in progress

A python package for analyzing Genome-Wide Association Summary Statistics (gwass). Currently, the package has two primary functions:

• Creating a table of SNPs with meta-data about derived and ancestral alleles / frequencies
• Cleaning up GWAS summary statistics and orienting the sign of effects to derived alleles

Currently the package has a specific focus such that it only works with the 1000 Genomes Project data but in the future this will be generalized.

• python3
• pysam
• numpy
• pandas

I recommend creating and activating a new conda python environment:

conda create -n <env_name> python
source activate <env_name>

Clone the repository and install into your python env

git clone https://github.com/jhmarcus/gwass
cd gwass/
make install

Make sure unit tests are working:

make test

wget ftp://ftp.1000genomes.ebi.ac.uk/vol1/ftp/release/20130502/ALL.wgs.phase3_shapeit2_mvncall_integrated_v5b.20130502.sites.vcf.gz
wget ftp://ftp.1000genomes.ebi.ac.uk/vol1/ftp/technical/reference/human_g1k_v37.fasta.gz
gunzip human_g1k_v37.fasta.gz

Assumes bcftools is installed (TODO: add bcftools command)

bcftools

wget http://portals.broadinstitute.org/collaboration/giant/images/0/01/GIANT_HEIGHT_Wood_et_al_2014_publicrelease_HapMapCeuFreq.txt.gz

Once the package has been installed it should add command-line executables to your path. One of these scripts will create a SNPs table given the 1KG sites vcf:

create-snps --sites <path>/ALL.wgs.phase3_shapeit2_mvncall_integrated_v5b.20130502.sites.biallelic.vcf.gz \
            --ref <path>/human_g1k_v37.fasta | \
            gzip -c > 1kg_phase3_snps.tsv.gz

note that this will take some time and is a pretty large file. The script will create the file 1kg_phase3_snps.tsv.gz with columns:

• chrom: chromsome number
• pos: reference genome position
• snp: rsID
• effect_allele: allele which effect sign is to be oriented to
• other_allele: other allele which effect sign is not oriented to
• ref_allele: the allele in the reference genome
• alt_allele: the alternate allele to the reference genome
• derived_allele: the derived allele determined by multi-way alignment
• ancestral_allele the ancestral allele determined by multi-way alignment
• minor_allele: the globally minor allele determined by 1KG Phase3
• major_allele: the globally major allele determined by 1KG Phase3
• allele_type: the type of the effect allele / other allele which is either derived_ancestral or minor_major
• ref_base_l2: the reference base two positions to the left of the SNP
• ref_base_l1: the reference base one positions to the left of the SNP
• ref_base_r1: the reference base one positions to the right of the SNP
• ref_base_r2: the reference base two positions to the right of the SNP
• f_sas: effect allele frequency in 1KG Phase3 in South Asians
• f_afr: effect allele frequency in 1KG Phase3 in Africans
• f_eas: effect allele frequency in 1KG Phase3 in East Asians
• f_eur: effect allele frequency in 1KG Phase3 in Europeans
• f_amr: effect allele frequency in 1KG Phase3 in Americans

note that this will take a bit of memory if loading the whole SNPs file

clean-summary-statistics --summary_statistics <path>/GIANT_HEIGHT_Wood_et_al_2014_publicrelease_HapMapCeuFreq.txt.gz \
                         --snps <path>/1kg_phase3_snps.tsv.gz
                         --out giant_height_summary_statistics

This will generate a file called giant_height_summary_statistics.tsv.gz with the same columns as the SNP table and two additional columns:

• beta_hat: the estimated effect size via OLS
• se: the standard error of the estimated effect size