cga tools and phasing

[Marie_Noir]

Hi all,

I am currently looking for tools extracting phase information from NextGen data. I have found so far gatk readBacked Phasing algorithm, which performs nice on SNPs but not on indels. There is also samtools phase, however it produces two files which have to be brought together again which is kind of cumbersome (and don't know yet if it handles indels).

I have read something in the complete genomic data file format description of a "HapLink" column (denoting phase information) in masterVAR file - but did not find any tool in the cga tool suite which extracts phase information - does anybody know how it works? Is it integrated in the generateMASTERvar routine?

And, a second question which arised from the first one: How is the Complete Genomics var file generated? I also did not find any routine for this purpose, but this is needed as input in the generateMASTERvar routine. However I have read that the VAR file is delivered if the company sequenced a genome for you. But is there any possibility to generate a Complete Genomics VAR file on my own (and, even one more question: do Complete Genomics use BAM or SAM files at any stage of their data analysis pipeline?)

I have also read the User Guide and data format description pdfs (to some extent) but still did not figure out the very beginning of their pipeline - i.e. how do they do alignment (or mapping) and how is the variant calling done?

Thanks in advance!

[scordes]

The Complete Genomics variation or var file is one of the key results provided in a data package that is delivered as part of our sequencing service. The entire data package is generated from our proprietary Complete Genomics Analysis Pipeline software that we are running internally and is currently not available for customers to run themselves. The master variation or masterVar file is based on the var file and provides a reformatted and highly annotated list of variant calls. Both var and masterVar files are standard files that Complete Genomics provides with each sequenced genome. Both files also contain a haplinkID that indicates phasing of nearby (within ~700 kb) variants. The algorithm that is being used to generate phase information populated in the HapLink field is a component of the Analysis Pipeline, and thus, is not customer accessible.

As you have discovered, we do provide CGA Tools, a suite of open source tools that support the downstream analysis, manipulation, and format conversion of the files that we deliver to customers. Complete Genomics reads and mapping files can be converted into SAM/BAM format using the map2sam and evidence2sam CGA Tools (http://cgatools.sourceforge.net).

A paper describing the computational techniques underlying our mapping, assembly, and variant calling process used in the Analysis Pipeline software was recently accepted in the Journal of Computational Biology (Carnevali, et. al., in press). If you have specific questions that we can address ahead of its publication, please feel free to contact us at [email protected].

[agel]

There is also samtools phase, however it produces two files which have to be brought together again which is kind of cumbersome (and don't know yet if it handles indels).

How does samtools phase work, exactly?

Another option is Haplotype Improver. It uses a combination of paired-end data and older algorithms like phase and fastPhase. Not sure about indels either though.