Adds support to SciDB for loading genotype data from VCF (Variant Call Format) files. It includes custom data types and loaders.
To compile, add the genotypes directory to the examples subdirectory in the SciDB source tree, and include an 'add_subdirectory("genotypes")' directive in the examples/CMakeLists.txt file.
To load a typical set of VCF files into a set of related SciDB arrays with the common basename "foobar" you would use the following commands on the SciDB coordinator node:
$ loadgt.sh -d foobar_samples.csv foobar *.vcf.gz
$ samples_load.sh foobar foobar_samples.csv
The 'foobar_samples.csv' is comma separated list of samples containing the list of samples which matches those in the VCF header files, the populations to which they belong, and their sex.
#col,sample,population,founder,sex
1,NA19625,ASW,true,F
2,NA19700,ASW,true,M
3,NA19701,ASW,true,F
4,NA19703,ASW,true,M
5,NA19704,ASW,true,F
For optimal performance of queries involving an entire population, this file should be sorted by population, then by sample. The sample columns in the VCF file will be reordered according to the column numbers listed in this file. An convience script is included which may be used to create a sample file which can then be editted by hand.
$ samples_collect_from_vcf.sh *.vcf.gz > foobar_samples.csv
To create an array containing allele counts for each population, for each variation site:
$ allele_counts_calculate.py foobar
This will create an array with the three dimensions that denote a variation (chrom, pos, and var) and two additional dimensions, population and allele. In the allele dimension, an index of -1 is the total number of alleles, 0 is the number of alleles that are the same as the reference allele, 1 is the first alternate, 2 is the second, and so forth.