def addDataset(label, feature_matrix, associations, method, source, description, comments, configfile, results_path, ds_date, disease, contact):
print "Adding " + source + " dataset to admin table with config " + configfile + " for label " + label
if (description == ""):
#not general, revisit this to enter all TCGA known cancers
if (label.find("brca") != -1 or label.find("BRCA") != -1):
description = "Breast"
if (label.find("ov") != -1 or label.find("OV") != -1):
description = description + "Ovarian"
if (label.find("gbm") != -1 or label.find("GBM") != -1):
description = description + "Glioblastoma"
if (label.find("coadread") != -1 or label.find("COAD") != -1 or label.find("coad") != -1 or label.find("crc") != -1 or label.find("CRC") != -1):
description = description + "ColoRectal"
if (label.find("cesc") != -1 or label.find("CESC") != -1):
description = description + "Cervical"
if (label.find("hnsc") != -1 or label.find("HNSC") != -1):
description = description + "HeadNeck"
if (label.find("kirc") != -1 or label.find("KIRC") != -1 or label.find("kirp") != -1 or label.find("KIRP") != -1):
description = description + "Kidney"
if (label.find("luad") != -1 or label.find("LUAD") != -1 or label.find("lusc") != -1 or label.find("LUSC") != -1):
description = description + "Lung"
if (label.find("stad") != -1 or label.find("STAD") != -1):
description = description + "Stomach"
if (label.find("nomask") != -1):
description = description
elif (label.find("mask") != -1):
description = description + " filtered"
if (comments == ""):
comments = "{matrix:"+feature_matrix+",associations:"+associations+"}"
inputfiles = "{matrix:"+feature_matrix+",associations:"+associations+"}"
currentDate = time.strftime("%m-%d-%y")
config = db_util.getConfig(configfile)
max_logpv = -1.0
if (os.path.exists(results_path + 'edges_out_' + label + '_meta.json')):
meta_json_file = open(results_path + 'edges_out_' + label + '_meta.json','r')
metaline = meta_json_file.read()
if (len(metaline) > 1):
try:
max_logpv = json.loads(metaline)["max_logpv"]
except ValueError:
max_logpv = -1
#okay that the max_logpv is not set
except:
print "Unexpected error:", sys.exc_info()[0]
raise
meta_json_file.close()
summary_json = ""
if (os.path.exists(results_path + "feature_summary_" + label + ".json")):
summary_file = open(results_path + "feature_summary_" + label + ".json", "r")
summary_json = summary_file.read().strip()
summary_file.close()
insertSql = "replace into tcga.regulome_explorer_dataset (label,method,source,contact,comments,dataset_date,description,max_logged_pvalue, input_files, default_display,disease,summary_json) values ('%s', '%s', '%s', '%s', '%s', '%s', '%s', %f, '%s', '%i', '%s', '%s');" %(label, method, source, contact, comments,ds_date,description, max_logpv, inputfiles, 1, disease, summary_json)
print "updating regulome_explorer_dataset\n" + insertSql
db_util.executeInsert(config, insertSql)
def populate_sample_meta(sampleList, config):
"""
sampleList needs to be a list of patients
"""
global dataset_label
labelTokens = dataset_label.split("_")
cancer_type = labelTokens[0]
clabel = ""
samColIndex = 0
for sam in sampleList:
#REPLACE INTO `tcga`.`SampleMeta` (sample_key,cancer_type,dataset_label,matrix_col_offset,meta_json) VALUES ('a' /*not nullable*/,'s' /*not nullable*/,'s' /*not nullable*/,0,'s');
insertSampleSql = "replace into sample_meta (sample_key,cancer_type,dataset_label,matrix_col_offset,meta_json) values ('%s', '%s', '%s', '%i', '%s');" %(sam, cancer_type,clabel,samColIndex,"{age:X,status:someStatus,comments:some comments}")
db_util.executeInsert(config, insertSampleSql)
samColIndex += 1
print "Done populating sample list for " + dataset_label
def populate_sample_meta(sampleList, config):
"""
sampleList needs to be a list of patients
"""
global dataset_label
labelTokens = dataset_label.split("_")
cancer_type = labelTokens[0]
clabel = ""
samColIndex = 0
for sam in sampleList:
#REPLACE INTO `tcga`.`SampleMeta` (sample_key,cancer_type,dataset_label,matrix_col_offset,meta_json) VALUES ('a' /*not nullable*/,'s' /*not nullable*/,'s' /*not nullable*/,0,'s');
insertSampleSql = "replace into sample_meta (sample_key,cancer_type,dataset_label,matrix_col_offset,meta_json) values ('%s', '%s', '%s', '%i', '%s');" % (
sam, cancer_type, clabel, samColIndex,
"{age:X,status:someStatus,comments:some comments}")
db_util.executeInsert(config, insertSampleSql)
samColIndex += 1
print "Done populating sample list for " + dataset_label
sys.exit(-1)
if operation.upper() == "ADD":
pathwaymembers = raw_input("Enter pathway members(required and comma separated) e.g.\nTP53,GENE1,GENE2...\n")
pathwayurl = raw_input("Enter pathway source url(optional)\n")
if len(pathwaymembers) < 1:
print "Invalid pathway defined, check your inputs"
sys.exit(-1)
# print "name %s\n members %s\n source %s\n url%s" %(pathwayname, pathwaymembers, pathwaysource, pathwayurl)
insertSql = "insert into random_forest.pathways values('%s', '%s', '%s', '%s')" % (
pathwaysource,
pathwayname,
pathwayurl,
pathwaymembers,
)
rc = db_util.executeInsert(config, insertSql)
if rc >= 0:
print "%s added" % pathwayname
else:
print "Problems with adding - return code is %i" % rc
elif operation.upper() == "DELETE":
deleteSql = "delete from random_forest.pathways where pname = '%s' and psource = '%s'" % (
pathwayname,
pathwaysource,
)
rc = db_util.executeInsert(config, deleteSql)
if rc >= 0:
print "%s removed" % pathwayname
else:
print "Problems with deleting - return code is %i" % rc
else:
operation = raw_input("ADD or DELETE pathways?(required)\n")
pathwayname = raw_input("Enter pathway name(required)\n")
pathwaysource = raw_input("Enter pathway source(required but custom okay)\n")
if (len(pathwayname) < 1 or len(pathwaysource) < 1):
print "Invalid pathway defined, check your inputs"
sys.exit(-1)
if (operation.upper() == "ADD"):
pathwaymembers = raw_input("Enter pathway members(required and comma separated) e.g.\nTP53,GENE1,GENE2...\n")
pathwayurl = raw_input("Enter pathway source url(optional)\n")
if (len(pathwaymembers) < 1):
print "Invalid pathway defined, check your inputs"
sys.exit(-1)
#print "name %s\n members %s\n source %s\n url%s" %(pathwayname, pathwaymembers, pathwaysource, pathwayurl)
insertSql = "insert into random_forest.pathways values('%s', '%s', '%s', '%s')" %(pathwaysource, pathwayname, pathwayurl,pathwaymembers)
rc = db_util.executeInsert(config, insertSql)
if (rc >= 0):
print "%s added" %pathwayname
else:
print "Problems with adding - return code is %i" % rc
elif (operation.upper() == "DELETE"):
deleteSql = "delete from random_forest.pathways where pname = '%s' and psource = '%s'" %(pathwayname, pathwaysource)
rc = db_util.executeInsert(config, deleteSql)
if (rc >= 0):
print "%s removed" %pathwayname
else:
print "Problems with deleting - return code is %i" % rc
else:
print "operation %s not supported" %(operation)
def addDataset(
label,
feature_matrix,
associations,
method,
source,
description,
comments,
configfile,
results_path,
ds_date,
disease,
contact,
):
print "Adding " + source + " dataset to admin table with config " + configfile + " for label " + label
if description == "":
# not general, revisit this to enter all TCGA known cancers
if label.find("brca") != -1 or label.find("BRCA") != -1:
description = "Breast"
if label.find("ov") != -1 or label.find("OV") != -1:
description = description + "Ovarian"
if label.find("gbm") != -1 or label.find("GBM") != -1:
description = description + "Glioblastoma"
if (
label.find("coadread") != -1
or label.find("COAD") != -1
or label.find("coad") != -1
or label.find("crc") != -1
or label.find("CRC") != -1
):
description = description + "ColoRectal"
if label.find("cesc") != -1 or label.find("CESC") != -1:
description = description + "Cervical"
if label.find("hnsc") != -1 or label.find("HNSC") != -1:
description = description + "HeadNeck"
if label.find("kirc") != -1 or label.find("KIRC") != -1 or label.find("kirp") != -1 or label.find("KIRP") != -1:
description = description + "Kidney"
if label.find("luad") != -1 or label.find("LUAD") != -1 or label.find("lusc") != -1 or label.find("LUSC") != -1:
description = description + "Lung"
if label.find("stad") != -1 or label.find("STAD") != -1:
description = description + "Stomach"
if label.find("nomask") != -1:
description = description
elif label.find("mask") != -1:
description = description + " filtered"
if comments == "":
comments = "{matrix:" + feature_matrix + ",associations:" + associations + "}"
inputfiles = "{matrix:" + feature_matrix + ",associations:" + associations + "}"
currentDate = time.strftime("%m-%d-%y")
config = db_util.getConfig(configfile)
max_logpv = -1.0
if os.path.exists(results_path + "edges_out_" + label + "_meta.json"):
meta_json_file = open(results_path + "edges_out_" + label + "_meta.json", "r")
metaline = meta_json_file.read()
if len(metaline) > 1:
try:
max_logpv = json.loads(metaline)["max_logpv"]
except ValueError:
max_logpv = -1
# okay that the max_logpv is not set
except:
print "Unexpected error:", sys.exc_info()[0]
raise
meta_json_file.close()
summary_json = ""
if os.path.exists(results_path + "feature_summary_" + label + ".json"):
summary_file = open(results_path + "feature_summary_" + label + ".json", "r")
summary_json = summary_file.read().strip()
summary_file.close()
insertSql = (
"replace into tcga.regulome_explorer_dataset (label,method,source,contact,comments,dataset_date,description,max_logged_pvalue, input_files, default_display,disease,summary_json) values ('%s', '%s', '%s', '%s', '%s', '%s', '%s', %f, '%s', '%i', '%s', '%s');"
% (
label,
method,
source,
contact,
comments,
ds_date,
description,
max_logpv,
inputfiles,
1,
disease,
summary_json,
)
)
print "updating regulome_explorer_dataset\n" + insertSql
db_util.executeInsert(config, insertSql)
