def test1(self):
pdbIds = ["1I1G"]
self.pdb = download_mmtf_files(pdbIds)
pdb_1 = self.pdb.flatMap(StructureToBioassembly()) \
.flatMap(StructureToProteinDimers(8, 20, False, True))
self.assertTrue(pdb_1.count() == 4)
def test2(self):
pdbIds = ["5NV3"]
self.pdb = download_mmtf_files(pdbIds, self.sc)
pdb_2 = self.pdb.flatMap(StructureToBioassembly()) \
.flatMap(StructureToProteinDimers(8, 20, False, True))
self.assertTrue(pdb_2.count() == 12)
def test4(self):
pdbIds = ["1BZ5"]
# C5-B4
# C6-B3
# D7-A2
# D8-A1
# E10-E9
self.pdb = download_mmtf_files(pdbIds, self.sc)
pdb_4 = self.pdb.flatMap(StructureToBioassembly()) \
.flatMap(StructureToProteinDimers(9, 20, False, True))
self.assertTrue(pdb_3.count() == 5)
def test3(self):
pdbIds = ["4GIS"]
# A3-A2
# A4-A1
# B5-A1
# B6-A2
# B6-B5
# B7-A3
# B7-A4
# B8-A4
# B8-B7
self.pdb = download_mmtf_files(pdbIds, self.sc)
pdb_3 = self.pdb.flatMap(StructureToBioassembly()) \
.flatMap(StructureToProteinDimers(8, 20, False, True))
self.assertTrue(pdb_3.count() == 9)
protein = mmtfReader.download_mmtf_files(["1STP"], sc)
# ## Map protein structure to Bioassembly, then map to protein dimers
# In[4]:
cutoffDistance = 8.0
contacts = 20
useAllAtoms = False
exclusive = True
dimers = protein.flatMap(StructureToBioassembly()) .flatMap(StructureToProteinDimers(cutoffDistance, contacts, useAllAtoms, exclusive))
# ## Count number of structures
# In[5]:
print(f"Number of structures : {dimers.count()}")
# ## Terminate Spark
# In[6]: