def submit():
files = glob.glob(path_snplist+'/*.txt') #OBS: folder also contains "not_mapped.log"
#files = ['/home/unix/ptimshel/git/snpsnap/samples/sample_10randSNPs_fewmatches.list']
files.sort()
processes = []
for (counter, filename) in enumerate(files, start=1):
pheno = os.path.splitext(os.path.basename(filename))[0]
logger.info( "processing file #%d/#%d: %s" % (counter, len(files), pheno) )
user_snps_file = filename # full path
output_dir = path_output_sub+"/"+pheno
HelperUtils.mkdirs(output_dir)
#TODO: consider the potential problems with 'use' environment
#TODO: reuse Python-2.7 && bsub [...]
# Put this inside LaunchBsub!!
command_shell = "python {program:s} --user_snps_file {snplist:s} --output_dir {outputdir:s} --distance_type ld --distance_cutoff 0.5 match --N_sample_sets {N} --max_freq_deviation {freq} --max_distance_deviation {dist} --max_genes_count_deviation {gene_count}".format(program=script2call, snplist=filename, outputdir=output_dir, N=10000, freq=1, dist=5, gene_count=5)
processes.append( LaunchBsub(cmd=command_shell, queue_name=queue_name, walltime=walltime, mem=mem, jobname=pheno, projectname='snpsnp', path_stdout=path_stdout, file_output=pheno+'.txt', no_output=False, email=email, logger=logger) ) #
for p in processes:
p.run()
time.sleep(args.pause)
return processes
def submit():
files = glob.glob(path_snplist+'/*.txt') #[0:2], OBS: folder also contains "not_mapped.log"
#files = ['/home/unix/ptimshel/git/snpsnap/samples/sample_10randSNPs_fewmatches.list']
files.sort()
processes = []
for (counter, filename) in enumerate(files, start=1):
filename = re.sub(r'[()]', '', filename) #### OBS: changing file names!
pheno = os.path.splitext(os.path.basename(filename))[0]
logger.info( "processing file #%d/#%d: %s" % (counter, len(files), pheno) )
user_snps_file = filename # full path
output_dir = path_output_sub+"/"+pheno
HelperUtils.mkdirs(output_dir)
command_shell = "python {program:s} --user_snps_file {snplist:s} --output_dir {outputdir:s} --distance_type ld --distance_cutoff 0.5 match --N_sample_sets {N} --max_freq_deviation {freq} --max_distance_deviation {dist} --max_genes_count_deviation {gene_count}".format(program=script2call, snplist=filename, outputdir=output_dir, N=N_sample_sets, freq=freq, dist=dist, gene_count=gene_count)
#command_seq = "--user_snps_file {snplist:s} --output_dir {outputdir:s} --distance_type ld --distance_cutoff 0.5 match --N_sample_sets {N} --max_freq_deviation {freq} --max_distance_deviation {dist} --max_genes_count_deviation {gene_count}".format(snplist=filename, outputdir=output_dir, N=1000, freq=5, dist=20, gene_count=20)
#print command_shell
processes.append( LaunchSubprocess(cmd=command_shell, path_stdout=path_stdout, logger=logger, jobname=pheno) ) #
#time.sleep(1)
#p.run_Log(file_output=pheno+'.txt') # writes stdout and stdout to "file_output" file in PATH path_stdout. NO WAITING since output goes to file
for p in processes:
p.run_Pipe()
return processes
###################################### Global params ######################################
queue_name = "hour" # [bhour, bweek] priority
#queue_name = "priority" # [bhour, bweek] priority
walltime="59" # hh:mmm, e.g. [24:00=1day | 10:00=10hrs | 120=2hrs | 1:0=1hrs
mem="1" # gb
#email='*****@*****.**'
email=False
script2call = "/home/unix/ptimshel/git/snpsnap/snpsnap_query.py" # Updated path
current_script_name = os.path.basename(__file__).replace('.py','')
path_snplist = "/cvar/jhlab/snpsnap/data/input_lists/gwascatalog_140201_listsBIGbim"
path_output_main = "/cvar/jhlab/snpsnap/data/query/gwascatalog"
path_output_sub = path_output_main + "/output"
HelperUtils.mkdirs(path_output_sub)
path_stdout = path_output_main + "/stdout"
HelperUtils.mkdirs(path_stdout)
###################################### ARGUMENTS ######################################
args = ParseArguments()
###################################### LOGGER ######################################
## Setup-logger
#logger = Logger(__name__, path_stdout).get() # gives __name__ == main
logger = Logger(current_script_name, path_stdout).get()
#loglevel = getattr(logging, args.logger_lvl.upper(), logging.INFO) # returns some numeric value
loglevel = getattr(logging, args.logger_lvl.upper()) # returns some numeric value
logger.setLevel(loglevel)
#logger.setLevel(logging.INFO)
#logger.setLevel(logging.WARNING)
