DISCAVAR VCF

VCF utility written in Python for calldata and annotation based filtering, interactive exploration and identification of disease causative variations.

Installation

PIP

To install DisCaVar, run the following from the directory: python setup.py build sdist pip install dist/discavar-vcf-0.1.tar.gz

NOTE: Not all required dependencies can be installed automatically using pip

Dependencies

For merging different vcf files into single cohort file, bcftools is required.

Running

From commandline with the executable discavar. See discavar --help for details

Overview statistics for a vcf file:

    discavar vcfstats --vcf sample_vcf.ann.vcf.gz

Cohort Analysis with cohort analysis task:

    discavar cohort --tsv test_input.tsv --report-outdir "Reports" --regions chrom_regions.bed --quality-gt 20 --vep-consequence "HIGH" --vep-significance "unknown_significance", "pathogenic", "risk_factor"

Interactively form a python shell by importing from discavar. See documentation for details.

e.g.:

    from discavar import Cohort
    
    sample_file = "cohorts_input.tsv"
    cohorts = Cohort.from_tsv(sample_file, build_cohorts=True)
    
    # subtraction analysis
    for cohort in cohorts:
        cohort.healthy_vs_diseased()

    # calldata filter
    var_filter = VariantFilter(
                            min_gq=20,
                            min_dp=10,
                            excl_intervals=["chr1"])

    # annotation filter
    ann_parser = VEPAnnotation(c.variants.get_headers())
    impacts = ["MODERATE", "HIGH", "MODIFIER"]
    significance = ["benign", "likely_benign"]
    ann_filter = AnnotationFilter(ann_parser,
                                  IMPACT__in=impacts,
                                  CLIN_SIG__in=significance)

    for cohort in cohorts:
        cohort.add_filters([var_filter, ann_filter])
        cohort.apply_filters()


    # plotting
    for cohort in cohorts:
        cohort.manhattan()
        cohort.vtypes_dist()

    # cluster analysis
    ...
    # custom queries
    ...