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 pdb_neighbors(pdb_f, pdb_id):
structure = PDBParser().get_structure(pdb_id, pdb_f)
atom_list = Selection.unfold_entities(structure, 'A')
ns = NeighborSearch(atom_list)
center_res = [res for res in structure.get_residues() if res.get_resname() in ['PTR','SEP','TPO']]
neighbors = []
for res in center_res:
if res.get_resname() == 'PTR':
atoms = [atom for atom in res.child_list if atom.get_name() in ['O1P', 'O2P', 'O3P', 'OH']]
elif res.get_resname() == 'SEP':
atoms = [atom for atom in res.child_list if atom.get_name() in ['O1P', 'O2P', 'O3P', 'OG']]
elif res.get_resname() == 'TPO':
atoms = [atom for atom in res.child_list if atom.get_name() in ['O1P', 'O2P', 'O3P', 'OG1']]
atom_neighbors = [ns.search(a.get_coord(),BOND_CUTOFF) for a in atoms]
atom_neighbors = [atom for atoms in atom_neighbors for atom in atoms]
atom_neighbors = list(set(atom_neighbors))
atom_neighbors = [atom for atom in atom_neighbors if 'N' in atom.get_name() or 'O' in atom.get_name()]
atom_neighbors = list(set(Selection.unfold_entities(atom_neighbors,'R')))
atom_neighbors = [r for r in atom_neighbors if r != res]
if len(atom_neighbors) > 0:
res = res.get_resname()+'_'+str(res.get_id()[1])+'_'+res.get_parent().get_id()
atom_neighbors = [n.get_resname()+'_'+str(n.get_id()[1])+'_'+n.get_parent().get_id() for n in atom_neighbors]
neighbors.append((pdb_id,res,atom_neighbors))
return neighbors
def within(resnum, angstroms, chain_id, model, use_ca=False, custom_coord=None):
"""See: https://www.biostars.org/p/1816/ https://www.biostars.org/p/269579/
Args:
resnum (int):
angstroms (float):
chain_id (str):
model (Model):
use_ca (bool): If the alpha-carbon atom should be used for the search, otherwise use the last atom of the residue
custom_coord (list): Custom XYZ coordinate to get within
Returns:
list: List of Bio.PDB.Residue.Residue objects
"""
# XTODO: documentation
# TODO: should have separate method for within a normal residue (can use "resnum" with a int) or a custom coord,
# where you don't need to specify resnum
atom_list = Selection.unfold_entities(model, 'A')
ns = NeighborSearch(atom_list)
if custom_coord: # a list of XYZ coord
target_atom_coord = np.array(custom_coord, 'f')
else:
target_residue = model[chain_id][resnum]
if use_ca:
target_atom = target_residue['CA']
else:
target_atom = target_residue.child_list[-1]
target_atom_coord = np.array(target_atom.get_coord(), 'f')
neighbors = ns.search(target_atom_coord, angstroms)
residue_list = Selection.unfold_entities(neighbors, 'R')
return residue_list
def get_residues_distance_distribution(data, r):
"""
This function computes the distribution of number of residues in certain radius around residues.
"""
files = glob.glob(data + "*_b.pdb")
dist = {}
files.sort()
file_counter = 0
for bound_pbd in files:
dist[bound_pbd] = []
l = []
for bound_pbd in files:
file_counter += 1
print "Protein " + str(file_counter) + "/" + str(len(files))
s, a, r = read_pdb_file(bound_pbd)
ns = NeighborSearch(a)
res_counter = 0
for res in r:
b = 0 * res.child_list[0].get_coord()
for atom in res.child_list:
b += atom.get_coord()
center = b / len(res.child_list)
l.append(len(ns.search(center, 100, "R")))
res_counter += 1
# print "Residue " + str(res_counter) + "out of " + str(len(r))
plot(np.bincount(l))
show()
print files
def search_connectivity(self, atom_center, rank, NS=None, previous=None):
if NS is None:
NS = NeighborSearch(self.model.atoms)
resultat = [
atom for atom in NS.search(center=atom_center.get_coord(),
radius=parametres.cutoff_bonds,
level="A")
if atom not in [atom_center, previous]
]
if rank == 1:
return resultat
else:
connectivity = {}
for atom in resultat:
resultat_temp = self.search_connectivity(atom_center=atom,
rank=rank - 1,
NS=NS,
previous=atom_center)
if len(resultat_temp) == 0:
connectivity[atom] = None
else:
connectivity[atom] = resultat_temp
return connectivity
def __init__(self, structure, chain_id, threshold): ns = NeighborSearch(list(structure.get_atoms())) for chain in structure[0]: if chain.id == chain_id: for residue in chain: if residue.id[0] == ' ': inter = 0 intra = 0 lon = 0 shor = 0 center = get_center(residue) for a in ns.search(center, threshold): # Iterate over contacts if a.get_full_id()[2] == residue.get_full_id()[2]: #if abs(int(a.get_full_id()[1]) - int(residue.get_full_id()[1])) > 3: intra += 1 else: inter += 1 if np.linalg.norm(a.get_coord()-center) < threshold/2: shor += 1 else: lon += 1 self.contacts[residue.get_full_id()] = (inter, intra, lon, shor)
def calculate_ic(structure, d_cutoff=5.0, selection=None):
"""
Calculates intermolecular contacts in a parsed structure object.
"""
atom_list = list(structure.get_atoms())
ns = NeighborSearch(atom_list)
all_list = ns.search_all(radius=d_cutoff, level="R")
if selection:
_sd = selection
def _chain(x):
return x.parent.id
ic_list = [
c for c in all_list
if (_chain(c[0]) in _sd and _chain(c[1]) in _sd) and (
_sd[_chain(c[0])] != _sd[_chain(c[1])])
]
else:
ic_list = [c for c in all_list if c[0].parent.id != c[1].parent.id]
if not ic_list:
raise ValueError("No contacts found for selection")
return ic_list
def search_subs(
self,
hx_atom,
cx_atom,
cy_atom,
hy_atom,
j_atoms,
chosen_j
):
""" Description:
Usage:
Parameters:
"""
subs_list = []
# j_atoms = [j[0] for j in j_atoms]
center = self.atom_dict[cy_atom][2]
ns = NeighborSearch(self.atom_list)
neighbors = ns.search(center, 1.7)
for neigh_atom in neighbors:
if neigh_atom.serial_number not in j_atoms:
subs_list.append(neigh_atom)
for subs in subs_list:
self.j_dict[chosen_j]["substituents"][subs.serial_number] = {
"SY": self.atom_dict[subs.serial_number][0],
"HX": self.atom_dict[hx_atom][0],
"CX": self.atom_dict[cx_atom][0],
"CY": self.atom_dict[cy_atom][0],
"element": self.atom_dict[subs.serial_number][1]
}
def _get_contacts(pdb_path, chain_rec, chain_lig, contact_dist):
structure = pdb_parser.get_structure('X', pdb_path)[0]
receptor = [structure[chain_rec_id] for chain_rec_id in chain_rec]
ligand = [structure[chain_lig_id] for chain_lig_id in chain_lig]
receptor_atoms = Selection.unfold_entities(receptor, 'A')
ns = NeighborSearch(receptor_atoms)
ligand_residues = Selection.unfold_entities(ligand, 'R')
contacts = set([])
contacts_lig = set([])
contacts_rec = set([])
for ligand_res in ligand_residues:
lig_resname = dindex_to_1[d3_to_index[ligand_res.get_resname()]]
lig_resnum = ligand_res.get_id()[1]
lig_chname = ligand_res.get_parent().get_id()
res_contacts = []
for lig_atom in ligand_res:
neighbors = ns.search(lig_atom.get_coord(), contact_dist)
res_contacts += Selection.unfold_entities(neighbors, 'R')
for receptor_res in res_contacts:
rec_resname = dindex_to_1[d3_to_index[receptor_res.get_resname()]]
rec_resnum = receptor_res.get_id()[1]
rec_chname = receptor_res.get_parent().get_id()
contacts.add((rec_resname, rec_resnum, rec_chname, lig_resname,
lig_resnum, lig_chname))
contacts_lig.add((lig_resname, lig_resnum, lig_chname))
contacts_rec.add((rec_resname, rec_resnum, rec_chname))
return contacts, contacts_rec, contacts_lig
def pdb_neighbors(pdb_f, pdb_id):
structure = PDBParser().get_structure(pdb_id, pdb_f)
atom_list = Selection.unfold_entities(structure, 'A')
ns = NeighborSearch(atom_list)
center = [(a, a.get_coord()) for a in structure.get_atoms()
if a.get_parent().get_resname() in ['PTR', 'SEP', 'TPO']
and a.get_name() in ['O1P', 'O2P', 'O3P', 'OH', 'OG', 'OG1']]
# if there are no phos-atomes, return
if len(center) == 0:
return ''
neighbors = {}
for a, c in center:
# set neighbor distance cutoff
neighbor_list = ns.search(c, BOND_CUTOFF)
residue_list = list(set(Selection.unfold_entities(neighbor_list, 'R')))
try:
neighbors[Selection.unfold_entities(a, 'R')[0]] = [
x for x in residue_list
if not x == Selection.unfold_entities(a, 'R')[0]
]
except:
continue
# add residue id and chain id
neighbors_full = dict([('_'.join(
[k.get_resname(),
str(k.get_id()[1]),
str(k.get_parent().get_id())]), [
'_'.join([
vi.get_resname(),
str(vi.get_id()[1]),
str(vi.get_parent().get_id())
]) for vi in v
]) for k, v in neighbors.iteritems()])
return neighbors_full
def collectSaltBridge(self, position, chain):
model = self.structure[0]
myres = model[chain][position]
atoms = Selection.unfold_entities(model, 'A') # A for atoms
ns = NeighborSearch(atoms)
if myres.get_resname() not in ['ARG', 'LYS', 'ASP', 'GLU', 'HIS']:
is_sb = 0
else:
for atom in myres:
atom_id = atom.get_full_id()
if atom_id[4][0] in ['NH1', 'NH2', 'NZ', 'NE2']:
close_atoms = ns.search(
atom.coord, 4.5
) # cutoff of 4 crieria fixed by Barlow, J M Thornton (PMID6887253) +0.5A to account for the unoptimised side chain
if any(atom in [atomtype.id for atomtype in close_atoms]
for atom in ['OE1', 'OE2', 'OD1', 'OD2']):
is_sb = 1
break
else:
is_sb = 0
break
elif atom_id[4][0] in ['OE1', 'OE2', 'OD1', 'OD2']:
close_atoms = ns.search(atom.coord, 4.5)
if any(atom in [atomtype.id for atomtype in close_atoms]
for atom in ['NH1', 'NH2', 'NZ', 'NE2']):
is_sb = 1
break
else:
is_sb = 0
break
return is_sb
def collectHbondsNumber(self, foldxIndivFile):
wt_mut_chain_id = whatMutation(foldxIndivFile)
prot_chain = wt_mut_chain_id[3]
prot_mut_id = int(wt_mut_chain_id[2])
model = self.structure[0]
myres = model[prot_chain][prot_mut_id]
atoms = Selection.unfold_entities(model, 'A') # A for atoms
ns = NeighborSearch(atoms)
resname = myres.get_resname()
h_bond_as_donor = []
h_bond_as_acceptor = []
for atom in myres:
if atom.name in donors:
#atoms = Selection.unfold_entities(model, 'A')
#print(atoms)
# cutoff of 3.2 angstroms D-A, strong mostly covalent according to
# Jeffrey, George A.; An introduction to hydrogen bonding, Oxford University Press, 1997.
close_atoms = ns.search(atom.coord, 3.2)
for close_atom in close_atoms:
full_atom_id = close_atom.get_full_id()
if (full_atom_id[3][1] != prot_mut_id) and (
full_atom_id[4][0]
in acceptors + main_chain_acceptors):
acceptor_atom_id = full_atom_id[3][1]
h_bond_as_donor.append(acceptor_atom_id)
# do the same thing fro acceptor atoms
if atom.name in acceptors:
close_atoms = ns.search(atom.coord, 3.2)
for close_atom in close_atoms:
full_atom_id = close_atom.get_full_id()
if (full_atom_id[3][1] != prot_mut_id) and (
full_atom_id[4][0] in donors + main_chain_donors):
acceptor_atom_id = full_atom_id[3][1]
h_bond_as_acceptor.append(acceptor_atom_id)
return len(h_bond_as_donor) + len(h_bond_as_acceptor)
def __calculate_separate_simbox_statistics(self, component, radius): #, cells_taken_by_component):
"""
calculates the number of grid cells describing each complex component.
calculation is done when all components are located in the center of simulation box
"""
self.cells_taken_by_component = {}
grid = Grid("cubic", radius, 10*radius, -1.)
grid.calculate_boundaries(component.pyrystruct.struct)
grid.generate_cubic_grid([])
#scitree = scipyconverter(grid.grid_cells)
ns = NeighborSearch(grid.grid_cells)
for atom in component.pyrystruct.struct.get_atoms():
center = array([atom.coord[0], atom.coord[1], atom.coord[2]])
if atom.vdw <= radius: rad = radius* sqrt(3)
else: rad = (radius*sqrt(3)) + atom.vdw
neighbours = ns.search(center, rad, 'A') #atom.vdw + radius
#neighbours = scitree.run_neisearch(center, rad, eps=0)
self.cells_taken_by_component = self.__calculate_simbox_statistics(neighbours) #, cells_taken_by_component)
if len(neighbours) == 0: print "OMGOMG"*10, component.pyrystruct.chain, atom.serial_number
del grid
del ns
neighbours = []
return len(self.cells_taken_by_component)
def CheckClashes(structure, chain):
"""
Checks for clashes at a radius = 2 between a PDB structure and all the atoms on the provided chain.
Returns True or False depending if number of different residues clashing exceeds 25.
Arguments:
-structure: PDB structure.
-chain: PDB chain structure to check if it has clashes with the main structure.
"""
# declare NeighborSearch() object instance with all the atoms from the structure (model 0), that includes all chains of that structure.
ns = NeighborSearch(unfold_entities(structure[0], 'A'))
# iterate over atoms in input chain, search for close residues
clashing_residues = set([])
for atom in chain.get_atoms():
close_res = ns.search(atom.get_coord(), radius=2, level="R")
try:
close_res.remove(atom.get_parent())
except ValueError:
pass
for res in close_res:
neighbor_res = (atom.get_parent(), res)
clashing_residues.add(neighbor_res)
if len(clashing_residues) > 25:
return True
return False
def get_residues_distance_distribution(data, r):
"""
This function computes the distribution of number of residues in certain radius around residues.
"""
files = glob.glob(data + "*_b.pdb")
dist = {}
files.sort()
file_counter = 0
for bound_pbd in files:
dist[bound_pbd] = []
l = []
for bound_pbd in files:
file_counter += 1
print "Protein " + str(file_counter) + "/" + str(len(files))
s, a, r = read_pdb_file(bound_pbd)
ns = NeighborSearch(a)
res_counter = 0
for res in r:
b = 0 * res.child_list[0].get_coord()
for atom in res.child_list:
b += atom.get_coord()
center = b / len(res.child_list)
l.append(len(ns.search(center, 100, "R")))
res_counter += 1
# print "Residue " + str(res_counter) + "out of " + str(len(r))
plot(np.bincount(l))
show()
print files
def contact_pairs(self):
"""obtain all residue pairs with atoms in close proximity"""
radius = 4.5
ns = NeighborSearch(self.atom_list)
self.contact_pairs = ns.search_all(radius,
level='R') #residues returned
def get_neighbor_chains(structure, options):
"""
Takes an structure and returns a dictionary with chains as keys and a list of chains as values holding the
chains with alpha carbons at less than 8 amstrongs from an alpha carbon of the key chain
:param structure: structure we want to check for clashes.
:return: dictionary with the clashes between chains.
"""
neighbor_chains = {}
ns = NeighborSearch(list(structure.get_atoms()))
for chain in structure.get_chains():
chains = list(structure.get_chains())
neighbor_chains[chain] = set([])
neighbor_dict = {}
for atom in [atom for atom in chain.get_atoms() if
atom.get_id() == 'CA' or atom.get_id() == 'P']: # For every alpha carbon or P in chain
for atom2 in ns.search(atom.get_coord(), 8, level='A'):
if atom2.get_id() == 'CA' or atom2.get_id() == 'P': # for every alpha carbon or P at 8 angstroms or less from atom
chain2 = atom2.get_parent().get_parent() # Getting to which chain it belongs
if chain2 != chain and chain2 not in neighbor_chains.keys():
# If it is not in the same chain and it is not already assessed
if chain2 not in neighbor_dict:
neighbor_dict[chain2] = 0
neighbor_dict[chain2] += 1
if options.verbose:
print('\n%s' % chain)
for close_chain, contacts in neighbor_dict.items():
print('%s: %s' % (close_chain, contacts))
if contacts > 8:
neighbor_chains[chain].add(close_chain)
return neighbor_chains
def check_clash(str_name, v=False):
"""check_clash, fract of clashes!
if zero contacts then error -> fix ->
Problem, contacts, str_name: 311 505 na-prot_13536.pdb
Sterical clashes 0.615841584158
c is counter
"""
if v: print('fn:', str_name)
structure = open(str_name)
#model = structure[0]
atoms_A = []
atoms_B = []
for line in structure.readlines():
if line[:4] == "ATOM":
#print line
at_nam = line[12:16].strip()
coor = [float(line[30:38]), float(line[38:46]), float(line[46:54])]
at = Atom.Atom(at_nam, coor, 0.0, 1.0, ' ', at_nam, 1, at_nam[0])
if line[21] == "A":
atoms_A.append(at)
elif line[21] == "B":
atoms_B.append(at)
else:
pass
#atoms_B = Selection.unfold_entities(structure[0]['B'], 'A')
#print len(atoms_A), len(atoms_B)
if len(atoms_A) > len(atoms_B):
less = atoms_B
more = atoms_A
else:
less = atoms_A
more = atoms_B
problem = 0
contacts = 0
ns = NeighborSearch(more)
for at in less:
neighbors = ns.search(array(at.get_coord()), 2.0, 'A')
if neighbors != []:
problem += 1
contacts += 1
else:
neighbors1 = ns.search(array(at.get_coord()), 4.0, 'A')
if neighbors1 != []:
contacts += 1
if v:
print('problem:', float(problem))
print('contacts:', float(contacts))
try:
fract = float(problem) / float(contacts)
except ZeroDivisionError:
fract = problem # or skip this structure
if v: print('ZeroDivison -- skip:', problem, contacts, str_name)
return fract
#print 'Contacts, str_name:', problem, contacts, str_name, "Sterical clashes ", fract
return fract
def check_clash(str_name, v=True):
"""check_clash, fract of clashes!
if zero contacts then error -> fix ->
Problem, contacts, str_name: 311 505 na-prot_13536.pdb
Sterical clashes 0.615841584158
c is counter
"""
print(str_name)
structure = open(str_name)
#model = structure[0]
atoms_A = []
atoms_B = []
for line in structure.readlines():
if line[:4] == "ATOM":
#print line
at_nam = line[12:16].strip()
coor = [float(line[30:38]),float(line[38:46]), float(line[46:54])]
at = Atom.Atom(at_nam,coor,0.0,1.0,' ',at_nam,1,at_nam[0])
if line[21] == "A":
atoms_A.append(at)
elif line[21] == "B":
atoms_B.append(at)
else: pass
#atoms_B = Selection.unfold_entities(structure[0]['B'], 'A')
#print len(atoms_A), len(atoms_B)
if len(atoms_A) > len(atoms_B):
less = atoms_B
more = atoms_A
else:
less = atoms_A
more = atoms_B
problem = 0
contacts = 0
ns=NeighborSearch(more)
for at in less:
neighbors=ns.search(array(at.get_coord()),2.0,'A')
if neighbors != []:
problem +=1
contacts +=1
else:
neighbors1=ns.search(array(at.get_coord()),4.0,'A')
if neighbors1 != []:
contacts +=1
if v:
print('problem:', float(problem))
print('contacts:', float(contacts))
try:
fract = float(problem)/float(contacts)
except ZeroDivisionError:
fract = problem # or skip this structure
print('ZeroDivison -- skip:', problem, contacts, str_name)
return fract
#print 'Contacts, str_name:', problem, contacts, str_name, "Sterical clashes ", fract
return fract
def compute_interaction_center(pdb_file, mutation_site):
"""Computes the geometric center of all heavy atoms interacting with the mutated residue.
Parameters
----------
pdb_file : str
Path to the PDB file containing the structure of the protein.
mutation_size : int
An integer designating the residue sequence ID.
Returns
-------
NumPy ndarray
The Cartesian coordinates of the geometric center
"""
pdb_parser = PDBParser(PERMISSIVE=1)
model = pdb_parser.get_structure(id='tmp', file=pdb_file)
# get all heavy atoms of the protein
all_heavy_atoms = [a for a in model.get_atoms() if a.element != 'H']
# get the mutated residue
mutation_res = None
for res in model.get_residues():
if res.get_id()[1] == int(mutation_site):
mutation_res = res
break
# get sidechain atoms
if mutation_res is not None:
heavy_atoms = [a for a in mutation_res.get_list() if a.element != 'H']
# if only four heavy atoms, aka, GLY, use CA
if len(heavy_atoms) == 4:
side_chain_atoms = [heavy_atoms[1]]
else:
side_chain_atoms = mutation_res.get_list()[4:]
else:
print('Invalid mutation site: {}'.format(mutation_site))
sys.exit(1)
# search for neighbnoring atoms
ns = NeighborSearch(atom_list=all_heavy_atoms, bucket_size=10)
all_interaction_atoms = []
for a in side_chain_atoms:
interaction_atoms = ns.search(center=a.coord, radius=5, level='A')
all_interaction_atoms += interaction_atoms
# remove duplicates
all_interaction_atoms = set(all_interaction_atoms)
# compute geometric center of all interaction atoms
geometric_center = np.zeros((3,))
for a in all_interaction_atoms:
geometric_center += a.coord / len(all_interaction_atoms)
return geometric_center
def get_interactions_between_chains(model, chain_id_1, chain_id_2, r_cutoff=6):
"""Calculate interactions between the residues of the two chains.
An interaction is defines as a pair of residues where at least one pair of atom
is closer than r_cutoff.
.. deprecated:: 1.0
Use python:fn:`get_interacting_residues` instead.
It gives you both the residue index and the resnum.
Returns
-------
OrderedDict
Keys are (residue_number, residue_amino_acid) tuples
(e.g. ('0', 'M'), ('1', 'Q'), ...).
Values are lists of (residue_number, residue_amino_acid) tuples.
(e.g. [('0', 'M'), ('1', 'Q'), ...]).
"""
try:
from Bio.PDB import NeighborSearch
except ImportError as e:
logger.warning('Importing Biopython NeighborSearch returned an error: {}'.format(e))
logger.warning('Using the the slow version of the neighbour-finding algorithm...')
return get_interactions_between_chains_slow(model, chain_id_1, chain_id_2, r_cutoff)
# Extract the chains of interest from the model
chain_1 = None
chain_2 = None
for child in model.get_list():
if child.id == chain_id_1:
chain_1 = child
if child.id == chain_id_2:
chain_2 = child
if chain_1 is None or chain_2 is None:
raise Exception('Chains %s and %s were not found in the model' % (chain_id_1, chain_id_2))
ns = NeighborSearch(list(chain_2.get_atoms()))
interactions_between_chains = OrderedDict()
for idx, residue_1 in enumerate(chain_1):
if residue_1.resname in AMINO_ACIDS and residue_1.id[0] == ' ':
resnum_1 = str(residue_1.id[1]) + residue_1.id[2].strip()
resaa_1 = convert_aa(residue_1.get_resname(), quiet=True)
interacting_residues = set()
for atom_1 in residue_1:
interacting_residues.update(ns.search(atom_1.get_coord(), r_cutoff, 'R'))
interacting_resids = []
for residue_2 in interacting_residues:
resnum_2 = str(residue_2.id[1]) + residue_2.id[2].strip()
resaa_2 = convert_aa(residue_2.get_resname(), quiet=True)
if residue_2.resname in AMINO_ACIDS and residue_2.id[0] == ' ':
interacting_resids.append((resnum_2, resaa_2,))
if interacting_resids:
interacting_resids.sort(
key=lambda x: int(''.join([c for c in x[0] if c.isdigit()])))
interactions_between_chains[(resnum_1, resaa_1)] = interacting_resids
return interactions_between_chains
def make_neighbors(self, fl_struct):
# create an empty NeighborsNet
nn = NeighborsNet()
# use NeighborSearch from Bio.PDB to compute distances
ns = NeighborSearch(list(self.bio_struct.get_atoms()))
# for each chain in structure that is not a dna-rna one
for fl_chain in fl_struct.get_chains():
if not fl_chain.rna_dna_chain:
# for each residue in this chain
for fl_res in fl_chain.residues:
# add a default entry
nn.add_default(fl_res)
# keep track of already inserted nieghbors (the search is mate for each atom in the residue)
already_have = []
# for each atom (coordinates) in the residue
for atom_coord in fl_res.atoms_coord:
# for each residue in range
for res in ns.search(atom_coord,
self.config["neighbors_range"],
level='R'):
# check if it is good atom and the same model, cause sometimes NS computes all models
if is_good_res(
res) and fl_res.model_id == res_model_id(
res) and not res.get_full_id(
) in already_have:
# try to get FlipperResidueAssociated
pos_2 = fl_struct.chains[res_chain_id(
res)].string_index_map.get(
res_string_index(res))
# print(fl_res.get_full_identifier(), res.get_full_id(), fl_struct.chains[res_chain_id(res)].string_index_map.get(res_string_index(res)))
if not pos_2 == None:
fl_res_2 = fl_struct.chains[res_chain_id(
res)].residues[pos_2]
# if the chain is the same
if fl_res.chain_id == fl_res_2.chain_id:
already_have.append(res.get_full_id())
if fl_res.pos_in_chain == fl_res_2.pos_in_chain:
continue
# if distance (as residue number) is less than threshold, then it is a short range neighbor
if abs(
fl_res.pos_in_chain -
fl_res_2.pos_in_chain
) < self.config[
"long_short_threshold"] and not fl_chain.have_gaps(
fl_res, fl_res_2):
nn.add_short(fl_res, fl_res_2)
# else it is a long rage neighbor
else:
nn.add_long(fl_res, fl_res_2)
# if it is not in the same chain it is an inter chain neighbor
else:
nn.add_inter(fl_res, fl_res_2)
already_have.append(res.get_full_id())
return nn
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 secondary_struc_cmap(chain,
sequence,
structure,
cutoff_distance=4.5,
cutoff_numcontacts=10,
exclude_neighbour=3,
ss_elements=['H', 'E', 'B', 'b', 'G']):
atom_list = Selection.unfold_entities(chain, 'A')
res_list = Selection.unfold_entities(chain, 'R')
res_names, numbering = [], []
for res in res_list:
res_names.append(res.get_resname())
numbering.append(res.get_id()[1])
numbering = np.array(numbering)
res_range = np.array(range(len(numbering)))
assert len(structure) == len(
numbering
), f'PDB file and Secondary structure map do not match!\n {chain.get_parent().get_parent().id} - PDB: {len(res_list)} Residues VS. STRIDE: {len(sequence)} Residues. '
ns = NeighborSearch(atom_list)
all_neighbours = ns.search_all(cutoff_distance, 'A')
struc_length = len(structure)
segment = np.zeros([struc_length], dtype='int')
nseg = 1
for i in range(struc_length):
if structure[i] in ss_elements:
segment[i] = nseg
if i == struc_length:
nseg += 1
elif structure[i + 1] != structure[i]:
nseg += 1
nseg -= 1
index_list = []
for atompair in all_neighbours:
res1 = res_range[numbering == atompair[0].get_parent().id[1]][0]
res2 = res_range[numbering == atompair[1].get_parent().id[1]][0]
if abs(res1 - res2) > exclude_neighbour:
if segment[res1] != 0 and segment[res2] != 0 and segment[
res1] != segment[res2]:
index_list.append((segment[res1] - 1, segment[res2] - 1))
index_list.sort()
count = Counter(index_list)
index = [values for values in count if count[values] >= cutoff_numcontacts]
return np.array(index), segment
def check_clash(model, chain_to_add, clash_distance=2.5):
"""
Checks wether a newly added chain has clashes with the rest of the structure. Returns a set containing the clashing chains and a boolean indicating whether the newly added chain is clashing with itself or not.
model: previous structure.
chain_to_add: new chain (Chain object from PDB.Chain).
clash_distance: indicates a distance threshold (in Angstroms) to consider two atoms as clashing.
"""
# Initialization of NeighbourSearch, that allows to find clashes
neighbor_object = NeighborSearch(list(model.get_atoms()))
structure_clashing_chains = set()
total_clashes = 0
# For each atom, clashes are compute
for atom in chain_to_add.get_atoms():
clashes = neighbor_object.search(atom.get_coord(), clash_distance)
if len(clashes) > 0: # If clashes are found...
# Increase the total number of clashes and add the conflicting chains to the set
for clash in clashes:
structure_clashing_chains.add(
clash.get_parent().get_parent().id)
total_clashes += 1
# In case the new chain is conflicting with several chains
if len(structure_clashing_chains) > 1 and total_clashes > 20:
return structure_clashing_chains, False
elif len(structure_clashing_chains
) == 1 and total_clashes > 20 and chain_to_add.id[1] == list(
structure_clashing_chains)[0][1]:
clash_chain = model[0][list(structure_clashing_chains)
[0]] # Conflictive chain
RMSD = superimpose(clash_chain, chain_to_add) # Compute RMSD
# If the RMSD is lower than 3.0, the chain is clashing with itself.
if RMSD <= 3.0:
return structure_clashing_chains, True
# Else the chain is clashing with another chain of the structure
else:
return structure_clashing_chains, False
elif total_clashes > 20:
return structure_clashing_chains, False
# No clashes found
else:
return None, False
def check_if_collide(self, component, point, radius):
"""
check whether aparticulat point does not collide with any components atoms
Return all component's atoms that have at least one
atom within radius of center for given point
"""
ns = NeighborSearch(list(component.get_atoms()))
point_center = array([point.x, point.y, point.z])
#we assume that pseudoresidue radius ;lis 1.5A as a collistion detection area.
found_collisions = len(ns.search(point_center, radius + 1.5, "A"))
return found_collisions
def interaction(self, pdb_id, filename, domain_1, domain_2):
"""Returns a dict with informations (atoms, residues...) if two domains
interact with each other, and returns False if not."""
print "Searching for interactions in "+pdb_id+"..."
# creates a strucuture object/class to extract atoms of the two domains
model = structure(pdb_id).get_model(pdb_id, filename)
residues_1 = structure(pdb_id).get_residues(model, domain_1)
residues_2 = structure(pdb_id).get_residues(model, domain_2)
atoms_1 = Selection.unfold_entities(residues_1, 'A')
atoms_2 = Selection.unfold_entities(residues_2, 'A')
# gets the serial numbers of the atoms
numbers_1 = structure(pdb_id).serial_numbers(atoms_1)
numbers_2 = structure(pdb_id).serial_numbers(atoms_2)
# the search starts here !
atoms = Selection.unfold_entities(model, 'A')
nsearch = NeighborSearch(atoms)
interacting_atoms_1 = []
interacting_atoms_2 = []
for atom in atoms:
if atom.get_serial_number() in numbers_1:
point = atom.get_coord()
# This is how we detect an interaction, we put 5 angstroms
# here.
# This is the simplest method we can use, and we're not sure
# that it is correct.
# Originally we have planned to go further by doing a surface
# and accesssion analysis, but we had no time.
# We hope we can talk about that during the talk.
neighbors = nsearch.search(point, 5)
for neighbor in neighbors:
if neighbor.get_serial_number() in numbers_2:
interacting_atoms_2.append(neighbor)
if atom not in interacting_atoms_1:
interacting_atoms_1.append(atom)
# returns a dict with all residues and atoms
if len(interacting_atoms_2) > 0:
infos = {}
infos['1'] = {}
infos['2'] = {}
# just get the parent residues for the list of atoms
interacting_residues_1 = structure(pdb_id).atoms2residues(
interacting_atoms_1)
interacting_residues_2 = structure(pdb_id).atoms2residues(
interacting_atoms_2)
infos['1']['atoms'] = interacting_atoms_1
infos['2']['atoms'] = interacting_atoms_2
infos['1']['residues'] = interacting_residues_1
infos['2']['residues'] = interacting_residues_2
return infos
else: return False
def getReferenceAtomsWithinSphere(self, residue, radius, caOnly=False):
if (not residue.__class__.__name__ == 'Residue'):
raise TypeError(
"The function process(residue, radius, caOnly) expects the first argument of class Bio.PDB.Residue"
)
ns = NeighborSearch(Selection.unfold_entities(self.ref_chain, 'A'))
ref_atoms = ns.search(residue['CA'].get_coord(), radius, level='A')
if (caOnly):
cas = []
for a in ref_atoms:
if (a.get_name() == 'CA'):
cas.append(a)
return cas
return ref_atoms
def get_neighbor_atoms(chain: ChainDesc, ligand: PhysicalResidue, produce_physical_atoms: bool = True) -> list: # a list of physical atoms
# use biopython neighboursearch to get list of AA atoms that close enough to ligand's atoms (using 10 angstroms)
# - get list of all chain's atoms except ligand's atoms
# - for each ligand's atom run neighbor search to find neighbors
# - add that neighbors to result list of neighbors (exclude duplicates)
# collect chain atoms
chain_atoms = list()
# get bio python residues of the chain
bio_residues = list(chain.chain.get_residues())
for residue in chain.get_residues(): # physical residue
# do not count atoms from ligand itself
if residue.get_residue_desc().get_short_name() == ligand.get_residue_desc().get_short_name():
continue
# do not count atoms from ligands
# TODO: refactor
if residue.get_residue_desc().get_short_name() not in database.get_amino_acids() + database.get_cofactors():
continue
elif residue.get_residue_desc().check_deeper():
continue
for atom in bio_residues[residue.get_index() - 1].get_atoms(): # BioPython atom!!!!
chain_atoms.append(atom)
neighbour_atoms = list()
for atom in ligand.get_atoms(): # physical atom
search = NeighborSearch(chain_atoms)
current_neighbours = search.search(atom.get_coords(), 10.0)
for neighbour in current_neighbours: # BioPython atom!!!!
if neighbour not in neighbour_atoms:
neighbour_atoms.append(neighbour)
if not produce_physical_atoms:
return neighbour_atoms
neighbour_physical_atoms = list()
for neighbour in neighbour_atoms:
terminus = None
if neighbour.get_parent().get_id()[1] == 1:
terminus = 'N'
elif 'OXT' in [a.get_id() for a in neighbour.get_parent().get_atoms()]:
terminus = 'C'
atom_desc = database.get_residue(neighbour.get_parent(), terminus).get_atom(neighbour.get_id())
physical_atom = PhysicalAtom(bio_atom=neighbour, atom_desc=atom_desc, coords=neighbour.get_coord())
neighbour_physical_atoms.append(physical_atom)
return neighbour_physical_atoms
def get_contacts(model, cutOff=5., minSeqDist=5):
contacts = list()
ns = NeighborSearch(list(model.get_atoms()))
foundPairs = ns.search_all(cutOff)
for pair in foundPairs:
fullAtomIDs = [pair[0].get_full_id(), pair[1].get_full_id()]
distance = pair[1] - pair[0]
weight = 1.
contacts.append((fullAtomIDs, distance, weight))
pass
return contacts
def search_stacking(self):
""" KC - stacking research in model """
#KC#CG# list of aromatics residues atoms
atom_for_search = []
for res in self.model.res:
if res.resname in ['PHE', 'TYR', 'TRP', 'HIS']:
for name in parametres.dico_cycles[res.resname]["cycle"]:
atom_for_search.append(res[name])
#KC#CG# list of residues pairs that have atoms pairs within 6A radius
self.aromatic_less_5 = NeighborSearch(atom_for_search).search_all(
6, level='R')
#KC#CG# search of stacking
self.stacking = []
for res1, res2 in self.aromatic_less_5:
if is_stacking(res1, res2):
if res1.get_id()[1] < res2.get_id()[1]:
self.stacking.append((res1, res2))
else:
self.stacking.append((res2, res1))
self.stacking_networks = self.search_network(self.stacking,
_type='stacking')
def pdb_dist(pdb_f, pdb_id):
structure = PDBParser().get_structure(pdb_id, pdb_f)
atom_list = Selection.unfold_entities(structure, 'A')
ns = NeighborSearch(atom_list)
center = [a for a in structure.get_atoms() if a.get_parent().get_resname(
) in ['PTR', 'SEP', 'TPO'] and a.get_name() in ['O1P', 'O2P', 'O3P', 'OH', 'OG', 'OG1']]
# neighbors = [a for a in structure.get_atoms() if a.get_parent().get_resname(
# ) in ['ARG'] and a.get_name() in ['NE', 'NH2', 'NH1']]
# neighbors = [a for a in structure.get_atoms() if a.get_parent().get_resname(
# ) in ['Lys'] and a.get_name() in ['NZ']]
neighbors = [a for a in structure.get_atoms() if a.get_parent().get_resname(
) in ['HIS'] and a.get_name() in ['ND1','NE2']]
def calc_dist(a, b):
vector = a.coord - b.coord
return np.sqrt(np.sum(vector * vector))
dist = {}
for c in center:
for n in neighbors:
value = calc_dist(c, n)
key = str(c.get_parent()) + '_' + str(n.get_parent())
if not key in dist.keys():
dist[key] = [value]
else:
dist[key].append(value)
for k, v in dist.items():
dist[k] = min(v)
dist = [v for k, v in dist.items()]
return dist
def filter_structure(pdb,chain,lig,lig_num):
# pdb_input = pdb_path + "/" + pdb + ":" + chain + ":" + lig + ":" + lig_num + ".atoms.pdb"
pdb_input = pdb_path + "/" + pdb + ".pdb"
pdb_output = output_path + "/" + pdb + ":" + chain + ":" + lig + ":" + lig_num + ".atoms.pdb"
structure = PDBParser(QUIET=1).get_structure(pdb, pdb_input)
atoms_pairs = NeighborSearch( list( structure.get_atoms() ) ).search_all(search_radius)
res_list = set()
for atom_pair in atoms_pairs:
res1 = atom_pair[0].parent
res_chain1 = res1.parent.id
res_name1 = res1.resname
res_num1 = str(res1.id[1])
res2 = atom_pair[1].parent
res_chain2 = res2.parent.id
res_name2 = res2.resname
res_num2 = str(res2.id[1])
if ( (res_chain1 == chain and res_name1 == lig and res_num1 == lig_num) or (res_chain2 == chain and res_name2 == lig and res_num2 == lig_num) ):
res_list.add(res1)
res_list.add(res2)
io = PDBIO()
io.set_structure(structure)
io.save(pdb_output, ResSelect(res_list))
def interchain_residue_contacts(self, chain_ids_1, chain_ids_2, radius):
""" Generate a list of residue contacts between two chains. """
all_chains = {chain.get_id(): chain for chain in self.get_chains()}
selected_chains = {
chain_id: chain
for chain_id, chain in all_chains.items()
if chain_id in chain_ids_1 + chain_ids_2
}
atoms = [
atom for chain_id, chain in selected_chains.items()
for atom in Selection.unfold_entities(chain, "A")
]
residue_contacts = NeighborSearch(atoms).search_all(radius, "R")
classified_contacts = defaultdict(list)
for contact in residue_contacts:
chain_1, chain_2 = [
residue.get_parent().get_id() for residue in contact
]
if chain_1 in chain_ids_1 and chain_2 in chain_ids_2:
classified_contacts[(chain_1, chain_2)].append({
chain_1:
contact[0],
chain_2:
contact[1]
})
elif chain_2 in chain_ids_1 and chain_1 in chain_ids_2:
classified_contacts[(chain_2, chain_1)].append({
chain_1:
contact[0],
chain_2:
contact[1]
})
return classified_contacts
def calculateNeighbors(filename, radius):
data = {}
structure = parser.get_structure(filename.split(".pdb")[0], filename)
atom_list = Selection.unfold_entities(structure, 'A') # A for atoms
residue_list = Selection.unfold_entities(structure, 'R') # R for residues
neighbor_search = NeighborSearch(atom_list)
for residue in residue_list:
resid = str(residue.get_id()[1])
contacts = []
for atom in residue.get_list():
contacts.extend(neighbor_search.search(atom.get_coord(), radius, level = "A"))
burial = len(contacts)/len(residue.get_list())
data[resid] = burial
return data
def calculate_ic(structure, d_cutoff=5.5, selection=None):
"""
Calculates intermolecular contacts in a parsed structure object.
"""
atom_list = list(structure.get_atoms())
ns = NeighborSearch(atom_list)
all_list = ns.search_all(radius=d_cutoff, level='R')
if selection:
_sd = selection_dict
ic_list = [c for c in all_list if (c[0].parent.id in _sd and c[1].parent.id in _sd)
and (_sd[c[0].parent.id] != _sd[c[1].parent.id]) ]
else:
ic_list = [c for c in all_list if c[0].parent.id != c[1].parent.id]
if not ic_list:
raise ValueError('No contacts found for selection')
return ic_list
def _remove_distant_hatatms(self, new_model, hetatm_chain):
"""Detach hetatms that are more than ``self.r_cutoff`` away from the main chain(s)."""
ns = NeighborSearch(list(new_model.get_atoms()))
hetatm_chain.id = [
c for c in reversed(string.ascii_uppercase) if c not in self.chain_ids][0]
res_idx = 0
while res_idx < len(hetatm_chain):
res_1 = hetatm_chain.child_list[res_idx]
in_contact = False
for atom_1 in res_1:
interacting_residues = ns.search(atom_1.get_coord(), self.r_cutoff, 'R')
if interacting_residues:
# logger.debug(res_1.id)
# logger.debug(interacting_residues)
in_contact = True
if in_contact:
res_idx += 1
continue
# logger.debug('Detaching child: {}'.format(res_1.id))
hetatm_chain.detach_child(res_1.id)
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))
structure = PDBParser().get_structure('X', args.pdb)
center_atoms = []
pymol_command = ""
all_atom_list = [atom for atom in structure.get_atoms() if atom.name == 'CA' ]
for k in args.chain1 :
chain_atoms = [atom for atom in structure[0][k].get_atoms() if atom.name == 'CA' ]
center_atoms += chain_atoms
#atom_list = [x for x in all_atom_list if x not in center_atoms]
for j in args.chain2 :
atom_list = [atom for atom in structure[0][j].get_atoms() if atom.name == 'CA' ]
ns = NeighborSearch(atom_list)
nearby_residues = {res for center_atom in center_atoms
for res in ns.search(center_atom.coord, 8.5, 'R')}
print "\nNeighbor residues in chain ", j, ": \n"
print sorted(res.id[1] for res in nearby_residues)
pymol_command = "show spheres, chain " + j + " and resi "
for m in sorted(res.id[1] for res in nearby_residues):
pymol_command = pymol_command + str(m) + "+"
print pymol_command[:-1] + " and name CA \n"
SCALAR_EXPRESSION %s_o = %s_startenergy + 4.91
SCALAR_EXPRESSION %s_s = %s
SCALAR_EXPRESSION %s_exp = exp(%s_s*(%s_currentenergy-%s_o))
SCALAR_EXPRESSION %s_k = %s
SCALAR_EXPRESSION %s_sig = (1-%s_k)+(%s_k/(1+%s_exp))
""") %(structure_id, structurefilename, correspondencefilename, resfilename, structure_id, startenergy, structure_id, structure_id, structure_id, s_value, structure_id, structure_id, structure_id, structure_id, structure_id, k_value, structure_id, structure_id, structure_id, structure_id)
fitnessfile.write(outstring)
fitnessstring = fitnessstring + str("*%s_sig") % (structure_id)
nucleosome = structure[0]
atom_list = Selection.unfold_entities(nucleosome, 'A') # A for atoms
neighbor_search = NeighborSearch(atom_list)
contacts_list = neighbor_search.search_all(radius, level = 'R')
repack_residues = []
for contact in contacts_list:
res1 = contact[0]
res2 = contact[1]
res1id = int(res1.get_id()[1])
chain1 = res1.get_parent()
chain1id = chain1.get_id()
res2id = int(res2.get_id()[1])
chain2 = res2.get_parent()
chain2id = chain2.get_id()
res1_in_patch = False
PDB = pdb_biomol_match.group(1)
BIOMOL_ID = pdb_biomol_match.group(2)
pdb_rcsb_asm_match = re.search(PDB_RCSB_ASM_REGEX, INPUT_FILENAME)
if pdb_rcsb_asm_match:
PDB = pdb_rcsb_asm_match.group(1)
BIOMOL_ID = pdb_rcsb_asm_match.group(2)
# LOAD STRUCTURE
structure = PDBParser().get_structure('structure', INPUT_FILE)
structure_atoms = list(structure.get_atoms())
logging.info('Loaded PDB structure (BioPython).')
# CONSTRUCT KDTREE
neighborsearch = NeighborSearch(structure_atoms)
logging.info('Constructured NeighborSearch.')
# GET INTERACTIONS
logging.info('Calculating interactions...')
for interaction_level in 'ARC':
if interaction_level in OUTPUTS:
logging.info('Calculating interactions for {}s...'.format(
LEVEL_MAP[interaction_level]))
pairs = neighborsearch.search_all(INTERACTION_THRESHOLD,
level=interaction_level)
def _build_interface(self, model, id, threshold, rsa_calculation, rsa_threshold, include_waters=False, *chains):
"""
Return the interface of a model
"""
self.threshold=threshold
# Recover chain list from initial unpacking
chain_list = self.chain_list
# Unfold atom list
atom_list = []
for c in model:
if c.id in chain_list:
atom_list.extend(Selection.unfold_entities(c,'A'))
# Using of NeighborSearch class in order to get the list of all residues at least than
# the threshold distance of each others
ns=NeighborSearch(atom_list)
pairs=ns.search_all(threshold, 'R')
if not pairs:
raise ValueError("No atoms found in the interface")
# Selection of residues pairs
# 1. Exclude water contacts
# 2. Filter same-chain contacts
# 3. Filter user-defined chain pairs
uniq_pairs=[]
for pair in pairs:
pair_resnames = (pair[0].resname, pair[1].resname)
pair_chains = (pair[0].parent.id, pair[1].parent.id)
if (not include_waters and 'HOH' in pair_resnames) or (pair_chains[0] == pair_chains[1]):
continue
if not (chains and not (pair_chains in chains)):
uniq_pairs.append(pair)
# Build the Interface
# 1. Iterate over the pair list
# 2. Add residues.
for resA, resB in uniq_pairs:
if resA not in self.interface:
self._add_residue(resA)
if resB not in self.interface:
self._add_residue(resB)
# Accessible surface area calculated for each residue
# if naccess setup on user computer and rsa_calculation
# argument is TRUE
if rsa_calculation and os.system('which naccess') == 0:
rsa_pairs=self._rsa_calculation(model, chain_list, rsa_threshold)
for res in rsa_pairs:
if res not in self.interface:
self._add_residue(res)
self._secondary_structure(model)
#interface=uniq_pairs
self.interface.uniq_pairs=uniq_pairs
for residue in structure.get_residues():
if residue.resname == args.ligand:
ligand = residue
break
if not ligand:
print('[!!] Ligand residue \'{0}\' not found in structure'.format(args.ligand), file=sys.stderr)
sys.exit(1)
# Calculate center of mass of the ligand
ligand_com = map(lambda x: sum(x)/len(x), zip(*[at.coord for at in ligand]))
ligand_com = np.asarray(ligand_com, dtype=np.float32)
# Calculate neighbors considering only aminoacid/nucleotide atoms (excl. waters, other ligands, etc)
sel_atoms = [at for at in structure.get_atoms() if at.parent.id[0] == ' ']
ns = NeighborSearch(sel_atoms)
neighbors = ns.search(ligand_com, 10.0, level='R') # 10A radius, return residues
# Calculate residue closer to each ligand atom and the respective distance
ligand_atoms = ligand.child_list
min_dist_list, _seen = [], set()
for l_at in ligand_atoms:
distances = []
for residue in neighbors:
for r_at in residue:
distances.append((r_at, l_at, r_at - l_at))
distances.sort(key=lambda x: x[-1])
min_dist = distances[0]
# One restraint per residue to keep the number of restraints small
def get_interacting_residues(model, r_cutoff=5, skip_hetatm_chains=True):
"""Return residue-residue interactions between all chains in `model`.
Parameters
----------
model : biopython.Model
Model to analyse.
Returns
-------
dict
A dictionary of interactions between chains i (0..n-1) and j (i+1..n).
Keys are (chain_idx, chain_id, residue_idx, residue_resnum, residue_amino_acid) tuples.
(e.g. (0, 'A', 0, '0', 'M'), (0, 1, '2', 'K'), ...)
Values are a list of tuples having the same format as the keys.
Examples
--------
You can reverse the order of keys and values like this::
complement = dict()
for key, values in get_interacting_chains(model):
for value in values:
complement.setdefault(value, set()).add(key)
You can get a list of all interacting chains using this command::
{(key[0], value[0])
for (key, values) in get_interacting_chains(model).items()
for value in values}
"""
from Bio.PDB import NeighborSearch
interactions_between_chains = dict()
# Chain 1
for chain_1_idx, chain_1 in enumerate(model):
if skip_hetatm_chains and chain_is_hetatm(chain_1):
message = (
"Skipping chain_1 with idx {} because it contains only hetatms."
.format(chain_1_idx)
)
logger.debug(message)
continue
chain_1_residue_ids = get_aa_residues(chain_1)
# Chain 2
for j, chain_2 in enumerate(model.child_list[chain_1_idx + 1:]):
chain_2_idx = chain_1_idx + 1 + j
if skip_hetatm_chains and chain_is_hetatm(chain_2):
message = (
"Skipping chain_2 with idx {} because it contains only hetatms."
.format(chain_2_idx)
)
logger.debug(message)
continue
chain_2_residue_ids = get_aa_residues(chain_2)
ns = NeighborSearch(list(chain_2.get_atoms()))
# Residue 1
for residue_1 in chain_1:
try:
residue_1_idx = chain_1_residue_ids.index(residue_1.id)
except ValueError:
continue
residue_1_resnum = str(residue_1.id[1]) + residue_1.id[2].strip()
residue_1_aa = convert_aa(residue_1.resname, quiet=True)
residue_1_key = (
chain_1_idx, chain_1.id, residue_1_idx, residue_1_resnum, residue_1_aa
)
interacting_residues = set()
for atom_1 in residue_1:
interacting_residues.update(ns.search(atom_1.get_coord(), r_cutoff, 'R'))
# Residue 2
interacting_residue_ids = []
for residue_2 in interacting_residues:
try:
residue_2_idx = chain_2_residue_ids.index(residue_2.id)
except ValueError:
continue
residue_2_resnum = str(residue_2.id[1]) + residue_2.id[2].strip()
residue_2_aa = convert_aa(residue_2.get_resname(), quiet=True)
residue_2_key = (
chain_2_idx, chain_2.id, residue_2_idx, residue_2_resnum, residue_2_aa
)
interacting_residue_ids.append(residue_2_key)
if interacting_residue_ids:
interactions_between_chains\
.setdefault(residue_1_key, set())\
.update(interacting_residue_ids)
return interactions_between_chains