def extract_feature(self):
seed(self.seed)
counter = 0
print_info_nn(" >>> Adding D2 category based shape distribution for database {0} ... ".format(self._database.name))
overall_time = datetime.now()
if not os.path.exists(self._get_dir_name()):
os.makedirs(self._get_dir_name())
for complex_name in self._database.complexes.keys():
protein_complex = self._database.complexes[complex_name]
proteins = [protein_complex.unbound_formation.ligand, protein_complex.unbound_formation.receptor]
for protein in proteins:
shape_dist_file = self._get_dir_name() + protein.name
if not os.path.exists(shape_dist_file + ".npy"):
counter += 1
if counter <= 15:
print_info_nn("{0}, ".format(protein.name))
else:
counter = 0
print_info("{0}".format(protein.name))
atoms = protein.atoms
neighbour_search = NeighborSearch(atoms)
distributions = np.zeros((len(protein.residues), self.number_of_bins))
for i in range(len(protein.residues)):
residue = protein.residues[i]
nearby_residues = neighbour_search.search(residue.center, self.radius, "R")
distributions[i, :] = self._compute_distribution(nearby_residues)
np.save(shape_dist_file, distributions)
distributions = np.load(shape_dist_file + ".npy")
for i in range(len(protein.residues)):
protein.residues[i].add_feature(Features.D2_CATEGORY_SHAPE_DISTRIBUTION, distributions[i, :])
print_info("took {0} seconds.".format((datetime.now() - overall_time).seconds))
def extract_feature(self):
seed(self.seed)
print_info_nn(
" >>> Adding D1 surface atoms shape distribution for {0} ... ".
format(self._database.name))
overall_time = datetime.now()
if not os.path.exists(self._get_dir_name()):
os.makedirs(self._get_dir_name())
for complex_name in self._database.complexes.keys():
protein_complex = self._database.complexes[complex_name]
proteins = [
protein_complex.unbound_formation.ligand,
protein_complex.unbound_formation.receptor
]
for protein in proteins:
shape_dist_file = self._get_dir_name() + protein.name
if not os.path.exists(shape_dist_file + ".npy"):
print_info("{0}".format(protein.name))
pdb_file_name = self._database.directory + pdb_directory + protein.name + '.pdb'
surface, normals = get_surface_atoms(pdb_file_name)
distributions = np.zeros(
(len(protein.residues), 2 * (self.number_of_bins + 1)))
for i in range(len(protein.residues)):
residue = protein.residues[i]
distributions[i, :] = self.get_distributions(
residue.center, surface, normals)
np.save(shape_dist_file, distributions)
distributions = np.load(shape_dist_file + ".npy")
for i in range(len(protein.residues)):
protein.residues[i].add_feature(
Features.D1_SURFACE_SHAPE_DISTRIBUTION,
distributions[i, :])
print_info("took {0} seconds.".format(
(datetime.now() - overall_time).seconds))
def extract_feature(self):
print_info_nn(" >>> Adding secondary structure for database {0} ... ".format(self._database.name))
overall_time = datetime.now()
counter = 0
if not os.path.exists(self.__get_dir_name()):
os.mkdir(self.__get_dir_name())
for complex_name in self._database.complexes.keys():
protein_complex = self._database.complexes[complex_name]
proteins = [protein_complex.unbound_formation.ligand, protein_complex.unbound_formation.receptor]
for protein in proteins:
protrusion_file = self.__get_dir_name() + protein.name
if not os.path.exists(protrusion_file+".npy"):
counter += 1
if counter <= 15:
print_info_nn("{0}, ".format(protein.name))
else:
counter = 0
print_info("{0}".format(protein.name))
pdb_file = self._database.directory + pdb_directory + protein.name + ".pdb"
result_dict = run_psaia(pdb_file)
protrusion_array = np.zeros((len(protein.residues), 5 + 5 + 5 + 6 + 6 + 1))
if result_dict is not None:
for index, residue in enumerate(protein.biopython_residues):
key = self.get_residue_id(residue.get_full_id())
if key in result_dict:
values = result_dict[key]
protrusion_array[index, :] = self._normalize_features(*values)
else:
print('key not found in PSAIA processing!')
np.save(protrusion_file, protrusion_array)
protrusion_array = np.load(protrusion_file+".npy")
for index, residue in enumerate(protein.residues):
residue.add_feature(Features.PROTRUSION_INDEX, protrusion_array[index, 21:])
print_info("took {0} seconds.".format((datetime.now() - overall_time).seconds))
def extract_feature(self):
secondary_structure_dict = dict(
zip(ss_abbreviations, range(len(ss_abbreviations))))
print_info_nn(
" >>> Adding secondary structure for database {0} ... ".format(
self._database.name))
overall_time = datetime.now()
counter = 0
if not os.path.exists(self.__get_dir_name()):
os.mkdir(self.__get_dir_name())
for complex_name in self._database.complexes.keys():
protein_complex = self._database.complexes[complex_name]
proteins = [
protein_complex.unbound_formation.ligand,
protein_complex.unbound_formation.receptor
]
for protein in proteins:
stride_x_file = self.__get_dir_name() + protein.name + ".npy"
if not os.path.exists(stride_x_file):
counter += 1
if counter <= 15:
print_info_nn("{0}, ".format(protein.name))
else:
counter = 0
print_info("{0}".format(protein.name))
pdb_file = self._database.directory + pdb_directory + protein.name + ".pdb"
n = len(protein.residues)
stride_x = stride_dict_from_pdb_file(pdb_file)
stride_x_array = np.zeros((n, 11))
for index, residue in enumerate(
protein.biopython_residues):
key = self.get_residue_id(residue.get_full_id())
if key in stride_x:
(_, s, phi, psi, asa, rasa) = stride_x[key]
if s not in secondary_structure_dict:
raise ValueError(
"unknown secondary structure! Add to dictionary!"
)
ss = np.zeros(len(secondary_structure_dict))
ss[secondary_structure_dict[s]] = 1
stride_x_array[index, :7] = ss
stride_x_array[index, 7] = phi
stride_x_array[index, 8] = psi
stride_x_array[index, 9] = asa
stride_x_array[index, 10] = rasa
np.save(stride_x_file, stride_x_array)
stride_x = np.load(stride_x_file)
for i, res in enumerate(protein.residues):
res.add_feature(Features.SECONDARY_STRUCTURE,
stride_x[i, :7])
res.add_feature(Features.PHI, stride_x[i, 7])
res.add_feature(Features.PSI, stride_x[i, 8])
res.add_feature(Features.ACCESSIBLE_SURFACE_AREA,
stride_x[i, 9])
res.add_feature(Features.RELATIVE_ACCESSIBLE_SURFACE_AREA,
stride_x[i, 10])
print_info("took {0} seconds.".format(
(datetime.now() - overall_time).seconds))
def get_rfpp(self):
if self.pyml_result is None:
return None
example_range = self.database.complexes_example_range
example_complex_map = {}
for complex_name in example_range:
interval = example_range[complex_name]
for example in range(interval[0], interval[2]):
example_complex_map[example] = complex_name
example_index_map = self.database.get_pair_index_map()
complex_length_folds = []
for fold in self.pyml_result:
complex_length_map = {}
for i in range(len(fold.L)):
complex_name = self.database.example_complex[fold.patternID[i]]
if complex_name not in complex_length_map:
complex_length_map[complex_name] = 0
complex_length_map[complex_name] += 1
complex_length_folds.append(complex_length_map)
rfpp = {}
fold_no = 0
for fold in self.pyml_result:
complex_performance_map = {}
complex_length_map = complex_length_folds[fold_no]
for i in range(len(fold.L)):
pid = fold.patternID[i]
complex_name = self.database.example_complex[pid]
if complex_name not in complex_performance_map:
example_no = complex_length_map[complex_name]
complex_performance_map[complex_name] = ndarray((example_no, 3))
complex_length_map[complex_name] = 0
perf_table = complex_performance_map[complex_name]
length = complex_length_map[complex_name]
perf_table[length, :] = [int(fold.Y[i]), int(fold.givenY[i]), fold.decisionFunc[i]]
complex_length_map[complex_name] += 1
number_of_examples_in_complex = perf_table.shape[0]
if complex_length_map[complex_name] == number_of_examples_in_complex:
sorted_perf = perf_table[(-perf_table[:, 2]).argsort()]
complex_performance_map[complex_name] = sorted_perf
for complex_name in complex_performance_map:
perf_table = complex_performance_map[complex_name]
for i in range(perf_table.shape[0]):
if perf_table[i, 0] > 0 and perf_table[i, 1] > 0:
rfpp[complex_name] = (i+1, perf_table.shape[0])
break
fold_no += 1
average = 0
for complex_name in rfpp:
rank, n = rfpp[complex_name]
percent = math.ceil((rank * 100) / n)
average += percent
print_info("{0} : {1}".format(complex_name, percent))
print_info("Average RFPP {0}".format(average))
return rfpp
def extract_feature(self):
counter = 0
overall_time = datetime.now()
number_of_amino_acids = len(standard_aa_names)
print_info_nn(" >>> Adding Half Surface Exposure ... ".format(self._database.name))
if not os.path.exists(self._get_dir_name()):
os.makedirs(self._get_dir_name())
for complex_name in self._database.complexes.keys():
protein_complex = self._database.complexes[complex_name]
proteins = [protein_complex.unbound_formation.ligand, protein_complex.unbound_formation.receptor]
for protein in proteins:
hse_file = self._get_dir_name() + protein.name
if not os.path.exists(hse_file + ".npy"):
counter += 1
if counter <= 15:
print_info_nn("{0}, ".format(protein.name))
else:
counter = 0
print_info("{0}".format(protein.name))
number_of_residues = len(protein.biopython_residues)
un = np.zeros(number_of_residues)
dn = np.zeros(number_of_residues)
uc = np.zeros((number_of_amino_acids, number_of_residues))
dc = np.zeros((number_of_amino_acids, number_of_residues))
for index, residue in enumerate(protein.biopython_residues):
u = self.get_side_chain_vector(residue)
if u is None:
un[index] = np.nan
dn[index] = np.nan
uc[:, index] = np.nan
dc[:, index] = np.nan
else:
residue_index = self._residue_index_table[residue.get_resname()]
uc[residue_index, index] += 1
dc[residue_index, index] += 1
neighbours_indices = protein.residues[index].get_feature(Features.RESIDUE_NEIGHBOURHOOD)
# print neighbours_indices
for neighbour_index in neighbours_indices:
if neighbour_index == -1:
break
neighbour_residue = protein.biopython_residues[int(neighbour_index)]
if is_aa(neighbour_residue) and neighbour_residue.has_id('CA'):
neighbour_vector = neighbour_residue['CA'].get_vector()
residue_index = self._residue_index_table[neighbour_residue.get_resname()]
if u[1].angle((neighbour_vector - u[0])) < np.pi / 2.0:
un[index] += 1
uc[residue_index, index] += 1
else:
dn[index] += 1
dc[residue_index, index] += 1
uc = (uc / (1.0 + un)).T
dc = (dc / (1.0 + dn)).T
hse_array = np.hstack((uc, dc))
np.save(hse_file, hse_array)
hse = np.load(hse_file + ".npy")
for i in range(len(protein.residues)):
protein.residues[i].add_feature(Features.HALF_SPHERE_EXPOSURE, hse[i, :])
print_info("took {0} seconds.".format((datetime.now() - overall_time).seconds))
def extract_feature(self):
seed(self.seed)
print_info_nn(
" >>> Adding D1 surface shape distribution for database {0} ... ".
format(self._database.name))
overall_time = datetime.now()
counter = 0
if not os.path.exists(self._get_dir_name()):
os.makedirs(self._get_dir_name())
for complex_name in self._database.complexes.keys():
protein_complex = self._database.complexes[complex_name]
proteins = [
protein_complex.unbound_formation.ligand,
protein_complex.unbound_formation.receptor
]
for protein in proteins:
shape_dist_file = self._get_dir_name() + protein.name
if not os.path.exists(shape_dist_file + ".npy"):
counter += 1
if counter <= 15:
print_info_nn("{0}, ".format(protein.name))
else:
counter = 0
print_info("{0}".format(protein.name))
atoms = protein.atoms
neighbour_search = NeighborSearch(atoms)
distributions = np.zeros(
(len(protein.residues), self.number_of_bins + 1))
for i in range(len(protein.residues)):
residue = protein.residues[i]
nearby_residues = [protein.biopython_residues[i]]
temp_nearby_residues = neighbour_search.search(
residue.center, self.radius, "R")
for nearby_residue in temp_nearby_residues:
if nearby_residue not in protein.biopython_residues:
continue
residues_index = protein.biopython_residues.index(
nearby_residue)
residue = protein.residues[residues_index]
if residue.get_feature(
Features.RELATIVE_ACCESSIBLE_SURFACE_AREA
) >= self.rASA_threshold:
nearby_residues.append(nearby_residue)
distributions[i, :] = self._compute_distribution(
nearby_residues, residue.center)
np.save(shape_dist_file, distributions)
distributions = np.load(shape_dist_file + ".npy")
for i in range(len(protein.residues)):
protein.residues[i].add_feature(
Features.D1_SURFACE_SHAPE_DISTRIBUTION,
distributions[i, :])
print_info("took {0} seconds.".format(
(datetime.now() - overall_time).seconds))
def extract_feature(self):
print_info_nn(
" >>> Adding secondary structure for database {0} ... ".format(
self._database.name))
overall_time = datetime.now()
if not os.path.exists(self.__get_dir_name()):
os.mkdir(self.__get_dir_name())
for complex_name in self._database.complexes.keys():
protein_complex = self._database.complexes[complex_name]
proteins = [
protein_complex.unbound_formation.ligand,
protein_complex.unbound_formation.receptor
]
for protein in proteins:
dssp_file = self.__get_dir_name() + protein.name + ".npy"
if not os.path.exists(dssp_file):
print_info_nn("... running DSSP for protein " +
protein.name)
start_time = datetime.now()
dssp = DSSP(
protein.structure[0], self._database.directory +
pdb_directory + protein.name + ".pdb")
dssp_array = np.ndarray((len(protein.residues), 6))
for (i, res) in enumerate(protein.biopython_residues):
(_, _, cid, rid) = res.get_full_id()
key = (cid, rid)
if key in dssp:
dssp_array[i, 2:] = (dssp[key])[2:]
else:
dssp_array[i, 2:] = [0, 0, 0, 0]
# print_error("WTH")
# sys.exit(0)
# print('here')
# pdb.set_trace()
# self.SS[:, index] = np.nan
# self.ASA[index] = np.nan
# self.rASA[index] = np.nan
# self.Phi[index] = np.nan
# self.Psi[index] = np.nan
np.save(dssp_file, dssp_array)
print_info("took {0} seconds.".format(
(datetime.now() - start_time).seconds))
dssp = np.load(dssp_file)
for i, res in enumerate(protein.residues):
# (_, s, ASA, rASA, phi, psi)
res.add_feature(Features.ACCESSIBLE_SURFACE_AREA, dssp[i,
2])
res.add_feature(Features.RELATIVE_ACCESSIBLE_SURFACE_AREA,
dssp[i, 3])
res.add_feature(Features.PHI, dssp[i, 4])
res.add_feature(Features.PSI, dssp[i, 5])
print_info("took {0} seconds.".format(
(datetime.now() - overall_time).seconds))
def __compute_profiles(self, db='nr', niter=3):
print_info_nn(" >>> Adding the profile features for dataset {0} ...".format(self._database.name))
start_time = datetime.now()
for complex_name in self._database.complexes.keys():
protein_complex = self._database.complexes[complex_name]
proteins = [protein_complex.unbound_formation.ligand, protein_complex.unbound_formation.receptor]
for protein in proteins:
fasta_file = self._database.directory + unbound_sequence_directory + protein.name + ".fasta"
output_file = self._database.directory + pssm_directory + protein.name
if not os.path.exists(output_file + ".mat"):
print_info("... processing protein {0} ... ".format(protein.name))
command = "cd {4} \n " \
"{5} " \
"-query {0} -db {1} -out {2}.psi.txt -num_iterations {3} -out_ascii_pssm {2}.mat" \
.format(fasta_file, db, output_file, niter, psiblast_db_folder, psiblast_executable)
print_info(command)
error_code = os.system(command)
if error_code == 0:
print_info('Successful!')
else:
print_error('Failed with error code {0}'.format(error_code))
pssm, psfm, info = ProfileExtractor.__parse_pssm_file(output_file + ".mat")
wpssm = ProfileExtractor.__get_wpsm(pssm)
wpsfm = ProfileExtractor.__get_wpsm(psfm)
for i, res in enumerate(protein.residues):
res.add_feature(Features.POSITION_SPECIFIC_SCORING_MATRIX, self._normalize(pssm[:, i]))
res.add_feature(Features.POSITION_SPECIFIC_FREQUENCY_MATRIX, self._normalize(psfm[:, i]))
res.add_feature(Features.WINDOWED_POSITION_SPECIFIC_SCORING_MATRIX, self._normalize(wpssm[:, i]))
res.add_feature(Features.WINDOWED_POSITION_SPECIFIC_FREQUENCY_MATRIX, self._normalize(wpsfm[:, i]))
print_info("took {0} seconds.".format((datetime.now() - start_time).seconds))
def __save_sequences_to_fasta(self):
for complex_name in self._database.complexes.keys():
protein_complex = self._database.complexes[complex_name]
proteins = [protein_complex.unbound_formation.ligand, protein_complex.unbound_formation.receptor]
for protein in proteins:
sequence = protein.sequence
fasta_file = self._database.directory + unbound_sequence_directory + protein.name + ".fasta"
if os.path.exists(fasta_file):
continue
print_info("... Saving sequence for protein " + protein.name)
f = open(fasta_file, "w+")
f.write(">{0}\n".format(protein.name))
f.write(sequence + "\n")
f.close()
def __extract_examples(self):
"""
This function returns the set of all positive and negative examples from DBD4 dataset. In protein complex C,
wth receptor R and ligand L, two residues r on R and r' on L are considered as a positive example if in the
bound form they are nearer than the threshold distance. All other pairs (r,r') with r on R and r' on L are
considered as negative examples. Extracted examples are saved in self.examples
"""
print_info("Finding the positive and negative examples in DBD4 ... {0}".format(self.positives_size))
start_time = datetime.now()
counter = 1
start_index = 0
neg_no = 0
pos_no = 0
for complex_name in self.complexes.keys():
print_info_nn("{0}/{1}... processing complex {2}".format(counter, len(self.complexes), complex_name))
protein_complex = self.complexes[complex_name]
bound_ligand_bio_residues = protein_complex.bound_formation.ligand.biopython_residues
bound_receptor_bio_residues = protein_complex.bound_formation.receptor.biopython_residues
bound_ligand_residues = protein_complex.bound_formation.ligand.residues
bound_receptor_residues = protein_complex.bound_formation.receptor.residues
pos = []
neg = []
for i in range(len(bound_ligand_bio_residues)):
for j in range(len(bound_receptor_bio_residues)):
bound_ligand_residue = bound_ligand_bio_residues[i]
bound_receptor_residue = bound_receptor_bio_residues[j]
l_atoms = [atom.get_coord() for atom in bound_ligand_residue.get_list()]
r_atoms = [atom.get_coord() for atom in bound_receptor_residue.get_list()]
dist_mat = cdist(l_atoms, r_atoms)
ligand_b2u = protein_complex.ligand_bound_to_unbound
receptor_b2u = protein_complex.receptor_bound_to_unbound
# if the residues have an unbound counterpart
# this is due to the fact that the unbound and bound formations may have slightly different residues
if bound_ligand_residues[i] in ligand_b2u and bound_receptor_residues[j] in receptor_b2u:
unbound_ligand_res = ligand_b2u[bound_ligand_residues[i]]
unbound_receptor_res = receptor_b2u[bound_receptor_residues[j]]
unbound_ligand_res_index = self.__get_residue_index(unbound_ligand_res)
unbound_receptor_res_index = self.__get_residue_index(unbound_receptor_res)
if dist_mat.min() < self.interaction_threshold:
pos.append((unbound_ligand_res_index, unbound_receptor_res_index, +1))
else:
neg.append((unbound_ligand_res_index, unbound_receptor_res_index, -1))
self.examples.extend(copy.copy(pos))
self.examples.extend(copy.copy(neg))
pos_no += len(pos)
neg_no += len(neg)
self.complexes_example_range[complex_name] = (
start_index, start_index + len(pos), start_index + len(neg) + len(pos))
print_info(" ( {0:03d}/{1:05d} ) -{2}".format(len(pos), len(neg), self.complexes_example_range[complex_name]))
start_index += len(pos) + len(neg)
counter += 1
all_e = pos + neg
for e in all_e:
self.example_complex["{0}_{1}".format(e[0], e[1])] = complex_name
print_info("Finding examples in DBD4 took " + str((datetime.now() - start_time).seconds) + " seconds. ")
print_info("The total number of examples found: " + str(pos_no + neg_no))
def extract_feature(self):
secondary_structure_dict = dict(zip(ss_abbreviations, range(len(ss_abbreviations))))
print_info_nn(" >>> Adding secondary structure for database {0} ... ".format(self._database.name))
overall_time = datetime.now()
counter = 0
if not os.path.exists(self.__get_dir_name()):
os.mkdir(self.__get_dir_name())
for complex_name in self._database.complexes.keys():
protein_complex = self._database.complexes[complex_name]
proteins = [protein_complex.unbound_formation.ligand, protein_complex.unbound_formation.receptor]
for protein in proteins:
stride_x_file = self.__get_dir_name() + protein.name + ".npy"
if not os.path.exists(stride_x_file):
counter += 1
if counter <= 15:
print_info_nn("{0}, ".format(protein.name))
else:
counter = 0
print_info("{0}".format(protein.name))
pdb_file = self._database.directory + pdb_directory + protein.name + ".pdb"
n = len(protein.residues)
stride_x = stride_dict_from_pdb_file(pdb_file)
stride_x_array = np.zeros((n, 11))
for index, residue in enumerate(protein.biopython_residues):
key = self.get_residue_id(residue.get_full_id())
if key in stride_x:
(_, s, phi, psi, asa, rasa) = stride_x[key]
if s not in secondary_structure_dict:
raise ValueError("unknown secondary structure! Add to dictionary!")
ss = np.zeros(len(secondary_structure_dict))
ss[secondary_structure_dict[s]] = 1
stride_x_array[index, :7] = ss
stride_x_array[index, 7] = phi
stride_x_array[index, 8] = psi
stride_x_array[index, 9] = asa
stride_x_array[index, 10] = rasa
np.save(stride_x_file, stride_x_array)
stride_x = np.load(stride_x_file)
for i, res in enumerate(protein.residues):
res.add_feature(Features.SECONDARY_STRUCTURE, stride_x[i, :7])
res.add_feature(Features.PHI, stride_x[i, 7])
res.add_feature(Features.PSI, stride_x[i, 8])
res.add_feature(Features.ACCESSIBLE_SURFACE_AREA, stride_x[i, 9])
res.add_feature(Features.RELATIVE_ACCESSIBLE_SURFACE_AREA, stride_x[i, 10])
print_info("took {0} seconds.".format((datetime.now() - overall_time).seconds))
def extract_feature(self):
print_info_nn(" >>> Adding residue depth for database {0} ... ".format(self._database.name))
overall_time = datetime.now()
counter = 0
if not os.path.exists(self._get_dir_name()):
os.makedirs(self._get_dir_name())
for complex_name in self._database.complexes.keys():
protein_complex = self._database.complexes[complex_name]
proteins = [protein_complex.unbound_formation.ligand, protein_complex.unbound_formation.receptor]
for protein in proteins:
residue_depth_file = self._get_dir_name() + protein.name + ".npy"
if not os.path.exists(residue_depth_file):
counter += 1
if counter <= 15:
print_info_nn("{0}, ".format(protein.name))
else:
counter = 0
print_info("{0}".format(protein.name))
pdb_file = self._database.directory + pdb_directory + protein.name + ".pdb"
rd = ResidueDepth(protein.structure[0], pdb_file)
rd_array = np.ndarray((len(protein.residues), 2)) # self.number_of_bins +
# surface = get_surface(pdb_file)
for (i, res) in enumerate(protein.biopython_residues):
(_, _, c, (h, rn, ic)) = res.get_full_id()
key = (c, (h, rn, ic))
if key in rd:
rdv = rd[key]
if rdv[0] is None:
rdv = (0, rdv[1])
print "WTH?"
if rdv[1] is None:
rdv = (rdv[0], 0)
print "WTH?"
rd_array[i, :2] = rdv
else:
print_error('WTH')
rd_array[i, :2] = [0, 0]
# rd_array[i, 2:] = self._compute_distribution_(surface, protein.residues[i].center)
np.save(residue_depth_file, rd_array)
surface_features = np.load(residue_depth_file)
for i, res in enumerate(protein.residues):
res.add_feature(Features.RESIDUE_DEPTH, self._normalize(surface_features[i, :2]))
print_info("took {0} seconds.".format((datetime.now() - overall_time).seconds))
def extract_feature(self):
seed(self.seed)
counter = 0
overall_time = datetime.now()
print_info_nn(
" >>> Adding D2 shape distribution for database {0} ... ".format(
self._database.name))
if not os.path.exists(self._get_dir_name()):
os.makedirs(self._get_dir_name())
for complex_name in self._database.complexes.keys():
protein_complex = self._database.complexes[complex_name]
proteins = [
protein_complex.unbound_formation.ligand,
protein_complex.unbound_formation.receptor
]
for protein in proteins:
shape_dist_file = self._get_dir_name() + protein.name
if not os.path.exists(shape_dist_file + ".npy"):
counter += 1
if counter <= 15:
print_info_nn("{0}, ".format(protein.name))
else:
counter = 0
print_info("{0}".format(protein.name))
atoms = protein.atoms
neighbour_search = NeighborSearch(atoms)
distributions = np.zeros(
(len(protein.residues), self.number_of_bins))
# distributions = np.zeros((len(protein.residues), self.number_of_bins+2))
for i in range(len(protein.residues)):
residue = protein.residues[i]
nearby_residues = neighbour_search.search(
residue.center, self.radius, "R")
distributions[i, :] = self._compute_distribution(
nearby_residues)
# distributions[i:, -1] = len(nearby_residues)
np.save(shape_dist_file, distributions)
distributions = np.load(shape_dist_file + ".npy")
for i in range(len(protein.residues)):
protein.residues[i].add_feature(
Features.D2_PLAIN_SHAPE_DISTRIBUTION,
distributions[i, :])
# protein.residues[i].add_feature(Features.NUMBER_OF_NEIGHBOURS, distributions[i, -1])
print_info("took {0} seconds.".format(
(datetime.now() - overall_time).seconds))
def extract_feature(self):
counter = 0
print_info_nn(" >>> Adding Residue Neighbourhood ... ")
overall_time = datetime.now()
if not os.path.exists(self._get_dir_name()):
os.makedirs(self._get_dir_name())
for complex_name in self._database.complexes.keys():
protein_complex = self._database.complexes[complex_name]
proteins = [protein_complex.unbound_formation.ligand, protein_complex.unbound_formation.receptor]
for protein in proteins:
residue_neighbourhood_file = self._get_dir_name() + protein.name
if not os.path.exists(residue_neighbourhood_file + ".npy"):
counter += 1
if counter <= 15:
print_info_nn("{0}, ".format(protein.name))
else:
counter = 0
print_info("{0}".format(protein.name))
neighbourhood = []
max_length = 0
for i, query_residue in enumerate(protein.residues):
neighbourhood.append([])
for j, neighbour_residue in enumerate(protein.residues):
# if i == j:
# continue
distance = cdist(query_residue.get_coordinates(), neighbour_residue.get_coordinates()).min()
similarity = np.exp(-(distance ** 2) / self._sigma)
if distance <= 7.5:
neighbourhood[-1].append(j)
if len(neighbourhood[-1]) > max_length:
max_length = len(neighbourhood[-1])
neighbourhood_array = -np.ones((len(protein.residues), max_length))
# print len(neighbourhood)
for i, residue_neighbourhood in enumerate(neighbourhood):
for j, neighbour_index in enumerate(neighbourhood[i]):
neighbourhood_array[i, j] = neighbourhood[i][j]
# print neighbourhood_array[i, :]
np.save(residue_neighbourhood_file, neighbourhood_array)
neighbourhood_array = np.load(residue_neighbourhood_file+".npy")
for index, residue in enumerate(protein.residues):
residue.add_feature(Features.RESIDUE_NEIGHBOURHOOD, neighbourhood_array[index, :])
print_info("took {0} seconds.".format((datetime.now() - overall_time).seconds))
def extract_feature(self):
print_info_nn(" >>> Adding secondary structure for database {0} ... ".format(self._database.name))
overall_time = datetime.now()
if not os.path.exists(self.__get_dir_name()):
os.mkdir(self.__get_dir_name())
for complex_name in self._database.complexes.keys():
protein_complex = self._database.complexes[complex_name]
proteins = [protein_complex.unbound_formation.ligand, protein_complex.unbound_formation.receptor]
for protein in proteins:
dssp_file = self.__get_dir_name() + protein.name + ".npy"
if not os.path.exists(dssp_file):
print_info_nn("... running DSSP for protein " + protein.name)
start_time = datetime.now()
dssp = DSSP(protein.structure[0], self._database.directory + pdb_directory + protein.name + ".pdb")
dssp_array = np.ndarray((len(protein.residues), 6))
for (i, res) in enumerate(protein.biopython_residues):
(_, _, cid, rid) = res.get_full_id()
key = (cid, rid)
if key in dssp:
dssp_array[i, 2:] = (dssp[key])[2:]
else:
dssp_array[i, 2:] = [0, 0, 0, 0]
# print_error("WTH")
# sys.exit(0)
# print('here')
# pdb.set_trace()
# self.SS[:, index] = np.nan
# self.ASA[index] = np.nan
# self.rASA[index] = np.nan
# self.Phi[index] = np.nan
# self.Psi[index] = np.nan
np.save(dssp_file, dssp_array)
print_info("took {0} seconds.".format((datetime.now() - start_time).seconds))
dssp = np.load(dssp_file)
for i, res in enumerate(protein.residues):
# (_, s, ASA, rASA, phi, psi)
res.add_feature(Features.ACCESSIBLE_SURFACE_AREA, dssp[i, 2])
res.add_feature(Features.RELATIVE_ACCESSIBLE_SURFACE_AREA, dssp[i, 3])
res.add_feature(Features.PHI, dssp[i, 4])
res.add_feature(Features.PSI, dssp[i, 5])
print_info("took {0} seconds.".format((datetime.now() - overall_time).seconds))
def __read_pdb_files(self):
print_info("Parsing the pdb files in directory {0} ....".format(
os.path.abspath(self.directory)))
ligand_bound_files = glob.glob(self.directory + pdb_directory +
"*_l_b.pdb")
ligand_bound_files.sort()
counter = 0
for ligand_bound_file in ligand_bound_files:
complex_name = basename(ligand_bound_file).replace("_l_b.pdb", "")
receptor_bound_file = ligand_bound_file.replace(
"_l_b.pdb", "_r_b.pdb")
ligand_unbound_file = ligand_bound_file.replace(
"_l_b.pdb", "_l_u.pdb")
receptor_unbound_file = ligand_bound_file.replace(
"_l_b.pdb", "_r_u.pdb")
print_info("Reading complex " + complex_name)
with warnings.catch_warnings():
warnings.simplefilter("ignore", PDBConstructionWarning)
ligand_bound = Protein(*read_pdb_file(ligand_bound_file))
receptor_bound = Protein(*read_pdb_file(receptor_bound_file))
ligand_unbound = Protein(*read_pdb_file(ligand_unbound_file))
receptor_unbound = Protein(
*read_pdb_file(receptor_unbound_file))
bound_formation = ProteinPair(ligand_bound, receptor_bound)
unbound_formation = ProteinPair(ligand_unbound,
receptor_unbound)
self.complexes[complex_name] = ProteinComplex(
complex_name, unbound_formation, bound_formation)
counter += 1
print_info("Total number of complexes processed : " + str(counter))
def extract_feature(self):
seed(self.seed)
print_info_nn(" >>> Adding D1 surface shape distribution for database {0} ... ".format(self._database.name))
overall_time = datetime.now()
counter = 0
if not os.path.exists(self._get_dir_name()):
os.makedirs(self._get_dir_name())
for complex_name in self._database.complexes.keys():
protein_complex = self._database.complexes[complex_name]
proteins = [protein_complex.unbound_formation.ligand, protein_complex.unbound_formation.receptor]
for protein in proteins:
shape_dist_file = self._get_dir_name() + protein.name
if not os.path.exists(shape_dist_file + ".npy"):
counter += 1
if counter <= 15:
print_info_nn("{0}, ".format(protein.name))
else:
counter = 0
print_info("{0}".format(protein.name))
atoms = protein.atoms
neighbour_search = NeighborSearch(atoms)
distributions = np.zeros((len(protein.residues), self.number_of_bins + 1))
for i in range(len(protein.residues)):
residue = protein.residues[i]
nearby_residues = [protein.biopython_residues[i]]
temp_nearby_residues = neighbour_search.search(residue.center, self.radius, "R")
for nearby_residue in temp_nearby_residues:
if nearby_residue not in protein.biopython_residues:
continue
residues_index = protein.biopython_residues.index(nearby_residue)
residue = protein.residues[residues_index]
if residue.get_feature(Features.RELATIVE_ACCESSIBLE_SURFACE_AREA) >= self.rASA_threshold:
nearby_residues.append(nearby_residue)
distributions[i, :] = self._compute_distribution(nearby_residues, residue.center)
np.save(shape_dist_file, distributions)
distributions = np.load(shape_dist_file + ".npy")
for i in range(len(protein.residues)):
protein.residues[i].add_feature(Features.D1_SURFACE_SHAPE_DISTRIBUTION, distributions[i, :])
print_info("took {0} seconds.".format((datetime.now() - overall_time).seconds))
def _load(self, file_name=None):
"""
This function load all the attributes of the class: positive and negative examples, ligands and receptors and
complex names are saved in pickle format.
"""
if file_name is None:
object_model_file_name = self.directory + pickle_directory + dbd4_object_model_file
else:
object_model_file_name = file_name
f = open(object_model_file_name)
print_info_nn("Loading the object model from {0} ... ".format(
object_model_file_name))
start_time = datetime.now()
(self.directory, self.complexes, self.residues,
self.complexes_example_range, self.examples,
self.example_complex) = cPickle.load(f)
f.close()
gc.collect()
print_info("took {0} seconds.".format(
(datetime.now() - start_time).seconds))
def _save(self, file_name=None):
"""
This function saves all the attributes of the class: positive and negative examples, ligands and receptors and
complex names are saved in pickle format.
"""
if not os.path.exists(self.directory + pickle_directory):
os.mkdir(self.directory + pickle_directory)
if file_name is None:
object_model_file_name = self.directory + pickle_directory + dbd4_object_model_file
else:
object_model_file_name = file_name
f = open(object_model_file_name, "wb")
print_info_nn("Saving the object model into {0} ... ".format(
object_model_file_name))
start_time = datetime.now()
cPickle.dump((self.directory, self.complexes, self.residues,
self.complexes_example_range, self.examples,
self.example_complex), f)
f.close()
print_info("took {0} seconds.".format(
(datetime.now() - start_time).seconds))
def _load(self, file_name=None):
"""
This function load all the attributes of the class: positive and negative examples, ligands and receptors and
complex names are saved in pickle format.
"""
if file_name is None:
object_model_file_name = self.directory + pickle_directory + dbd4_object_model_file
else:
object_model_file_name = file_name
f = open(object_model_file_name)
print_info_nn("Loading the object model from {0} ... ".format(object_model_file_name))
start_time = datetime.now()
(self.directory,
self.complexes,
self.residues,
self.complexes_example_range,
self.examples,
self.example_complex) = cPickle.load(f)
f.close()
gc.collect()
print_info("took {0} seconds.".format((datetime.now() - start_time).seconds))
def _save(self, file_name=None):
"""
This function saves all the attributes of the class: positive and negative examples, ligands and receptors and
complex names are saved in pickle format.
"""
if not os.path.exists(self.directory + pickle_directory):
os.mkdir(self.directory + pickle_directory)
if file_name is None:
object_model_file_name = self.directory + pickle_directory + dbd4_object_model_file
else:
object_model_file_name = file_name
f = open(object_model_file_name, "wb")
print_info_nn("Saving the object model into {0} ... ".format(object_model_file_name))
start_time = datetime.now()
cPickle.dump((self.directory,
self.complexes,
self.residues,
self.complexes_example_range,
self.examples,
self.example_complex), f)
f.close()
print_info("took {0} seconds.".format((datetime.now() - start_time).seconds))
def extract_feature(self):
counter = 0
overall_time = datetime.now()
print_info_nn(" >>> Adding B Factor ... ".format(self._database.name))
if not os.path.exists(self._get_dir_name()):
os.makedirs(self._get_dir_name())
for complex_name in self._database.complexes.keys():
protein_complex = self._database.complexes[complex_name]
proteins = [
protein_complex.unbound_formation.ligand,
protein_complex.unbound_formation.receptor
]
for protein in proteins:
b_factor_filename = self._get_dir_name() + protein.name
if not os.path.exists(b_factor_filename + ".npy"):
counter += 1
if counter <= 15:
print_info_nn("{0}, ".format(protein.name))
else:
counter = 0
print_info("{0}".format(protein.name))
b_factor_array = np.zeros(len(protein.residues))
for (index,
residue) in enumerate(protein.biopython_residues):
b_factor_array[index] = max(
[atom.get_bfactor() for atom in residue])
np.save(b_factor_filename, b_factor_array)
b_factor_array = np.load(b_factor_filename + ".npy")
# print b_factor_array
for i in range(len(protein.residues)):
protein.residues[i].add_feature(Features.B_VALUE,
b_factor_array[i])
print_info("took {0} seconds.".format(
(datetime.now() - overall_time).seconds))
def __read_pdb_files(self):
print_info("Parsing the pdb files in directory {0} ....".format(os.path.abspath(self.directory)))
ligand_bound_files = glob.glob(self.directory + pdb_directory + "*_l_b.pdb")
ligand_bound_files.sort()
counter = 0
for ligand_bound_file in ligand_bound_files:
complex_name = basename(ligand_bound_file).replace("_l_b.pdb", "")
receptor_bound_file = ligand_bound_file.replace("_l_b.pdb", "_r_b.pdb")
ligand_unbound_file = ligand_bound_file.replace("_l_b.pdb", "_l_u.pdb")
receptor_unbound_file = ligand_bound_file.replace("_l_b.pdb", "_r_u.pdb")
print_info("Reading complex " + complex_name)
with warnings.catch_warnings():
warnings.simplefilter("ignore", PDBConstructionWarning)
ligand_bound = Protein(*read_pdb_file(ligand_bound_file))
receptor_bound = Protein(*read_pdb_file(receptor_bound_file))
ligand_unbound = Protein(*read_pdb_file(ligand_unbound_file))
receptor_unbound = Protein(*read_pdb_file(receptor_unbound_file))
bound_formation = ProteinPair(ligand_bound, receptor_bound)
unbound_formation = ProteinPair(ligand_unbound, receptor_unbound)
self.complexes[complex_name] = ProteinComplex(complex_name, unbound_formation, bound_formation)
counter += 1
print_info("Total number of complexes processed : " + str(counter))
def main():
print_info("Starting the experiment")
start_time = datetime.now()
seed = 1
#number_of_samples = 5000
number_of_samples = 20000
dbd4 = DBD4(size=number_of_samples, ratio=-1, seed=seed)
mtrand.seed(seed)
feature_sets = [
#[
# Features.WINDOWED_POSITION_SPECIFIC_SCORING_MATRIX,
# Features.WINDOWED_POSITION_SPECIFIC_FREQUENCY_MATRIX,
#],
# [
# Features.WINDOWED_POSITION_SPECIFIC_SCORING_MATRIX,
# Features.D2_PLAIN_SHAPE_DISTRIBUTION
# ],
# [
# Features.WINDOWED_POSITION_SPECIFIC_SCORING_MATRIX,
# Features.D1_PLAIN_SHAPE_DISTRIBUTION
# ],
# [
# Features.WINDOWED_POSITION_SPECIFIC_SCORING_MATRIX,
# Features.RELATIVE_ACCESSIBLE_SURFACE_AREA,
# Features.D2_SURFACE_SHAPE_DISTRIBUTION
# ],
[
Features.WINDOWED_POSITION_SPECIFIC_SCORING_MATRIX,
Features.RELATIVE_ACCESSIBLE_SURFACE_AREA,
Features.D1_SURFACE_SHAPE_DISTRIBUTION
],
# [
# Features.WINDOWED_POSITION_SPECIFIC_SCORING_MATRIX,
# Features.D2_CATEGORY_SHAPE_DISTRIBUTION
# ],
# [
# Features.WINDOWED_POSITION_SPECIFIC_SCORING_MATRIX,
# # Features.PROTRUSION_INDEX,
# # Features.B_VALUE,
# Features.HALF_SPHERE_EXPOSURE,
# Features.SECONDARY_STRUCTURE,
# Features.WINDOWED_POSITION_SPECIFIC_FREQUENCY_MATRIX,
# Features.POSITION_SPECIFIC_SCORING_MATRIX,
# Features.POSITION_SPECIFIC_FREQUENCY_MATRIX,
# Features.RELATIVE_ACCESSIBLE_SURFACE_AREA,
# # # Features.PHI,
# # # Features.PSI,
# # Features.RELATIVE_ACCESSIBLE_SURFACE_AREA,
# Features.D2_SURFACE_SHAPE_DISTRIBUTION,
# # Features.D1_SURFACE_SHAPE_DISTRIBUTION,
# # Features.D2_PLAIN_SHAPE_DISTRIBUTION,
# # Features.D1_SURFACE_SHAPE_DISTRIBUTION,
# Features.RESIDUE_DEPTH
# ]
]
results = []
for feature_set in feature_sets:
print_special("Feature set {0}".format(feature_set))
e = Experiment(feature_set, dbd4, Classifier.SVM)
e.run(number_of_bins=20, radius=15, number_of_samples=-1, seed=seed, gamma=0.5, save=True, folds=5, rASA=.5)
results.append(e.pyml_result)
print_info("Took {0} seconds.".format((datetime.now() - start_time).seconds))
save_results(number_of_samples, results, feature_sets)
def __extract_examples(self):
"""
This function returns the set of all positive and negative examples from DBD4 dataset. In protein complex C,
wth receptor R and ligand L, two residues r on R and r' on L are considered as a positive example if in the
bound form they are nearer than the threshold distance. All other pairs (r,r') with r on R and r' on L are
considered as negative examples. Extracted examples are saved in self.examples
"""
print_info(
"Finding the positive and negative examples in DBD4 ... {0}".
format(self.positives_size))
start_time = datetime.now()
counter = 1
start_index = 0
neg_no = 0
pos_no = 0
for complex_name in self.complexes.keys():
print_info_nn("{0}/{1}... processing complex {2}".format(
counter, len(self.complexes), complex_name))
protein_complex = self.complexes[complex_name]
bound_ligand_bio_residues = protein_complex.bound_formation.ligand.biopython_residues
bound_receptor_bio_residues = protein_complex.bound_formation.receptor.biopython_residues
bound_ligand_residues = protein_complex.bound_formation.ligand.residues
bound_receptor_residues = protein_complex.bound_formation.receptor.residues
pos = []
neg = []
for i in range(len(bound_ligand_bio_residues)):
for j in range(len(bound_receptor_bio_residues)):
bound_ligand_residue = bound_ligand_bio_residues[i]
bound_receptor_residue = bound_receptor_bio_residues[j]
l_atoms = [
atom.get_coord()
for atom in bound_ligand_residue.get_list()
]
r_atoms = [
atom.get_coord()
for atom in bound_receptor_residue.get_list()
]
dist_mat = cdist(l_atoms, r_atoms)
ligand_b2u = protein_complex.ligand_bound_to_unbound
receptor_b2u = protein_complex.receptor_bound_to_unbound
# if the residues have an unbound counterpart
# this is due to the fact that the unbound and bound formations may have slightly different residues
if bound_ligand_residues[
i] in ligand_b2u and bound_receptor_residues[
j] in receptor_b2u:
unbound_ligand_res = ligand_b2u[
bound_ligand_residues[i]]
unbound_receptor_res = receptor_b2u[
bound_receptor_residues[j]]
unbound_ligand_res_index = self.__get_residue_index(
unbound_ligand_res)
unbound_receptor_res_index = self.__get_residue_index(
unbound_receptor_res)
if dist_mat.min() < self.interaction_threshold:
pos.append((unbound_ligand_res_index,
unbound_receptor_res_index, +1))
else:
neg.append((unbound_ligand_res_index,
unbound_receptor_res_index, -1))
self.examples.extend(copy.copy(pos))
self.examples.extend(copy.copy(neg))
pos_no += len(pos)
neg_no += len(neg)
self.complexes_example_range[complex_name] = (start_index,
start_index +
len(pos),
start_index +
len(neg) + len(pos))
print_info(" ( {0:03d}/{1:05d} ) -{2}".format(
len(pos), len(neg),
self.complexes_example_range[complex_name]))
start_index += len(pos) + len(neg)
counter += 1
all_e = pos + neg
for e in all_e:
self.example_complex["{0}_{1}".format(e[0],
e[1])] = complex_name
print_info("Finding examples in DBD4 took " +
str((datetime.now() - start_time).seconds) + " seconds. ")
print_info("The total number of examples found: " +
str(pos_no + neg_no))
def main():
seed = 1
number_of_samples = 20000
dbd4 = DBD4(size=number_of_samples, ratio=1, thresh=6, seed=seed)
SecondaryStructureExtractor(dbd4).extract_feature()
for complex_name in dbd4.complexes:
print_info(complex_name)
c = dbd4.complexes[complex_name]
b_ligand, b_receptor = (c.bound_formation.ligand,
c.bound_formation.receptor)
u_ligand, u_receptor = (c.unbound_formation.ligand,
c.unbound_formation.receptor)
b_ligand_bio_residues, b_receptor_bio_residues = b_ligand.biopython_residues, b_receptor.biopython_residues
# b_l_ns, b_r_ns = NeighborSearch(b_ligand.atoms), NeighborSearch(b_receptor.atoms)
u_l_ns, u_r_ns = NeighborSearch(u_ligand.atoms), NeighborSearch(
u_receptor.atoms)
index = randint(1, 30)
positives = 0
negatives = 0
lb2u = c.ligand_bound_to_unbound
rb2u = c.receptor_bound_to_unbound
p = False
n = False
with PdfPages('{0}/geometry/figures/{1}.pdf'.format(
reports_directory, complex_name)) as pdf:
try:
for i in range(len(b_ligand_bio_residues)):
for j in range(len(b_receptor_bio_residues)):
if b_ligand.residues[
i] not in lb2u or b_receptor.residues[
j] not in rb2u:
continue
l_atoms = b_ligand_bio_residues[i].get_list()
r_atoms = b_receptor_bio_residues[j].get_list()
dist_mat = cdist(
[atom.get_coord() for atom in l_atoms],
[atom.get_coord() for atom in r_atoms])
if p and n:
print "getting out of loop..."
raise GetOutOfLoop
if dist_mat.min() < dbd4.interaction_threshold:
if p:
continue
positives += 1
if positives != index:
continue
# b_l_points, b_l_dist = get_coords(b_l_ns, b_ligand, b_ligand.residues[i], 0.5, False)
# b_r_points, b_r_dist = get_coords(b_r_ns, b_receptor, b_receptor.residues[j], 0.5, False)
b_l_points, b_l_dist = get_coords(
u_l_ns, u_ligand, lb2u[b_ligand.residues[i]],
0.5, False)
b_r_points, b_r_dist = get_coords(
u_r_ns, u_receptor,
rb2u[b_receptor.residues[j]], 0.5, False)
u_l_points, u_l_dist = get_coords(
u_l_ns, u_ligand, lb2u[b_ligand.residues[i]],
0.5, True)
u_r_points, u_r_dist = get_coords(
u_r_ns, u_receptor,
rb2u[b_receptor.residues[j]], 0.5, True)
fig = plt.figure()
ax = fig.add_subplot(111, projection='3d')
ax.scatter(b_l_points[:, 0],
b_l_points[:, 1],
b_l_points[:, 2],
c='r')
ax.scatter(b_r_points[:, 0],
b_r_points[:, 1],
b_r_points[:, 2],
c='b')
plt.title(
"Interacting Residues Bound Conformation")
pdf.savefig()
plt.close()
fig = plt.figure()
ax = fig.add_subplot(111, projection='3d')
ax.scatter(u_l_points[:, 0],
u_l_points[:, 1],
u_l_points[:, 2],
c='r')
ax.scatter(u_r_points[:, 0],
u_r_points[:, 1],
u_r_points[:, 2],
c='b')
plt.title(
"Interacting Surface Residues Bound Conformation"
)
pdf.savefig()
plt.close()
plt.figure()
plt.plot(u_l_dist)
plt.plot(u_r_dist)
plt.legend([
"bound ligand {0}".format(i),
"bound receptor {0}".format(j),
"unbound ligand", "unbound receptor"
])
pdf.savefig()
plt.close()
p = True
else:
if n:
continue
lb2u = c.ligand_bound_to_unbound
rb2u = c.receptor_bound_to_unbound
b_l_points, b_l_dist = get_coords(
u_l_ns, u_ligand, lb2u[b_ligand.residues[i]],
0.5, False)
b_r_points, b_r_dist = get_coords(
u_r_ns, u_receptor,
rb2u[b_receptor.residues[j]], 0.5, False)
u_l_points, u_l_dist = get_coords(
u_l_ns, u_ligand, lb2u[b_ligand.residues[i]],
0.5, True)
u_r_points, u_r_dist = get_coords(
u_r_ns, u_receptor,
rb2u[b_receptor.residues[j]], 0.5, True)
fig = plt.figure()
ax = fig.add_subplot(111, projection='3d')
ax.scatter(b_l_points[:, 0],
b_l_points[:, 1],
b_l_points[:, 2],
c='r')
ax.scatter(b_r_points[:, 0],
b_r_points[:, 1],
b_r_points[:, 2],
c='b')
plt.title(
"Non-Interacting Residues Bound Conformation")
pdf.savefig()
plt.close()
fig = plt.figure()
ax = fig.add_subplot(111, projection='3d')
ax.scatter(u_l_points[:, 0],
u_l_points[:, 1],
u_l_points[:, 2],
c='r')
ax.scatter(u_r_points[:, 0],
u_r_points[:, 1],
u_r_points[:, 2],
c='b')
plt.title(
"Non-Interacting Surface Residues Bound Conformation"
)
pdf.savefig()
plt.close()
plt.figure()
plt.plot(u_l_dist)
plt.plot(u_r_dist)
plt.legend([
"bound ligand {0}".format(i),
"bound receptor {0}".format(j),
"unbound ligand", "unbound receptor"
])
pdf.savefig()
plt.close()
n = True
except GetOutOfLoop:
pass
def extract_feature(self):
counter = 0
overall_time = datetime.now()
number_of_amino_acids = len(standard_aa_names)
print_info_nn(" >>> Adding Half Surface Exposure ... ".format(
self._database.name))
if not os.path.exists(self._get_dir_name()):
os.makedirs(self._get_dir_name())
for complex_name in self._database.complexes.keys():
protein_complex = self._database.complexes[complex_name]
proteins = [
protein_complex.unbound_formation.ligand,
protein_complex.unbound_formation.receptor
]
for protein in proteins:
hse_file = self._get_dir_name() + protein.name
if not os.path.exists(hse_file + ".npy"):
counter += 1
if counter <= 15:
print_info_nn("{0}, ".format(protein.name))
else:
counter = 0
print_info("{0}".format(protein.name))
number_of_residues = len(protein.biopython_residues)
un = np.zeros(number_of_residues)
dn = np.zeros(number_of_residues)
uc = np.zeros((number_of_amino_acids, number_of_residues))
dc = np.zeros((number_of_amino_acids, number_of_residues))
for index, residue in enumerate(
protein.biopython_residues):
u = self.get_side_chain_vector(residue)
if u is None:
un[index] = np.nan
dn[index] = np.nan
uc[:, index] = np.nan
dc[:, index] = np.nan
else:
residue_index = self._residue_index_table[
residue.get_resname()]
uc[residue_index, index] += 1
dc[residue_index, index] += 1
neighbours_indices = protein.residues[
index].get_feature(
Features.RESIDUE_NEIGHBOURHOOD)
# print neighbours_indices
for neighbour_index in neighbours_indices:
if neighbour_index == -1:
break
neighbour_residue = protein.biopython_residues[
int(neighbour_index)]
if is_aa(neighbour_residue
) and neighbour_residue.has_id('CA'):
neighbour_vector = neighbour_residue[
'CA'].get_vector()
residue_index = self._residue_index_table[
neighbour_residue.get_resname()]
if u[1].angle((neighbour_vector -
u[0])) < np.pi / 2.0:
un[index] += 1
uc[residue_index, index] += 1
else:
dn[index] += 1
dc[residue_index, index] += 1
uc = (uc / (1.0 + un)).T
dc = (dc / (1.0 + dn)).T
hse_array = np.hstack((uc, dc))
np.save(hse_file, hse_array)
hse = np.load(hse_file + ".npy")
for i in range(len(protein.residues)):
protein.residues[i].add_feature(
Features.HALF_SPHERE_EXPOSURE, hse[i, :])
print_info("took {0} seconds.".format(
(datetime.now() - overall_time).seconds))
def main():
seed = 1
number_of_samples = 20000
dbd4 = DBD4(size=number_of_samples, ratio=1, thresh=6, seed=seed)
SecondaryStructureExtractor(dbd4).extract_feature()
for complex_name in dbd4.complexes:
print_info(complex_name)
c = dbd4.complexes[complex_name]
b_ligand, b_receptor = (c.bound_formation.ligand, c.bound_formation.receptor)
u_ligand, u_receptor = (c.unbound_formation.ligand, c.unbound_formation.receptor)
b_ligand_bio_residues, b_receptor_bio_residues = b_ligand.biopython_residues, b_receptor.biopython_residues
# b_l_ns, b_r_ns = NeighborSearch(b_ligand.atoms), NeighborSearch(b_receptor.atoms)
u_l_ns, u_r_ns = NeighborSearch(u_ligand.atoms), NeighborSearch(u_receptor.atoms)
index = randint(1, 30)
positives = 0
negatives = 0
lb2u = c.ligand_bound_to_unbound
rb2u = c.receptor_bound_to_unbound
p = False
n = False
with PdfPages('{0}/geometry/figures/{1}.pdf'.format(reports_directory, complex_name)) as pdf:
try:
for i in range(len(b_ligand_bio_residues)):
for j in range(len(b_receptor_bio_residues)):
if b_ligand.residues[i] not in lb2u or b_receptor.residues[j] not in rb2u:
continue
l_atoms = b_ligand_bio_residues[i].get_list()
r_atoms = b_receptor_bio_residues[j].get_list()
dist_mat = cdist([atom.get_coord() for atom in l_atoms], [atom.get_coord() for atom in r_atoms])
if p and n:
print "getting out of loop..."
raise GetOutOfLoop
if dist_mat.min() < dbd4.interaction_threshold:
if p:
continue
positives += 1
if positives != index:
continue
# b_l_points, b_l_dist = get_coords(b_l_ns, b_ligand, b_ligand.residues[i], 0.5, False)
# b_r_points, b_r_dist = get_coords(b_r_ns, b_receptor, b_receptor.residues[j], 0.5, False)
b_l_points, b_l_dist = get_coords(u_l_ns, u_ligand, lb2u[b_ligand.residues[i]], 0.5, False)
b_r_points, b_r_dist = get_coords(u_r_ns, u_receptor, rb2u[b_receptor.residues[j]], 0.5,
False)
u_l_points, u_l_dist = get_coords(u_l_ns, u_ligand, lb2u[b_ligand.residues[i]], 0.5, True)
u_r_points, u_r_dist = get_coords(u_r_ns, u_receptor, rb2u[b_receptor.residues[j]], 0.5,
True)
fig = plt.figure()
ax = fig.add_subplot(111, projection='3d')
ax.scatter(b_l_points[:, 0], b_l_points[:, 1], b_l_points[:, 2], c='r')
ax.scatter(b_r_points[:, 0], b_r_points[:, 1], b_r_points[:, 2], c='b')
plt.title("Interacting Residues Bound Conformation")
pdf.savefig()
plt.close()
fig = plt.figure()
ax = fig.add_subplot(111, projection='3d')
ax.scatter(u_l_points[:, 0], u_l_points[:, 1], u_l_points[:, 2], c='r')
ax.scatter(u_r_points[:, 0], u_r_points[:, 1], u_r_points[:, 2], c='b')
plt.title("Interacting Surface Residues Bound Conformation")
pdf.savefig()
plt.close()
plt.figure()
plt.plot(u_l_dist)
plt.plot(u_r_dist)
plt.legend(["bound ligand {0}".format(i), "bound receptor {0}".format(j), "unbound ligand",
"unbound receptor"])
pdf.savefig()
plt.close()
p = True
else:
if n:
continue
lb2u = c.ligand_bound_to_unbound
rb2u = c.receptor_bound_to_unbound
b_l_points, b_l_dist = get_coords(u_l_ns, u_ligand, lb2u[b_ligand.residues[i]], 0.5, False)
b_r_points, b_r_dist = get_coords(u_r_ns, u_receptor, rb2u[b_receptor.residues[j]], 0.5,
False)
u_l_points, u_l_dist = get_coords(u_l_ns, u_ligand, lb2u[b_ligand.residues[i]], 0.5, True)
u_r_points, u_r_dist = get_coords(u_r_ns, u_receptor, rb2u[b_receptor.residues[j]], 0.5,
True)
fig = plt.figure()
ax = fig.add_subplot(111, projection='3d')
ax.scatter(b_l_points[:, 0], b_l_points[:, 1], b_l_points[:, 2], c='r')
ax.scatter(b_r_points[:, 0], b_r_points[:, 1], b_r_points[:, 2], c='b')
plt.title("Non-Interacting Residues Bound Conformation")
pdf.savefig()
plt.close()
fig = plt.figure()
ax = fig.add_subplot(111, projection='3d')
ax.scatter(u_l_points[:, 0], u_l_points[:, 1], u_l_points[:, 2], c='r')
ax.scatter(u_r_points[:, 0], u_r_points[:, 1], u_r_points[:, 2], c='b')
plt.title("Non-Interacting Surface Residues Bound Conformation")
pdf.savefig()
plt.close()
plt.figure()
plt.plot(u_l_dist)
plt.plot(u_r_dist)
plt.legend(["bound ligand {0}".format(i), "bound receptor {0}".format(j), "unbound ligand",
"unbound receptor"])
pdf.savefig()
plt.close()
n = True
except GetOutOfLoop:
pass