Intersection based Ensemble calling approaches using inputs from multiple callers.

Prepare variant inputs for ensemble calling approaches.
Creates normalized, bgzipped tabix indexed inputs from VCF files.

Realignment based Ensemble calling approaches using inputs from multiple callers.
Uses tools from the Marth lab and Erik Garrison to realign and recall given a
set of possible variants in a genomic region.

Sort VCF sample columns to have a consistent order between multiple inputs.
Variant callers order called outputs differently and this ensures they are
consistent to feed into ensemble calling.

Merge multiple VCF files together, running in parallel over genomic regions.

Split variation calls into small genomic regions for parallel processing.

Performing squaring off of variant call sets, recalling at all sample positions.
This converts a merged dataset with no calls at positions not assessed in the
sample, into a fully 'square' merged callset with reference calls at positions
without evidence for a variant, distinguishing true no-calls from reference
calls.

Create VCF headers using algorithms contained in Picard/Tribble tools.
This does the best job of cleanly merging and organizing headers from
multiple variant calling approaches.

Utilities for manipulating VCF files

Helper functions to retrieve information from htsjdk VariantContext
objects, which represent variant data stored in VCF files.