 1's notion of alignment "stratum", and its distinction between "Maq-like" and "end-to-end" modes. In Bowtie 2 all alignments lie along a continuous spectrum of alignment scores where the scoring scheme, similar to Needleman-Wunsch and Smith-Waterman.
For bowtie1, using the -m option should suppress all alignments for a read with more than -m found alignments. Look at the manual if that's unclear, they give a number of examples. .

Hej, does that mean that bowtie2 does not need to have these options because they are already included anyway?

[dpryan]

It's not that they're included but rather that they're not needed. The usage of MAPQ is superior in all ways and the previous methods should be forgotten entirely.

[bommel]

It's not that they're included but rather that they're not needed. The usage of MAPQ is superior in all ways and the previous methods should be forgotten entirely.

Thanks a lot I will have a look at my results