if not result["pvalue"] == "NA":
p = float(result["pvalue"])
eqtls[snpId][geneSymbol]["tissues"][result["tissueId"]] = {"pvalue": p}
print "Writing to eQTL file..."
if os.path.isfile(input_dir + "/eqtls.txt"):
upsert = raw_input("Appending output to existing eQTL file (%s). Continue? [y/N] " % input_dir + "/eqtls.txt")
if not upsert.lower() == 'y':
print "Did not write to existing SNP database."
return
with open(input_dir + "/eqtls.txt",'a') as eqtlfile:
for snp in eqtls.keys():
for gene in eqtls[snp].keys():
if not gene == "snp_info":
for tissue in eqtls[snp][gene]["tissues"].keys():
eqtlfile.write("%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\n" % (snp,eqtls[snp]["snp_info"]["chr"],eqtls[snp]["snp_info"]["locus"],gene,eqtls[snp][gene]["gene_chr"],eqtls[snp][gene]["gene_start"],eqtls[snp][gene]["gene_end"],eqtls[snp][gene]["cell_lines"],eqtls[snp][gene]["cis?"],eqtls[snp][gene]["p_thresh"],tissue,eqtls[snp][gene]["tissues"][tissue]["pvalue"]))
if __name__ == "__main__":
parser = argparse.ArgumentParser(description="Tool for re-sending failed GTEx requests, and optionally appending the results to an existing eQTLs file.")
parser.add_argument("-i","--input_dir",required=True,help="The output directory from the CoDeS3D run containing failed_requests.")
parser.add_argument("-c","--config",default="docs/conf.py",help="The configuration file to be used in this instance (default: conf.py)")
parser.add_argument("-p","--num_processes",type=int,default=1,help="Desired number of processes for multiprocessing (default: 1).")
args = parser.parse_args()
config = configparser.ConfigParser()
config.read(args.config)
gene_database_fp = config.get("Defaults","GENE_DATABASE_FP")
reqLists = parse_failed_requests(args.input_dir + "/failed_GTEx_requests.txt")
snps = codes3d.parse_snps_files(args.input_dir + "/snps.txt")
genes = codes3d.parse_genes_files(args.input_dir + "/genes.txt")
get_GTEx_responses(reqLists,snps,genes,gene_database_fp,args.num_processes,args.input_dir)
nargs='+',
help=
"Space-separated list of cell lines to include (others will be ignored). NOTE: Mutually exclusive with EXCLUDE_CELL_LINES."
)
parser.add_argument(
"-x",
"--exclude_cell_lines",
nargs='+',
help=
"Space-separated list of cell lines to exclude (others will be included). NOTE: Mutually exclusive with INCLUDE_CELL_LINES."
)
parser.add_argument(
"-o",
"--output_dir",
default="hiCquery_output",
help=
"The directory in which to output results (\"hiCquery_output\" by default)."
)
args = parser.parse_args()
config = configparser.ConfigParser()
config.read(args.config)
hic_data_dir = config.get("Defaults", "HIC_DATA_DIR")
if not os.path.isdir(args.output_dir):
os.makedirs(args.output_dir)
snps = codes3d.parse_snps_files(args.snps_files)
interactions = codes3d.find_interactions(snps, hic_data_dir,
args.include_cell_lines,
args.exclude_cell_lines,
args.output_dir)
type=int,
default=1,
help="Desired number of processes for multiprocessing (default: 1).")
parser.add_argument(
"-f",
"--fdr_threshold",
type=float,
default=0.05,
help=
"The FDR threshold to consider an eQTL statistically significant (default: 0.05)."
)
parser.add_argument(
"-o",
"--output_dir",
default="hiCquery_output",
help=
"The directory in which to output results (\"hiCquery_output\" by default)."
)
args = parser.parse_args()
config = configparser.ConfigParser()
config.read(args.config)
eqtl_data_dir = config.get("Defaults", "EQTL_DATA_DIR")
gene_database_fp = config.get("Defaults", "GENE_DATABASE_FP")
if not os.path.isdir(args.output_dir):
os.makedirs(args.output_dir)
snps = codes3d.parse_snps_files([args.snps_file])
genes = codes3d.parse_genes_files([args.genes_file])
eqtls, num_sig = codes3d.find_eqtls(snps, genes, eqtl_data_dir,
gene_database_fp, args.fdr_threshold,
args.local_databases_only,
args.num_processes, args.output_dir)