Roche 454 amplicon alignment by BWA-SW

[adri.ldg]

Hi,
I am using BWA-SW for the alignment of 454 amplicons and later detection of variants. When I do my proof of concept, I find that if the variant is in 2nd bp 3' or 5', BWA-SW does't map 1st and 2nd bases so variant caller doesn't find the variant.

Could anyone help me with the parameters should I use? By the moment I am using this command:

./bwa bwasw -a 1 -b 3 -q 3 -r 1 -d 0 ${ref} $files/file.fq > ${alignments}/file.sam

Thank you very much in advance! Any hint will be usefull!

[Torst]

BWA is a "local" aligner. If you require that the WHOLE query sequence must be included in the alignment, you need a "glocal" aligner (global in query, local in target). You can use glsearch36 from the FASTA36 package, or exonerate with the correct --model option.

[lh3]

Use bwa-mem. It deals with the problem in a better way.

[adri.ldg]

Thank you very much! I will try both options

[Torst]

FYI - lh3 is the author of bwa-mem, but he seems to know what he is doing ;-)

I think bwa-mem still focuses on local alignment, but if it can find a glocal alignment that is nearly as good (high scoring) as the local alignment, it will report that instead.

[adri.ldg]

Brilliant! Thank you so much!

bwa mem finds mismatches at de begining of the amplicon reads!

glsearch36 can help me for other things too