Using SNP/Indel Callers with 454

[adaptivegenome]

Guys,

Sorry for the mundane question but after nearly exclusively using Illumina data and tools like DINDEL (developed for Illumina data), I find myself needing to call variants in 454 reads. I am well aware of the homopolymer problem for 454 but I wanted to ask generally how GATK UnifiedGenotyper and other variant callers work with 454 data. I have a sample sequenced with GAII and 454 and I want to call mutations in both and use the 454 data to validate the GAII data. But I don't want to bias the analysis if tools like GATK do not work correctly with 454.

Anyways I apologize for the lengthy question. Any insights would be appreciated.

[colindaven]

We found the alignment to be quite difficult, as Newbler doesn't produce SAM format.

We tried a number of aligners, but ended up using ClcGenomics workbench for convenience.

SNP calling works fine with Samtools, so I don't see why it wouldn't work with GATK. Homopolymers haven't proved to be a major problem after reference alignment.

[adaptivegenome]

We sequenced the same sample with GAII and 454. We mapped both samples back to a reference and now I want to call SNPs/Indels. I want to use 454 to validate calls from the Illumina data. My question is whether you can use the same variant callers for Illumina and 454. I know at least some are optimized for Illumina and so I do not know if it is appropriate to use with 454.

[Zam]

I've used Cortex (see http://dx.doi.org/10.1038/ng.1028 and/or cortexassembler.sourceforge.net) on both Illumina and 454 data with good results. The only thing I might do differently with 454 is use the homopolymer cutoff option (break a read at a homopolymer).

If you try it, you will need to start again with fastq rather than going from your mapping though I'm afraid.

best

Zam

[colindaven]

We haven't had any problems (if you leave aside BAQ) comparing Samtools SNP calls from Illumina and 454 data after reference alignment. I am not sure if samtools is optimised only for Illumina, but I doubt it.

We are happy with results so far, and some have been wet lab validated.

[bgulko]

I don't know if it would be helpful, but there is a Genotype caller I wrote for my Lab's use that has been tested on Illumina, 454, SOLiD and Sanger. It requires a samtools pileup as input, but is fully Bayesian and doesn't bias towards the reference. If its helpful, have a look at http://compgen.bscb.cornell.edu/GPhoCS/BSNP/
-- Brad

[adaptivegenome]

I don't know if it would be helpful, but there is a Genotype caller I wrote for my Lab's use that has been tested on Illumina, 454, SOLiD and Sanger. It requires a samtools pileup as input, but is fully Bayesian and doesn't bias towards the reference. If its helpful, have a look at http://compgen.bscb.cornell.edu/GPhoCS/BSNP/
-- Brad

Thanks Brad. This looks interesting. We will try it out. Have you done something similar for indels?