align target sequence

[colinmolter]

Hello,
I have used the GATKv3.0 pipeline to align and call variants on whole exome data. Recently, i was asked to get the variants from raw data sequenced using ION PGM. The data I get have been obtained by sequencing around a list of hundreds of variants. It means that i get a pretty small fastq file.

I know that using -L option it is possible to call variants around specific intervals (it save hours). My question is :
Is there a non-commercial tool that enables to fasten the alignment process by looking just around targetted intervals on my reference genome.

eg. for bwa, something like:

bwa bwasw hg19ref.fa [COLOR="DarkGreen"]-L intervals.bed[/COLOR] mysample.fastq

thanks
colin

[swbarnes2]

Hello,
I have used the GATKv3.0 pipeline to align and call variants on whole exome data. Recently, i was asked to get the variants from raw data sequenced using ION PGM. The data I get have been obtained by sequencing around a list of hundreds of variants. It means that i get a pretty small fastq file.

I know that using -L option it is possible to call variants around specific intervals (it save hours). My question is :
Is there a non-commercial tool that enables to fasten the alignment process by looking just around targetted intervals on my reference genome.

eg. for bwa, something like:

bwa bwasw hg19ref.fa [COLOR="DarkGreen"]-L intervals.bed[/COLOR] mysample.fastq

thanks
colin

In theory, you can just make a smaller reference genome to align to, which consists of only those stretches of DNA that you care about. In practice, you might get the aligner forcing reads that should not go there to align there, because it's going to try and put them somewhere. The last thing you want is to think you see a SNP, when it is in fact a bunch of reads that should have aligned eleswhere being forced to align in your target region.

You will get the most accurate alignment if you align to a reference that is as close as possible to your sample's sequence, but that means the whole genome.

[colinmolter]

Thanks for the answer,
making a smaller reference genome is of course a possibility. The only problem with that is that I will get the position adjusted to that reference genome and not to the whole genome. This means that I will have to do a conversion of the position after alignment. I was more thinking of tools having that option integrated.
(Looking to the net, it seems that roche's gs-amplicon might do that though i haven't tried it).

I was thinking that there might be software like bwa, bowtie2, soap, etc. that could manage this.
thanks
colin