Feature Request: Automatically Validate .db models

[cfoguet]

Test that:

• -All the variants in .db file are in the covariance table
• -either rsids or VarIDs can match in harmonized GWAS file from GWAS catalogue (either rsid or VarID)

Have a test GWAS (harmonized file from GWAS catalogue) that can be used to test that PrediXcan.py (https://github.com/hakyimlab/MetaXcan) works (matches most SNPS and gets association pvalues even if not significant) with this input:

python SPrediXcan.py
--model_db_path "${model_db_fname}"
--gwas_file $gwas_file
--covariance $covariance_fname
--snp_column rsid --keep_non_rsid --model_db_snp_key rsid --separator "," --effect_allele_column effect_allele
--non_effect_allele_column other_allele
--or_column odds_ratio
--pvalue_column p_value
--output_file $out_name

--snp_column rsid and --model_db_snp_key rsid need to be changed to--snp_column VarID and --model_db_snp_key varID if varID is used to match SNPS

If the covariance table is very large, it should be able run with the stream covariance mode (stream covariant requires that the covariance table is sorted by gene)

python SPrediXcan.py
--model_db_path "${model_db_fname}"
--gwas_file $gwas_file
--covariance $covariance_fname --stream_covariance
--snp_column rsid --keep_non_rsid --model_db_snp_key rsid --separator "," --effect_allele_column effect_allele
--non_effect_allele_column other_allele
--or_column odds_ratio
--pvalue_column p_value
--output_file $out_name

When the covariance matrix dosen't need to be streamed the analysis pretty fast (<5 minutes), otherwise it can take ~4 hours