class PosSelClusterTest(unittest.TestCase):
def setUp(self,):
session = Session()
# family_id = 19187
# protein_id = 1151960
# domain_id = 1302435
family_id = 18883
protein_id = 1063014
domain_id = 1212855
self.family = session.query(Family).get(family_id)
self.protein = session.query(Protein).get(protein_id)
self.domain = session.query(Domain).get(domain_id)
self.ps_sites = get_possel_sites(self.family)
(self.pdb_id, self.pdb_chain, self.struct) = get_domain_struct(self.domain)
print "Testing on Family {0}, Protein {1}, Domain {2}, Structure {3}, PDB {4}{5}".format(
family_id, protein_id, domain_id, self.struct.id, self.pdb_id, self.pdb_chain
)
print "Sites of +Sel for family: ", self.ps_sites
self.pdb_struct = BioPDBStruct(self.pdb_id, self.pdb_chain, debug=True)
self.cluster_id_str = "Family {0}, Protein {1}, Domain {2}, Structure {3}".format(
family_id, protein_id, domain_id, self.struct.id
)
def testRunClusterAnalysis(self,):
domain_sites = get_domain_sites(self.family, self.protein, self.domain, self.ps_sites)
print "Local sites for domain {0}: ".format(self.domain.id), domain_sites
self.pdb_struct.cluster_analysis(
domain_sites, sample_size=10, store_file="ps-cluster-test.pkl", tag=self.cluster_id_str, report=True
)
def cluster_driver(family_id):
session = Session()
family = session.query(Family).get(family_id)
if family == None:
raise Exception("Family {0} could not be fetched from the database".format(family_id))
# Get sites for family (+Sel and TODO: firedb)
ps_sites = get_possel_sites(family)
# DEBUG
print "Start cluster analysis for family {0}".format(family.id)
print "Family {0} sites of +sel: ".format(family.id), ps_sites
# Get repr protein from family (first one)
protein = family.proteins[0]
# Attempt clustering on structures for all domains in protein
for domain in protein.domains:
try:
(pdb_id, pdb_chain, struct) = get_domain_struct(domain)
except Exception as e:
print e
print "Domain {0} (Family {1}, Protein {2}) has no valid structure. Skipping..".format(
domain.id, family.id, protein.id
)
continue
# Get domain-specific sites
domain_sites = get_domain_sites(family, protein, domain, ps_sites)
# DEBUG
print "Family {0}, Protein {1}, Domain {2} local sites: ".format(family.id, protein.id, domain.id), domain_sites
# Create BioPDBStruct object to call cluster analysis on
pdb_struct = BioPDBStruct(pdb_id, pdb_chain, debug=True)
cluster_id_str = "Family {0}, Protein {1}, Domain {2}, Structure {3}".format(
family.id, protein.id, domain.id, struct.id
)
try:
pdb_struct.cluster_analysis(
domain_sites, sample_size=samples, store_file=results_file, tag=cluster_id_str, report=True
)
except Exception as e:
print e
print "Can not complete clustering analysis on domain {0}. Skipping..".format(domain.id)
continue
# DEBUG
print "Clustering for domain/structs in Family {0} (Protein {1}) complete".format(family.id, protein.id)