def get_ensembl_organism_names(self, organism):
"""returns the specified organism name file contents, using
the EnsEMBL path"""
logging.info('RSAT - get_ensembl_organism_names(%s)', organism)
return util.read_url("/".join([self.base_url, RsatDatabase.DIR_PATH,
organism + '_EnsEMBL',
RsatDatabase.ORGANISM_NAMES_PATH]))
def get_obo_url(self,version):
url = 'http://viewvc.geneontology.org/viewvc/GO-SVN/ontology-releases/%s/' %(version)
body = util.read_url(url)
revision_pat = re.compile(r'<a href="/viewvc/GO-SVN\?view=revision&revision=(\d+)"')
m = revision_pat.search(body)
rev = m.group(1)
obo_url = 'http://viewvc.geneontology.org/viewvc/GO-SVN/ontology-releases/%s/go-basic.obo?revision=%s' %(version,rev)
return obo_url
CHECKPOINT_INTERVAL = 100
CHECKPOINT_FILE = None
if __name__ == '__main__':
print('cMonkey (Python port) (c) 2011-2012, Institute for Systems Biology')
print('This program is licensed under the General Public License V3.')
print('See README and LICENSE for details.\n')
if len(sys.argv) < 3:
print('Usage: ./run_cmonkey.sh <organism-code> <ratio-file> ' +
'[string-file] [checkpoint-file]')
else:
string_file=None
if len(sys.argv) > 3:
string_file = sys.argv[3]
if len(sys.argv) > 4:
CHECKPOINT_FILE = sys.argv[4]
matrix_factory = dm.DataMatrixFactory([dm.nochange_filter, dm.center_scale_filter])
matrix_filename = sys.argv[2]
if matrix_filename.startswith('http://'):
indata = util.read_url(matrix_filename)
infile = util.DelimitedFile.create_from_text(indata, has_header=True, quote='\"')
else:
infile = util.DelimitedFile.read(matrix_filename, has_header=True, quote='\"')
matrix = matrix_factory.create_from(infile)
cmonkey_run = cmonkey_run.CMonkeyRun(sys.argv[1], matrix,
string_file=string_file) # num_cluster=250 for halo_ref
cmonkey_run.run()
