def test_gm(log_files, source_dataframe):
print "Testing gene map and network map"
ni = NetworkInfo( "net_info_table","c2.cp.biocarta.v4.0.symbols.gmt")
genes = []
for p in ni.get_pathways():
genes.append(set(ni.get_genes(p)))
for log in log_files:
with open(log, 'r') as l:
for run_json in l:
run = json.loads( run_json )
for allele in run['alleles']:
source_allele = "%s-%s" % ( run['file_id'], allele )
working_dir = find_working(
run['data_matrices'][source_allele] )
gm_file = join( working_dir,
run['data_matrices'][ source_allele ]['gm'])
nm_file = join( working_dir,
run['data_matrices'][ source_allele ]['nm'])
gm = np.load(gm_file)
nm = np.load(nm_file)
for i in range(1, len(nm)):
start = nm[i-1]
end = nm[i]
if start < end:
base_set = genes[i-1]
my_net = gm[2*start:2*end]
my_net_genes = source_dataframe.index[my_net]
gene_set = set(my_net_genes)
assert gene_set.issubset(base_set), "Genes do not match network genes"
assert (end - start) == scipy.misc.comb(len(gene_set),2, exact=1), "Wrong number of genes"
print "Gene map and network map tests ... PASSED"
def networks(self):
if self._pathways is None:
ni = NetworkInfo( *self.net_info )
self._pathways = ni.get_pathways()
self.logger.debug("Got Networks")
return self._pathways
def deprecated_networks(self):
if self._pathways is None:
print self.net_info
ni = NetworkInfo( *self.net_info )
self._pathways = ni.get_pathways()
return self._pathways