def analyze_dihedral(self):
"""
Deprecated. Please use class Dihedral_Analisys
"""
angles = list()
cov_atm_lig = self.ligand.child_dict[self.ligand_dict['cov_atm']]
##dihedral between CA < CB < SG < ligand
angle_1 = math.degrees(
bp.calc_dihedral(self.covalent_res.child_dict['CA'].get_vector(),
self.covalent_res.child_dict['CB'].get_vector(),
self.covalent_atm_res.get_vector(),
cov_atm_lig.get_vector()))
angles.append(angle_1)
##all dihedral of CB < SG < ligand-covalent-atom < other ligand atoms
ns = bp.NeighborSearch(list(self.ligand.get_atom()))
neigh = ns.search(cov_atm_lig.get_coord(), 2)
neigh = filter(lambda x: x.name != self.ligand_dict['cov_atm'],
neigh) # removes the atom itself
for i in neigh:
ang = math.degrees(
bp.calc_dihedral(
self.covalent_res.child_dict['CB'].get_vector(),
self.covalent_atm_res.get_vector(),
cov_atm_lig.get_vector(), i.get_vector()))
angles.append(ang)
open('/'.join([self.path, DIHEDRAL_OUTPUT]), 'w').write(
reduce(lambda x, ang: ' '.join([x, str(ang)]), angles, ''))
def annotate_fallback(chain_list):
"""
If neither DSSR nor MC-Annotate are available, we use an ad-hoc implementation of canonical
basepair detection as fallback.
This does not work well for missing atoms or modified residues.
"""
kdtree = bpdb.NeighborSearch(
[atom for chain in chain_list for atom in chain.get_atoms()])
pairs = kdtree.search_all(10, "R")
basepairs = {}
# Sorted, so conflicting basepairs are deterministically solved
for res1, res2 in sorted(pairs):
if res1.resname.strip() not in RNA_RESIDUES or res1.id[0].startswith(
"H_"):
continue
if res2.resname.strip() not in RNA_RESIDUES or res2.id[0].startswith(
"H_"):
continue
labels = {res1.resname.strip(), res2.resname.strip()}
try:
is_bp = is_basepair_pair(res1, res2)
if is_bp:
res1_id = fgr.resid_from_biopython(res1)
res2_id = fgr.resid_from_biopython(res2)
if res1_id in basepairs:
warnings.warn("More than one basepair detected for {}."
" Ignoring {}-{} because {}-{} is already"
" part of the structure".format(
res1_id, res1_id, res2_id, res1_id,
basepairs[res1_id]))
continue
if res2_id in basepairs:
warnings.warn("More than one basepair detected for {}."
" Ignoring {}-{} because {}-{} is already"
" part of the structure".format(
res2_id, res2_id, res1_id, res2_id,
basepairs[res2_id]))
continue
basepairs[res1_id] = res2_id
basepairs[res2_id] = res1_id
except KeyError as e:
log.debug("Missing atom %s. %s has atoms %s, %s has atoms %s", e,
res1, res1.child_dict, res2, res2.child_dict)
pass
seq_ids = []
for chain in sorted(chain_list, key=lambda x: x.id):
for residue in chain:
seq_ids.append(fgr.resid_from_biopython(residue))
bpseq = ""
chain_dict = {c.id: c for c in chain_list}
for i, seqid in enumerate(seq_ids):
if seqid in basepairs:
bp = seq_ids.index(basepairs[seqid]) + 1
else:
bp = 0
bpseq += "{} {} {}\n".format(
i + 1, chain_dict[seqid.chain][seqid.resid].resname.strip(), bp)
return bpseq, seq_ids
def get_atom_neighbors(self, atom, atoms, cov_dist=2):
ns = bp.NeighborSearch(atoms)
neighbors = ns.search(atom.get_coord(), cov_dist)
covalent = [
atm for atm in neighbors
if not (atm.name.startswith('H') or atm.name == atom.name)
] #filter Hydrogens & the atom itself
return covalent
def is_contact(res_1, other_atoms, cutoff):
for atom in res_1:
ns = PDB.NeighborSearch(other_atoms)
center = atom.get_coord()
neighbors = ns.search(center, cutoff) # 5.0 for distance in angstrom
residue_list = PDB.Selection.unfold_entities(neighbors,
'R') # R for residues
if len(residue_list) > 0:
return True
return False
def interchain_contacts(struct):
all_atoms = bpdb.Selection.unfold_entities(struct, 'A')
ns = bpdb.NeighborSearch(all_atoms)
pairs = ns.search_all(2.8)
ic_pairs = []
for (a1, a2) in pairs:
if a1.parent.parent != a2.parent.parent:
ic_pairs += [(a1, a2)]
return ic_pairs
def noncovalent_distances(chain, cutoff=0.3):
'''
Print out the distances between all non-covalently bonded atoms
which are closer than cutoff to each other.
:param chain: The Bio.PDB chain.
:param cutoff: The maximum distance
'''
all_atoms = bpdb.Selection.unfold_entities(chain, 'A')
ns = bpdb.NeighborSearch(all_atoms)
contacts = ns.search_all(cutoff)
return [ftuv.magnitude(c[1] - c[0]) for c in contacts if not is_covalent(c)]
def num_noncovalent_clashes(chain):
'''
Check if a chain has non-covalent clashes. Non-covalent clashes are found
when two atoms that aren't covalently linked are within 1.8 A of each other.
:param chain: The chain to evaluate
:param return: The number of non-covalent clashes.
'''
all_atoms = bpdb.Selection.unfold_entities(chain, 'A')
ns = bpdb.NeighborSearch(all_atoms)
contacts = ns.search_all(1.9)
return len([c for c in contacts if not is_covalent(c)])
def neighbourhood(atoms,radius,show):
pairs=set()
for r1,r2 in PDB.NeighborSearch(atoms).search_all(radius,'A'):
chain_list_1 = PDB.Selection.unfold_entities([r1], 'C')
chain_list_2 = PDB.Selection.unfold_entities([r2], 'C')
res_list_1 = PDB.Selection.unfold_entities([r1], 'R')
res_list_2 = PDB.Selection.unfold_entities([r2], 'R')
if chain_list_1[0].get_id() != chain_list_2[0].get_id():
printverbose(sys.stdout,show,(" %s %s (Atom %s) (chain %s) contact with %s %s (Atom %s) (chain %s) \n"%
(res_list_1[0].get_resname(),res_list_1[0].get_id()[1],r1.get_id(),chain_list_1[0].get_id(),
res_list_2[0].get_resname(),res_list_2[0].get_id()[1],r2.get_id(),chain_list_2[0].get_id())))
pairs.add( ( res_list_1[0], chain_list_1[0].get_id(), res_list_2[0], chain_list_2[0].get_id() ) )
pairs.add( ( res_list_2[0], chain_list_2[0].get_id(), res_list_1[0], chain_list_1[0].get_id() ) )
return pairs
def _find_covalent_atom(self):
atoms = list(self.ligand.get_atom())#ligand atoms
for res in self.structure.get_residues():
if res.id[0] == ' ':#add all residues atoms
atoms = atoms + res.get_list()
ns = bp.NeighborSearch(atoms)
neighbors = ns.search_all(self.covalent_distance)
covalent = [cpl for cpl in neighbors if (cpl[0].parent.id[0] == ' ' and cpl[1].parent.id == self.ligand.id) or (cpl[0].parent.id == self.ligand.id and cpl[1].parent.id[0] == ' ' )] # filter only intermolecule bonds
covalent = [cpl for cpl in covalent if not (cpl[0].name.startswith('H') or cpl[1].name.startswith('H'))] #filter Hydrogens
cov_cpl = min(covalent, key = lambda cpl: abs(cpl[0] - cpl[1]))
## decouple ligand & receptor covalent atoms
self.receptor_cov_atom = [atom for atom in cov_cpl if atom.parent.id[0] == ' '][0]
self.ligand_cov_atom = [atom for atom in cov_cpl if atom.parent.id == self.ligand.id][0]
def enumerate_interactions_kdtree(model):
relevant_atoms = [
a for a in model.get_atoms() if a.name[0] in ["C", "N", "O"]
]
if not relevant_atoms:
return set()
kdtree = bpdb.NeighborSearch(relevant_atoms)
pairs = kdtree.search_all(6, "A")
res_pair_list = set()
for a1, a2 in pairs:
if a1.name not in all_side_chains and a2.name not in all_side_chains:
continue
p1 = a1.get_parent()
p2 = a2.get_parent()
if p1.id == p2.id:
continue
elif p1 < p2:
res_pair_list.add((p1, p2))
else:
res_pair_list.add((p2, p1))
interacting_residues = set()
for res1, res2 in res_pair_list:
rna_res = None
other_res = None
if res1.resname.strip(
) in RNA_RESIDUES and not res1.id[0].startswith("H_"):
rna_res = res1
else:
other_res = res1
if res2.resname.strip(
) in RNA_RESIDUES and not res2.id[0].startswith("H_"):
rna_res = res2
else:
other_res = res2
if rna_res is None or other_res is None:
continue
log.debug("%s(chain %s) and %s(chain %s, resname %s) are close",
rna_res, rna_res.parent.id, other_res, other_res.parent.id,
other_res.resname.strip())
# Only consider C and N. So no ions etc
if any(a.name[0] in ["C", "N"] for a in other_res.get_atoms()):
interacting_residues.add(rna_res)
else:
log.debug("but %s has wrong atoms %s", other_res,
list(a.name for a in other_res.get_atoms()))
log.debug("Interacting: %s", interacting_residues)
return interacting_residues
def are_clashing(chain_one, chain_two, max_clashes=300, contact_distance=1.0):
"""
Compares the CA atoms of two chains and checks for clashes according to the
contact distance.
Returns a boolean.
"""
atoms_one = [atom for atom in chain_one.get_atoms() if atom.get_id() == 'CA']
atoms_two = [atom for atom in chain_two.get_atoms() if atom.get_id() == 'CA']
ns = pdb.NeighborSearch(atoms_one)
clashes = 0
for atom_two in atoms_two:
for atom in ns.search(atom_two.get_coord(), contact_distance, 'A'):
clashes += 1
if clashes == max_clashes:
if main.options.verbose:
sys.stderr.write("Clash Found!\n")
return True
return False
def has_clashes(move_atoms, model):
"""Compares the atoms backbone atoms of the moving chain with the backbone atoms of the model"""
backbone = {"CA", "C1\'"}
chain_atoms = [atom for atom in move_atoms
if atom.id in backbone] # Gets only the backbone atoms
model_atoms = [atom for atom in model.get_atoms() if atom.id in backbone]
ns = PDB.NeighborSearch(
model_atoms
) # Generates a neigbour search tree to speed up distance calculations
clashes = 0
for atom in chain_atoms:
clashes += bool(ns.search(
atom.coord,
2)) # If this atom shows clashes, add 1 to the clashes counter
if clashes / len(
chain_atoms
) >= 0.03: # If more than 3% of atoms show clashes return yes
return True
else: # Otherwise return no
return False
def clashing_filter(chain_one, chain_two, max_clashes=30, contact_distance=2.0):
"""
Given a maximum number of clashes and a contact distance, the function checks for clashes
in the alpha carbons (CA) of two different chains (chain_one and chain_two).
Returns true if the maximum number of clashes is reached, else returns false.
"""
atoms_one = [atom for atom in chain_one.get_atoms() if atom.get_id() == 'CA' or atom.get_id() =="C5'"]
atoms_two = [atom for atom in chain_two.get_atoms() if atom.get_id() == 'CA' or atom.get_id() =="C5'"]
ns = pdb.NeighborSearch(atoms_one)
clashes = 0
for atom_two in atoms_two:
for atom in ns.search(atom_two.get_coord(), contact_distance, 'A'):
clashes += 1
if clashes == max_clashes:
if main.options.verbose:
sys.stderr.write("Clash Found!\n")
return True
return False
def has_clashes(move_atoms, model, clash_dist):
"""Compares the backbone atoms of the moving chain with the backbone atoms of the model. Returns a boolean value based on the presence and abundancy of clashes.
Keyword arguments:
move_atoms -- set of atoms from the chain that is potentially going to be added to the macrocomplex
model -- current macrocomplex model in construction
clas_dist -- minimum clash distance between 2 atoms. The default minimum is 2 A
Considerations:
Alpha carbons are used as backbone atoms.
Threshold: if 3% of atoms of the chain that is been comparing to the model shows clashes, <True> is returned. Hence, that chain won't be added to the model."""
backbone = {"CA", "C1\'"}
chain_atoms = [atom for atom in move_atoms if atom.id in backbone]
model_atoms = [atom for atom in model.get_atoms() if atom.id in backbone]
ns = PDB.NeighborSearch(model_atoms)
clashes = 0
for atom in chain_atoms:
clashes += bool(ns.search(atom.coord, float(clash_dist)))
if clashes / len(chain_atoms) >= 0.03:
return True
else:
return False
def HydrogenBondFilter(receptor_atoms, molecule,affinity):
ligand_id = os.path.basename(molecule).replace(".pdb","")
ligand = parser.get_structure(ligand_id, molecule)
checklist = []
global targethydrogen, use_angle
### Loop through list of all atoms we know contribute to hydrogen bonds from the protein. At least one is essential for function so poses only pass if they form at least one with the defined targets
for target in targets:
target_res = target.get_parent()
protein_hydrogen = [atom for atom in bp.Selection.unfold_entities(target.get_parent(),"A") if ("H" in atom.get_id()) and target - atom <= 1.1]
if protein_hydrogen:
use_angle = True
targethydrogen = protein_hydrogen[0]
else:
use_angle = False
PosePassInfo = []
### Grab the hydrogen bond info of each pose of each drug. Unfold the atoms of the drug and the combine it with the atoms of the the receptor. Then Search is performed on close atoms
for posenum, pose in enumerate(ligand):
pose_atoms = bp.Selection.unfold_entities(pose, "A")
all_atoms = receptor_atoms + pose_atoms
ns = bp.NeighborSearch(all_atoms)
### Search for atoms that are within X angstrom of defined target. Then check whether found atoms are ligand atoms and if they are capable of hydrogen bond formation
close_atoms = ns.search(target.coord, 3.6)
Hbond_LigAtoms = [atom for atom in close_atoms if (atom in pose_atoms) and (atom.get_name().translate(None,"0123456789") in Filter_atoms)]
### Check if H-bond is within defined parameters per pose and put that in a list per drug. Poses that pass have their ligand efficiency written, poses that don't have 0.0
Pass = HBondCheck(Hbond_LigAtoms,pose_atoms,posenum,affinity,target)
PosePassInfo.append(Pass)
### In case of multiple possible hydrogen bonds a list is kept of poses that have passed at least one of them
if checklist:
for posenum, entry in enumerate(PosePassInfo):
if (entry != 0.0) and (checklist[posenum] == 0.0):
checklist[posenum] = entry
else:
checklist = PosePassInfo
return checklist
def process_queue(id):
pdir = "/home/kenneth/proj/proMin/proteins/hagai/2018/pdbs"
df = dfHag.iloc[id]
# print(df)
pdbFile = os.path.join(pdir, df.pdb + ".pdb")
#Create a PDBParser object
parser = PDBParser() #PERMISSIVE = 0 will list all errors with PDB file
STR = parser.get_structure(df.pdb, os.path.join(pdbFile))
envComp = ""
atoms = STR[0][df.chain].get_atoms()
atomID = ""
for atom in atoms:
# print(atom.get_fullname())
# print(atom.get_fullname().replace(' ',''))
if atom.get_name() == df.atomName:
resID = atom.get_parent().get_full_id()
# print(resID,'blah')
if (resID[3][0].find(df.resName) > -1) and (int(resID[3][1])
== df.resNum):
atomID = resID[3]
if atomID == '':
print(df.coord, resID, 'blank')
continue
chain = STR[0][df.chain]
atom = chain[atomID][atom.get_name()]
atom_list = PDB.Selection.unfold_entities(STR[0], 'A')
ns = PDB.NeighborSearch(atom_list)
atoms = ns.search(atom.get_coord(), 2.8, level='A')
# print(atoms)
envComp = df.coord + "," + df.resName + "," + ",".join(
map(str, list(resID))) + "," + get_environment(atoms)
# print(envComp)
process_queue.q.put(envComp)
return envComp
def Collision_Check(model_atom_list, addition_atom_list, radius):
"""Return a list of tuple containing the residue id interacting and the coordinates in conflict
Input:
-model = list of atoms take as reference for the collitions
-addition = list of atoms being us to check against the model
-radius = integrer required for become the empty radious (Amstrongs) around each atom
Output:
-collisions_list = list of tuples of coordinates where atoms collide
"""
model_ns = pdb.NeighborSearch(model_atom_list)
collisions_list = []
for atom in addition_atom_list:
collision_list = model_ns.search(atom.get_coord(), radius, "R")
collisions_list.extend([
tuple([
str(atom.get_parent().get_id()[1]),
str(x.get_id()[1]),
str(atom.get_coord())
]) for x in collision_list
])
return collisions_list
STR = get_cif_STR(cifPath)
try:
RES = STR[0][chain][("H_" + LIG, int(resID), " ")]
except Exception as e:
print(" ", i.Microenvironment_ID, ("H_" + LIG, int(resID), " "))
print(e)
continue
resName = RES.get_resname()
feATOMS = [atom for atom in RES.get_atoms() if "FE" in atom.get_name()]
atom_list = PDB.Selection.unfold_entities(STR, "A")
ns = PDB.NeighborSearch(atom_list)
for atom in feATOMS:
atomName = atom.get_name()
feAtom = {
"ligName": resName,
"ligChain": chain,
"resID": resID,
"ligAtom": atomName,
"Metal": i.Metal,
"ligCoords": atom.get_coord()
}
annotations = {attr: getattr(i, attr) for attr in hagaiNames}
neighbors = ns.search(atom.get_coord(), 3, level='A')
neigh_dict = {
i.Microenvironment_ID: {
directory = sys.argv[1] # Folder with the chain pdbs
distance = float(sys.argv[2])
directory_out = os.path.join(directory, 'pairs') # Folder where pairs will be saved
PDBparser = PDB.PDBParser(QUIET=True)
done = []
for pdb_1 in list(filter(lambda x: x.endswith('.pdb'), os.listdir(directory))):
for pdb_2 in list(filter(lambda x: x.endswith('.pdb'), os.listdir(directory))):
if (pdb_1 == pdb_2):
continue # Not to duplicate itself
elif not (pdb_1, pdb_2) in done or not (pdb_2, pdb_1) in done: # Not to duplicate pairs
done.append((pdb_1, pdb_2))
structure1 = PDBparser.get_structure(pdb_1[:-4], os.path.join(directory,pdb_1))
structure2 = PDBparser.get_structure(pdb_2[:-4], os.path.join(directory,pdb_2))
for chain in structure2.get_chains():
atoms = list(chain.get_atoms())
ns = PDB.NeighborSearch(atoms) # An object to search chains near an atom
for target_atom in structure1.get_atoms():
near = ns.search(target_atom.coord, distance)
if(near and (pdb_1[-5:-4],pdb_2[-5:-4]) not in done and (pdb_2[-5:-4],pdb_1[-5:-4]) not in done): # If there's an atom near, they interact
print('Joining chain {a} and {b}'.format(a=pdb_1[-5:-4], b=pdb_2[-5:-4]))
done.append((pdb_1[-5:-4],pdb_2[-5:-4]))
io = PDB.PDBIO()
structure1[0].child_list += [chain for chain in structure2.get_chains()] # Joins all the chains
io.set_structure(structure1)
if(not os.path.exists(os.path.join(directory_out))):
os.mkdir(os.path.join(directory_out))
p = re.search('^[A-Za-z0-9]+_([A-Za-z0-9]+).pdb$', pdb_1 ).group(1)
q = re.search('^[A-Za-z0-9]+_([A-Za-z0-9]+).pdb$', pdb_2 ).group(1)
io.save(os.path.join(directory_out,pdb_1[:-5]+p+q+'.pdb'))
print(os.path.join(directory,pdb_1[:-4]+'.fa'))
if(os.path.exists(os.path.join(directory,pdb_1[:-4]+'.fa'))) and os.path.exists(os.path.join(directory,pdb_2[:-4]+'.fa')):
def process_queue(id):
# print(id)
# print(c)
# print(i)
# print("i:", i)
# iDict = list_tuples_to_dict(chunks)
# print(type(iDict))
# print(iDict.keys())
# iDict = list_to_dict(i)
# print(id)
# pbar.update(1)
series = feDF.loc[feDF['Microenvironment_ID'] == id].to_dict()
# print(list(series['Metal'].values())[0])
# annotations = {attr:series[attr] for attr in hagaiNames}
pdb, LIG, chain, resID, function = re.split('[._]', id)
# # print(pdb,LIG,chain,resID,function,i.Metal)
cifPath = os.path.join(pdir, pdb + ".cif")
STR = get_cif_STR(cifPath)
try:
RES = STR[0][chain][("H_" + LIG, int(resID), " ")]
except Exception as e:
print(" ", series['Microenvironment_ID'],
("H_" + LIG, int(resID), " "))
print(e)
# continue
resName = RES.get_resname()
feATOMS = [atom for atom in RES.get_atoms() if "FE" in atom.get_name()]
atom_list = PDB.Selection.unfold_entities(STR[0], "A")
ns = PDB.NeighborSearch(atom_list)
#return dataframe
output = []
for atom in feATOMS:
atomName = atom.get_name()
feAtom = {
"id_hagai": id,
"ligName": resName,
"ligChain": chain,
"resID": resID,
"ligAtom": atomName,
"Metal": list(series['Metal'].values())[0],
"ligX": atom.get_coord()[0],
"ligY": atom.get_coord()[1],
"ligZ": atom.get_coord()[2]
} #,"micro_annotations":annotations}
neighbors = ns.search(atom.get_coord(), 3, level='A')
# output.append((id,len(neighbors)))
c = 0
for neigh in neighbors:
if neigh - atom != 0:
feAtom["N_" + str(c) +
"_resName"] = neigh.get_parent().get_resname()
feAtom["N_" + str(c) + "_atomName"] = neigh.get_name()
feAtom["N_" + str(c) +
"_resNum"] = neigh.get_parent().get_id()[1]
feAtom["N_" + str(c) + "_x"] = neigh.get_coord()[0]
feAtom["N_" + str(c) + "_y"] = neigh.get_coord()[1]
feAtom["N_" + str(c) + "_z"] = neigh.get_coord()[2]
feAtom["N_" + str(c) + "_dist"] = neigh - atom
feAtom["N_" + str(c) +
"_dx"] = neigh.get_coord()[0] - atom.get_coord()[0]
feAtom["N_" + str(c) +
"_dy"] = neigh.get_coord()[1] - atom.get_coord()[1]
feAtom["N_" + str(c) +
"_dz"] = neigh.get_coord()[2] - atom.get_coord()[2]
count += 1
output.append(feAtom)
# neigh_dict = {id:{"feAtom":feAtom,"Microenvironment_annotations":annotations,"Neighbors":{"Neighbor_"+str(idx): {"N_"+str(idx)+"_resName":neigh.get_parent().get_resname(),"N_"+str(idx)+"_atomName":neigh.get_name(),"N_"+str(idx)+"_resNum":neigh.get_parent().get_id()[1],"N_"+str(idx)+"_atomCoord":neigh.get_coord(),"N_"+str(idx)+"_distance":neigh-atom,"N_"+str(idx)+"_dx":neigh.get_coord()[0]-atom.get_coord()[0],"N_"+str(idx)+"_dy":neigh.get_coord()[1]-atom.get_coord()[1],"N_"+str(idx)+"_dz":neigh.get_coord()[2]-atom.get_coord()[2]} for idx, neigh in enumerate(neighbors) if neigh-atom != 0 }}}
# print(neigh_dict)
process_queue.q.put(output)
return output
def is_steric_clash(structure, rotating_chain, distance_for_clash=2.5):
"""Check if there is a steric clash between a rotating chain and current structure. Return False (no clash), 1 (clash between two different chains) or 2 (same chain). Also returns the ids of the chains in structure that are clashing
Keyword arguments:
structure -- whole structure
rotating_chain -- chain to be analyzed with respect to structure
distance_for_clash -- threshold to consider clash between atoms. Default = 2.5 (Armstrongs)
Clash criteria: at least 20 of the atoms are at a lower distance than distance_for_clash
Same chain criteria: RMSD between them <= 3.0 """
# initialize the neighbor search
NS = pdb.NeighborSearch(list(structure.get_atoms()))
clashing_chains = set() # the set of clashing chains (in structure)
n_clashes = 0 # the number of clashes
for at in rotating_chain.get_atoms():
neighbors = NS.search(at.get_coord(), distance_for_clash)
if len(neighbors) > 0:
for neigh in neighbors:
clashing_chains.add(neigh.get_parent().get_parent().id)
n_clashes += 1
if len(clashing_chains) > 1 and n_clashes > 20:
# a clash against different chains:
val_to_return = 1
elif len(clashing_chains) == 1 and n_clashes > 20 and rotating_chain.id[
1] == list(clashing_chains)[0][1]:
# a clash MAYBE because you are trying to superimpose something in the place it was already
# define the clashing chain:
clash_chain = structure[0][list(clashing_chains)[0]]
res_chain1 = list(clash_chain.get_residues())
res_chain2 = list(rotating_chain.get_residues())
# so first we obtain a list of the common residues
common_res_s1 = get_list_of_common_res(res_chain1, res_chain2)
common_res_s2 = get_list_of_common_res(res_chain2, res_chain1)
# then we obtain a list of atom objects to use it later
common_atoms_s1 = np.array([
list(x.get_coord())
for x in get_atom_list_from_res_list(common_res_s1)
])
common_atoms_s2 = np.array([
list(x.get_coord())
for x in get_atom_list_from_res_list(common_res_s2)
])
# debug the atoms provided
if len(common_atoms_s1) == len(common_atoms_s2):
RMSD = rmsd.kabsch_rmsd(common_atoms_s1, common_atoms_s2)
else:
RMSD = 1000 # like being different
if RMSD <= 3.0:
# it is the same chain
val_to_return = 2
else:
# it is another chain or the same with different structure
val_to_return = 1
elif n_clashes > 20:
# it is ine chain and the previous conditions are not fullfilled
val_to_return = 1
else:
# no clash
val_to_return = False
if val_to_return:
return val_to_return, clashing_chains
else:
return val_to_return, None
def count_receptor_contacts(cls, paths, complexname, receptor_chain,
ligand_chain, nwindows, dbf, model_db_file,
query_dict):
"""
Count number of paths contacting each receptor
"""
wd = os.path.dirname(model_db_file)
pdb_kwargs = dict(complexname=complexname,
receptor_chain=receptor_chain,
ligand_chain=ligand_chain,
nwindows=nwindows)
pdbwindowid = "{complexname}{receptor_chain}{ligand_chain}{nwindows}".format(
**pdb_kwargs)
outfile = os.path.join(wd, "{0}_path_contacts.pdb".format(pdbwindowid))
path_score_file = os.path.join(
wd, "{0}_receptor_occupancy.csv".format(pdbwindowid))
if not shared.missing(outfile) and not shared.missing(path_score_file):
logging.debug("%s exists", outfile)
return
cutoff = 5.0
residue_fmt = "{chain}_{resname}{resid[1]}"
def make_key(residue):
__, __, chainid, residueid = residue.get_full_id()
return residue_fmt.format(chain=chainid,
resname=residue.get_resname(),
resid=residueid)
# Drop window1 (modelid) column
orig_window_vars = [
x for x in paths.columns.values.tolist() if x.startswith("window")
]
for window_var in orig_window_vars:
paths = paths.drop(window_var, axis=1)
# Get model filepaths for paths
filepaths = cls.get_paths(paths[['pathsid']],
dbf=dbf,
model_db_file=model_db_file,
query_dict=query_dict)
window_vars = [
x for x in filepaths.columns.values.tolist()
if x.startswith("window")
]
get_files = lambda row: [row[w] for w in window_vars]
parser = PDB.PDBParser(QUIET=True)
# Remove hydrogens
get_structure = lambda x: parser.get_structure(
os.path.splitext(os.path.basename(x))[0], shared.strip_h(x))
modelid = 0
receptor_contacts = collections.defaultdict(set)
for x, row in filepaths.iterrows():
pathsid = row['pathsid']
path_files = get_files(row)
for fn in path_files:
structure = get_structure(fn)
atoms = [
atom for chain in structure[modelid] for residue in chain
for atom in residue
]
if not atoms:
raise PlotPathsError("No atoms in %s" % fn)
ns = PDB.NeighborSearch(atoms)
search = ns.search_all(radius=cutoff, level="R")
for res1, res2 in search:
__, __, c1, r1 = res1.get_full_id()
__, __, c2, r2 = res2.get_full_id()
# Skip if chains are both ligand or both receptor
if (c1 == ligand_chain) == (c2 == ligand_chain):
continue
if c1 in receptor_chain:
key = make_key(res1)
elif c2 in receptor_chain:
key = make_key(res2)
else:
raise PlotPathsError("Neither %s nor %s is receptor" %
(c1, c2))
receptor_contacts[key].add(pathsid)
# Convert from defaultdict to normal dict
receptor_contacts = dict(receptor_contacts)
# Count paths contacting each receptor residue
emptyset = set()
# Chains have been combined
r_ch = receptor_chain[0]
# Deliberately using last structure from loop
for residue in structure[modelid][r_ch]:
key = make_key(residue)
mypaths = receptor_contacts.get(key, emptyset)
count = len(mypaths)
for atom in residue:
atom.set_bfactor(count)
# Write out structure with b-factor
#structure[modelid].detach_child(ligand_chain)
#io = PDB.PDBIO()
#io.set_structure(structure)
#io.save(outfile)
# Count receptor contacts for each path
path_score_dict = collections.defaultdict(int)
for contacts in receptor_contacts.itervalues():
n_contacts = len(contacts)
for pathid in contacts:
path_score_dict[pathid] += n_contacts
path_score_df = pd.DataFrame(path_score_dict.items(),
columns=["pathid", "occupancyscore"])
path_score_df.to_csv(path_score_file, index=False)
def __init__(self, protein_pdbpath: str, distance: float = 1.5):
parser = PDB.PDBParser(QUIET=True, PERMISSIVE=True)
s = parser.get_structure('protein', protein_pdbpath)
self.kd = PDB.NeighborSearch(list(s.get_atoms()))
self.radius = distance
HSEB = PDB.HSExposureCB(s)
HSEB_dict = HSEB.property_dict
HSEB_keys = HSEB.property_keys
HSEB_list = HSEB.property_list
depth = PDB.ResidueDepth(s)
dep_dict = depth.property_dict
dep_keys = depth.property_keys
dep_list = depth.property_list
dssp = PDB.DSSP(s, "3skpFH.pdb")
dssp_dict = dssp.property_dict
nb_dict = {}
nb = PDB.NeighborSearch(ca_list)
for a in ca_list:
t = nb.search(a.get_coord(), 8)
aa = a.get_parent()
aa_id = (aa.get_parent().get_id(), aa.get_id())
nb_dict[aa_id] = t
dic = {}
dic["res_id"] = []
for a in aa_list:
dic["res_id"].append(a.get_id())
dic["res_name"] = []
for a in aa_list:
dic["res_name"].append(a.get_resname())
def calculate_sphere_variance(structure, chain, md_df, mapping):
"""
Calculates the sphere variance and returns the corresponding statistic
Works on a single seq/structure
:param structure: Bio.PDB structure object
:param chain: Chain ID as str
:param md_df: Dataframe containing sensitivity (masked_dump file)
:param mapping: Dict mapping sensitivity coordinates to PDB coordinates
:return: A data frame containing the statistics per GO
"""
# get the sequence:
seq_len = len(md_df['AA'][1:].tolist())
# get the list of gos:
gos = {c.split('_')[0] for c in md_df.columns if c.startswith('GO:')}
allowed = set(mapping.values())
# get the residiues:
residues = structure[0][chain]
sphere_variances_per_go = {go: [] for go in gos}
mean_variances = {go: 0. for go in gos}
mean_fakes = {go: np.zeros(bs) for go in gos}
p_vals = {}
emp_vals = {}
bs_mean = {}
bs_median = {}
# for residue in protein
n_clean = 0
for seq_idx in range(-1, seq_len):
if seq_idx == -1:
# set all to nan:
for go in gos:
sphere_variances_per_go[go].append(np.nan)
elif seq_idx not in mapping:
for go in gos:
sphere_variances_per_go[go].append(np.nan)
continue
else:
try:
res = residues[mapping[seq_idx]]
ca = res['CA']
except:
for go in gos:
sphere_variances_per_go[go].append(np.nan)
continue
# get the neighbors:
glob_resseq = mapping[seq_idx][1]
center = ca.get_coord()
search = pdb.NeighborSearch(atom_list=list(structure.get_atoms()))
neighbors = search.search(center=center, radius=radius, level="R")
n_clean_neighbors = 0
imps = {x: [] for x in gos}
fakes = {x: [] for x in gos}
for n in neighbors:
het_atm, resseq, _ = n.id
if het_atm.strip() == '' and resseq >= 0:
if abs(resseq -
glob_resseq) >= exclude and n.id in allowed:
for go in gos:
try:
imps[go].append(n.sensitivity[go])
fakes[go].append(n.fake[go])
n_clean_neighbors += 1
except AttributeError:
break
if n_clean_neighbors == 0:
for go in gos:
sphere_variances_per_go[go].append(np.nan)
continue
for go in gos:
current_var = np.nanvar(imps[go])
current_fake = np.nanvar(fakes[go], axis=0)
sphere_variances_per_go[go].append(current_var)
mean_variances[go] += np.nan_to_num(current_var)
mean_fakes[go] += np.nan_to_num(current_fake)
n_clean += 1
# normalize the variances:
if n_clean > 1:
for go in gos:
mean_variances[go] /= n_clean
mean_fakes[go] /= n_clean
# do the test
_, p_vals[go] = stats.ttest_1samp(mean_fakes[go],
popmean=mean_variances[go],
nan_policy="omit")
emp_vals[go] = (
1 + np.sum(mean_fakes[go] <= mean_variances[go])) / (1 + bs)
bs_mean[go] = np.nanmean(mean_fakes[go])
bs_median[go] = np.nanmedian(mean_fakes[go])
# now convert the sphere vars to df and append:
# sphere_variances_df = pd.DataFrame.from_dict(sphere_variances_per_go, orient='columns')
# sphere_variances_df.columns = ['%s_%s_svar' % (c, sens_to_use) for c in sphere_variances_per_go] # as the keys were go terms
# md_df_svar = pd.concat([md_df, sphere_variances_df], axis=1)
record = {
"mean_variances": mean_variances,
"p_vals": p_vals,
"emp_vals": emp_vals,
"bs_mean": bs_mean,
"bs_median": bs_median
}
df_svar = pd.DataFrame.from_records(record)
return df_svar
def build_complex(threshold, distance, stoichiometry, sequences, structures, verbose, initial):
'''
This is the core function of the program.
Takes a list of structures and tries to join them in a single model, taking into account
diferent parameters, such as an intial model, a threshold for the pairwise alignments and
the distance to consider that two chains are in the same position and it must be discarded
(different atoms cannot occupy the same space)
'''
# Initiate empty structure and check some errors
seqs = get_fastas_from_structs(structures, sequences)
# Initializing an empty structure to save the model
full_structure = PDB.Structure.Structure('full')
full_structure.add(PDB.Model.Model('model'))
# We will need to assign ids. For now is limited to up to 64 different chains, but more single characters can be added
ids = _chain_id()
iters = 0
pair_num = len(structures)
# Prepare a dict that saves how many chains of every one in the stoichiometry has the model in construction
if(stoichiometry):
current_number_of_chains = {chain_id:0 for chain_id in set(stoichiometry)}
# We need at least 2 pdbs with an interaction. Can be the same one repeated. This is done to avoid trouble
if(pair_num < 2):
raise ValueError('Needed at least 2 pdbs to superpose')
failed = []
current = 0
done = []
# If an initial structure was given, introduce it into the model
if(initial):
if verbose: print('Initializing complex')
# Obtaining the sequences of the chains
init_seqs = get_fastas_from_structs([initial], sequences)
seqs[full_structure.id] = {}
# For each chain, add it to the model with sequential ids: A->B-> etc
for idx, chain in enumerate(initial.get_chains()):
chain_id = next(ids)
seqs[full_structure.id][chain_id] = init_seqs[initial.id][chain.id]
chain_new = PDB.Chain.Chain(chain_id)
chain_new.child_list = list(chain.get_residues())
model = next(full_structure.get_models())
# If stoichiometry is selected, then count which sequences will be added
if(stoichiometry):
chain_sequence = _get_chain_sequence(chain_new)
for seq in sequences:
alignment = pairwise2.align.globalxx(chain_sequence, seq.seq, one_alignment_only=True)
# This chain corresponds to this stoichiometry chain
if(alignment[0][2]/len(alignment[0][0])> threshold and
current_number_of_chains[seq.id] < stoichiometry[seq.id]):
model.child_list.append(chain_new)
current_number_of_chains[seq.id] += 1
if(stoichiometry):
count = 0
for chain in current_number_of_chains:
if current_number_of_chains[chain] == stoichiometry[chain]:
count += 1
if count == len(stoichiometry):
print('Stoichiometry fullfilled!')
return full_structure
model.child_list.append(chain_new)
done.append(initial.id)
if verbose: print('Starting to build')
# The main loop of the function
while True:
if verbose:
print('Loop #%i' % current)
print('Current number of chains: %i' % len(list(full_structure.get_chains())))
# All the structures where added correctly (should only be for non stoichiometric uses)
if(not stoichiometry and len(done) == len(structures)):
break
structure = structures[current % len(structures)]
if(structure.id in done):
current += 1
continue # If is already in, continue to the next one
count = list(failed.count(element) for element in failed)
# If failed has an element twice, that means that no more available chains could be added
# Therefor, it will start an endless loop, as the structure remain equal no matter which
# of the left structures is trying to be added. We stop it here
if 3 in count: # We are repeating structures
if verbose: print('Some pdbs could not be joined.\nAnd endless loop started and no more chains could be added\nFinishing the model in the current state')
break
if len(list(full_structure.get_chains())) == 0:
if verbose: print('Initializing complex')
seqs[full_structure.id] = {}
for idx, chain in enumerate(structure.get_chains()):
chain_id = next(ids) # Selecting the next id available
seqs[full_structure.id][chain_id] = seqs[structure.id][chain.id] # Assigning the new chains its sequence
# Empty new chain with the new id
chain_new = PDB.Chain.Chain(chain_id)
# Adding the residues to the new chain and adding the chain to the new model
chain_new.child_list = list(chain.get_residues())
model = next(full_structure.get_models())
# Check stoichiometry if needed
if(stoichiometry):
# Need the exact sequence to align
chain_sequence = _get_chain_sequence(chain_new)
# Align with every sequence in the fasta to know which sequence is
for seq in sequences:
alignment = pairwise2.align.globalxx(chain_sequence, seq.seq, one_alignment_only=True)
try:
current_number_of_chains[seq.id]
except KeyError:
# The sequence chain is not in the stoichiometry
raise errors.chain_in_stoic_not_in_fasta(seq.id)
# If homologous and does not surpass the stoichiometry, add it
if(alignment[0][2]/len(alignment[0][0])> threshold and
current_number_of_chains[seq.id] < stoichiometry[seq.id]):
model.child_list.append(chain_new)
current_number_of_chains[seq.id] += 1
count = 0
for chain in current_number_of_chains:
if current_number_of_chains[chain] == stoichiometry[chain]:
count += 1
if count == len(stoichiometry):
print('Stoichiometry fullfilled!')
return full_structure
model.child_list.append(chain_new)
done.append(structure.id)
current += 1
continue
# The Enseble class keeps code ordered and abstracts the alignment
# and superimposition of the code
pair = Ensemble(full_structure, structure)
pair_seqs = (seqs[full_structure.id], seqs[structure.id])
alignment = pair.get_best_alignment(pair_seqs)
# If score is > 0.95, they should be homologs
for align in alignment:
if(align[1] > threshold):
#try:
atoms_of_chains = pair.superimpose(align[0][0], align[0][1])
#except PDB.PDBExceptions.PDBException:
# current += 1
# continue
iters += 1
neighbor = PDB.NeighborSearch(list(full_structure.get_atoms()))
clashes = 0
for chain in atoms_of_chains:
for atom in atoms_of_chains:
for atom in chain.get_atoms():
close_atoms = neighbor.search(atom.get_coord(), distance)
# If there are atoms within 2 angstroms, consider a clash
if len(close_atoms) > 0:
clashes += 1
if (clashes < 10 and pair.rms < 0.05):
# The chain can be added, so let's add it
for chain in atoms_of_chains:
chain_id = next(ids) # Chosing an id
chain2 = PDB.Chain.Chain(chain_id) # An empty chain which will add the residues to join
chain2.child_list += list(chain.get_residues())
model = next(full_structure.get_models())
if(stoichiometry): # Taking the stoichiometry into account
chain_sequence = _get_chain_sequence(chain)
for seq in sequences:
alignment = pairwise2.align.globalxx(chain_sequence, seq.seq, one_alignment_only=True)
# This chain corresponds to this stoichiometry chain
if(alignment[0][2]/len(alignment[0][0])> threshold and
current_number_of_chains[seq.id] < stoichiometry[seq.id]):
model.child_list.append(chain2)
current_number_of_chains[seq.id] += 1
count = 0
for chain in current_number_of_chains:
if current_number_of_chains[chain] == stoichiometry[chain]:
count += 1
if verbose: print('Current number of chains: %i' % sum(current_number_of_chains.values()))
if count == len(stoichiometry):
if verbose: print('Stoichiometry fullfilled!')
return full_structure
else:
model.child_list.append(chain2)
if verbose and stoichiometry: print('Current number of chains: %i' % sum(current_number_of_chains.values()))
done.append(structure.id)
failed = []
break
else:
failed.append(structure.id)
# Pdbs clash
current += 1
return full_structure