p.add_option("--pi",action='store',dest="pi_name",default=None,
help="List data for PI(s) matching PI_NAME (can use glob-style "
"patterns)")
p.add_option("--unassigned",action='store_true',dest="unassigned",default=False,
help="List data for projects where PI is not assigned")
opts,args = p.parse_args()
# Collect data
audit_data = {}
unassigned = []
undetermined = []
for d in args:
for dirn in utils.list_dirs(d):
dirn = os.path.join(d,dirn)
#print "Examining %s" % dirn
try:
run = AnalysisDir(dirn)
for p in run.get_projects():
if p.name == "undetermined":
undetermined.append((p,get_size(p.dirn)))
continue
pi = p.info.PI
if pi is None:
# PI is not assigned
p.info['run'] = os.path.basename(dirn)
unassigned.append(p)
continue
elif opts.pi_name is not None:
if not fnmatch.fnmatch(pi,opts.pi_name):
# Skip non-matching name
continue
#print "\t%s: %s" % (p.name,pi)
p.add_option("--unassigned",
action='store_true',
dest="unassigned",
default=False,
help="List data for projects where PI is not assigned")
opts, args = p.parse_args()
# Collect data
audit_data = {}
unassigned = []
undetermined = []
for d in args:
for dirn in utils.list_dirs(d):
dirn = os.path.join(d, dirn)
#print "Examining %s" % dirn
try:
run = AnalysisDir(dirn)
for p in run.get_projects():
if p.name == "undetermined":
undetermined.append((p, get_size(p.dirn)))
continue
pi = p.info.PI
if pi is None:
# PI is not assigned
p.info['run'] = os.path.basename(dirn)
unassigned.append(p)
continue
elif opts.pi_name is not None:
if not fnmatch.fnmatch(pi, opts.pi_name):
# Skip non-matching name
continue
#print "\t%s: %s" % (p.name,pi)
if __name__ == "__main__":
p = optparse.OptionParser(usage="%prog DIR [PROJECT]")
p.add_option("-i","--init",action="store_true",dest="init",default=False,
help="initialise metadata file for the selected project (nb can "
"only be applied to one project at a time)")
p.add_option("-u","--update",action="append",dest="update",
help="update the metadata in the selected project by specifying "
"key=value pairs e.g. user='******' (nb can only be applied "
"to one project at a time)")
opts,args = p.parse_args()
if not args or len(args) > 2:
p.error("Need to supply a run directory and an optional project name")
# Try and load the data
try:
run = AnalysisDir(args[0])
print "Loaded data from %s" % args[0]
print "Found %d projects" % run.n_projects
if run.undetermined is not None:
print "Found 'undetermined' analysis"
print "Found %d sequencing data directories" % run.n_sequencing_data
except Exception,ex:
sys.stderr.write("Failed to load data for %s: %s\n" % (args[0],ex))
sys.exit(1)
# Aquire the projects
try:
target_project = args[1]
except IndexError:
target_project = None
projects = run.get_projects(pattern=target_project)
if not projects:
# Collect arguments
if len(args) != 4:
p.error("Wrong args")
galaxy_url = args[0]
library_name = args[1]
dest = args[2]
rundir = args[3]
# Subset of projects
if opts.projects is not None:
projects = opts.projects.split(',')
else:
projects = None
# Get analysis dir
analysis_dir = AnalysisDir(rundir)
print "Run name: %s" % analysis_dir.run_name
print "Run number: %s" % analysis_dir.metadata.run_number
print "Date stamp: %s" % analysis_dir.date_stamp
try:
# Create the directory structure on the server
build_library_directory(analysis_dir, dest, projects=projects)
# Upload to the Galaxy instance
create_data_library(galaxy_url,
library_name,
analysis_dir,
dest,
projects=projects,
no_verify=opts.no_verify)
except Exception as ex:
logging.critical("Export failed: %s" % ex)