Skip to content


Subversion checkout URL

You can clone with HTTPS or Subversion.

Download ZIP
Commits on Feb 16, 2015
  1. Brad Chapman

    Support ##contig headers with only ID attributes. Generated by bcftoo…

    chapmanb authored
    …ls 1.2 when inputs have no ##contig information
Commits on Oct 24, 2014
  1. David Caplan
Commits on Oct 10, 2014
  1. Martijn Vermaat
Commits on Sep 17, 2014
  1. amwenger

    Simplify _format_sample logic

    amwenger authored
    The attribute is no longer set to None for
    uncalled calls, which means that _format_sample can now
    rely on obtaining the original sample genotype.
  2. amwenger
  3. amwenger

    Close file handles in TestUncalledGenotypes tests

    amwenger authored
    Warning about open file handles muddle the output of unit tests
    and are a potentially confusing factor to those interpreting
    the tests.
  4. amwenger

    Add test cases for uncalled genotypes support

    amwenger authored
    * Remember the ploidity of uncalled genotypes such that
      the sample genotypes written by PyVCF.Writer match the
      sample genotypes read by PyVCF.Reader.
    * For uncalled _Calls, gt_nums and gt_bases are None;
      gt_alleles is a list of "None" with a length of _Call.ploidity.
Commits on Sep 9, 2014
  1. Chris Lasher

    Bugfix: SNP records with N as ALT now noted as SNPs.

    gotgenes authored
    The VCF 4.0 and newer specifications say the ALT field is a comma
    separated list that includes "base Strings made up of the bases
    A,C,G,T,N". Notably, the last case was not handled by `Record.is_snp`,
    causing it to erroneously report `False` for records with "N" as the ALT.
  2. Martijn Vermaat

    Partial support for VCFv4.2

    martijnvermaat authored
    - Add R as an INFO field count (number of alleles including reference).
    - Support the optional Source and Version keys on INFO metainformation.
    Thanks alot @travc for contributing these fixes!
    See #172
Commits on Jul 6, 2014
  1. Martijn Vermaat

    Don't crash on metadata lines without value

    martijnvermaat authored
    The spec actually does not allow for metadata lines without value, but we
    shouldn't crash on them.
    Fixes #168
Commits on Jun 25, 2014
  1. Martijn Vermaat

    Merge pull request #166 from martijnvermaat/format-none

    martijnvermaat authored
    Don't crash when FORMAT is set to the missing value (.)
  2. Martijn Vermaat

    Don't crash when FORMAT is set to the missing value (.)

    martijnvermaat authored
    It is not valid according to the spec, but issue #164 shows a VCF file
    where the FORMAT column contains just a dot character. We have no way
    of interpreting the subsequent genotype columns in that case, so this
    patch ignores them.
  3. Martijn Vermaat

    Allow flag INFO field to be declared as string

    martijnvermaat authored
    As reported in #164, we previously crashed on flag INFO fields declared
    as strings (and the number of values declared as 1). This is indeed not
    according to spec, but we should probably allow it anyway.
Commits on Jun 8, 2014
  1. Martijn Vermaat

    Merge pull request #148 from mgymrek/master

    martijnvermaat authored
    making alternate allele frequency work in the case of non-diploid genotypes
Commits on May 20, 2014
  1. Chris Lasher

    Adds _Record.affected_start and .affected_end.

    gotgenes authored
    These coordinates should represent the zero-based, half-open region of
    the reference sequence affected by all the events included in ALT. These
    coordinates allow the user to identify precisely which bases are altered
    by the events in the record.
    Provides more thorough documentation on the coordinate schemes for
    _Record.POS, .start, and .end.
Commits on May 14, 2014
  1. Chris Lasher

    Fixes fetch documentation in package docstring.

    gotgenes authored
    This corrects several lines that relate to the changes to fetch brought
    in by Pull Request #156.
  2. Chris Lasher Martijn Vermaat

    Reader.fetch uses zero-based, half-open coordinates.

    gotgenes authored martijnvermaat committed
    These changes make the behavior of Reader.fetch consistent with with
    pysam.Tabixfile, which uses the zero-based, half-open coordinate system
    for Tabixfile.fetch. See
    Previously, PyVCF's Reader.fetch declared no particular coordinate
    system. Since the method quietly deducted 1 from the start position,
    apparently it assumed users were going to input a one-based coordinate
    there. However, users familiar with pysam's Tabixfile for other formats
    get an unexpected surprise when variants ahead of the start coordinate
    start getting returned by Reader.fetch.
    As _Record.start and _Record.end are in the ZBHO coordinate system, it
    adds to the consistency that fetch take start and end coordinates in
    ZBHO, so the same _Record instance could be retrieved using its .CHROM,
    .start, and .end coordinates.
    This change also removes the prior behavior of fetch of returning a
    single _Record instance if given only chrom and start coordinates, by
    implicitly doing a Tabixfile.fetch(chrom, start-1, start). The new
    behavior when omitting the end parameter is to return an iterator of
    _Record instances starting at start and continuing through the end of
    the chromosome chrom. Again, this is the behavior consistent with
    pysam.Tabixfile.fetch, and is what users ought to expect.
    This change also allows the user to omit both the start and end
    positions. In this case, an iterable of _Record instances for all
    records for the particular chromosome chrom will be returned, which
    again, is consistent with Tabixfile.fetch. This behavior also resolves
    Issue #123 "Cannot fetch() whole chromosome".
Commits on May 13, 2014
  1. Chris Lasher
  2. Chris Lasher

    Skips broken test for PyPy.

    gotgenes authored
  3. Chris Lasher
  4. Chris Lasher

    Marks skipped tests as skipped, not passed.

    gotgenes authored
    Decorates tests that are potentially skipped, as well as broken tests
    that are always skipped, as being skipped, rather than indicating
    falsely that these tests have passed (the result of premature return
    statements prior to any assertions in the tests).
    This introduces another dependency for Python 2.6, the unittest2 module,
    which back-ported this functionality from Python 2.7 and Python 3.
Commits on Mar 6, 2014
  1. Melissa Gymrek
  2. Melissa Gymrek
  3. Melissa Gymrek
  4. Melissa Gymrek
Commits on Feb 22, 2014
  1. Lenna Peterson
  2. Lenna Peterson

    Merge branch 'master' of into lenna

    lennax authored
Commits on Feb 21, 2014
  1. casbon

    fix missing .pyx

    casbon authored
Commits on Feb 10, 2014
  1. casbon

    bump version

    casbon authored
  2. casbon

    Added tests for walk_together with more complex inputs

    datagram authored casbon committed
  3. casbon

    Test data for testing the fix for issue #140

    datagram authored casbon committed
  4. casbon

    finished fixing edge case where 'other' is None

    datagram authored casbon committed
  5. casbon

    Fixed edge case where all inputs are empty, simplified logic

    Ben Weisburd authored casbon committed
  6. casbon

    Fixed spacing and wrapping in, removed test for old walk_tog…

    Ben Weisburd authored casbon committed
    …ether arg (eq function), fixed edge case in _AltRecord
  7. casbon

    Fix for issue #140, add vcf_record_sort_key arg

    datagram authored casbon committed
    - Added 'vcf_record_sort_key' to allow user to specify arbitrary chromosome ordering.
    - Fixed issue #140 by making sure to emit all records from the current chromosome before moving on to the next one. This takes care of the problem in most typical cases (eg. when all files have records for all contigs), but not in some edge cases, in which case the 'vcf_record_sort_key' arg can be used to fully solve the problem by explicitly defining the chromosome order.
Something went wrong with that request. Please try again.