[Bioc-devel] Bioc needs better support for variants

On Fri, Apr 15, 2011 at 7:19 AM, Martin Morgan <mtmorgan at fhcrc.org
<mailto:mtmorgan at fhcrc.org>> wrote:

    On 04/15/2011 06:05 AM, Vincent Carey wrote:

        I will comment on my limited view and progress.  I need to work from
        an exemplar.  I committed cheung2010 in the experimental data
        archive
        (devel only).  This relates to PMID 20856902, genetics of expression
        in immortalized B cells.

        There are 147 individuals with hapmap phase 3 genotypes and hgfocus
        arrays (:-( but about 45 have RNA-seq data in GEO.  fastq is
        available
        with the SRAtools fastq-dump and you can get the sra data reasonably
        quickly using ascp.  I will eventually make a sample from their
        RNA-seq data available in this package to look at SNP-driven
        allele-specific expression and other aspects of SNP-dependent
        expression regulation.

        Probably there is DNA-seq data out there on these coriell cell lines
        but for the moment I will be looking at the chip-based SNPs and
        imputation on those.  Better representations for 8 million SNP per
        sample would probably come in handy, but breaking them up by
        chromosome in SnpMatrix instances is OK so far.  I think we have to
        recognize that in any of these paradigms discrete calls are
        often not
        going to cut it, and uncertainty representations will be important.

        VCF representations of indels in 1000 genomes are available, but I
        don't know that we have good tools for importing and modeling those.
        Another exemplar that should be considered.

        On Fri, Apr 15, 2011 at 7:16 AM, Michael Lawrence
        <lawrence.michael at gene.com <mailto:lawrence.michael at gene.com>>
          wrote:

            Hi guys,

            Congrats on the release. For this next one, one focus, in my
            opinion, should
            be on analyzing variants in the context of sequencing data.
            This includes
            infrastructure for things like calling variants (in DNA and
            RNA), as well as
            determining their effects (e.g., coding and splicing
            changes). It would be
            good if we could come up with a plan. If we had one, we
            could commit some
            resources here to the problem.

            Is anyone willing to help out on this? What do you guys think?


    We could certainly play a role in annotation of variants and support
    for interfacing with established 3rd party formats. Obviously also
    the representation of variants that overlap with IRanges /
    Biostrings infrastructure. Martin



Great, this is in line with what I was thinking. We need a way to
formally represent sets of variants, as well as transcripts and proteins
(i.e., something based on a GRange[List]). Then we can map between
coordinate systems and request the consequences of mutations. I was
looking at the Ensembl variations Perl API;  it might be good for
inspiration.

Is there somewhere like a wiki where we could start hashing this out?

Michael



            Thanks,
            Michael

                    [[alternative HTML version deleted]]