genomvar 0.3.0

Case 1

Common task in genome variant analysis is: there are two VCF files (for example obtained from variant caller #1 and caller #2) and the differences should be investigated.

First we read the VCF files into genomvar genomvar.varset.VariantSet objects which hold the variants with underlying data contained in INFO fields:

>>> from genomvar.varset import VariantSet
>>> vs1 = VariantSet.from_vcf('caller1.out.vcf.gz',parse_info=True)
>>> vs2 = VariantSet.from_vcf('caller2.out.vcf.gz',parse_info=True)

To find variants detected by caller #1 but not caller #2 diff method is used. Then differences are exported to numpy for futher analysis:

>>> diff = vs1.diff(vs2)
>>> recs = diff.to_records() # recs is a numpy structured dtype array
>>> recs[['chrom','start','end','ref','alt','vartype']]
[('chr1',  1046755,  1046756, 'T', 'G', 'SNP')
 ('chr1',  1057987,  1057988, 'T', 'C', 'SNP')
  ...,
 ('chr19', 56434340, 56434341, 'A', 'G', 'SNP')
 ('chrY', 56839067, 56839068, 'A', 'G', 'SNP')]
>>> recs['INFO']['DP'].mean() # recs['INFO']['DP'] is a numpy ndarray
232.18819746028257

Case 2

There is a smaller variant file obtained from the data and a bigger one usually obtained from a database. Variants in the former should be “annotated” with some data associated with variants in the latter.

This case is different from the previous in that DB file might not comfortably fit into memory. Class genomvar.varset.VariantSetFromFile can be used for this purpose:

>>> vs = varset.VariantSet.from_vcf('vcf_of_interest.vcf')
>>> dbSNP = varset.VariantSetFromFile('DBSNP.vcf.gz', index=Trueg)
>>> annots = []
>>> for vrt in vs.iter_vrt():
>>>     m = dbSNP.match(vrt)
>>>     annots.append(m[0].attrib['id'] if m else None)
>>> annots
[None, None, 'rs540057607', 'rs367710686', 'rs940651103', ...]

Here genomvar.varset.VariantSet.match method is used. It searches for variants with the same genomic alteration as argument variant and returns a list of those. Then VCF ID field can be accessed from those matching variants in attrib['id'] (dbSNP rs numbers in this particular case).