def build_phi_psi_model(pdb_filename):
parser = PDBParser()
structure = parser.get_structure("sample", Path(PDBdir, pdb_filename))
model = structure[0]
chain = model["A"]
seq = ""
phi_diangle = []
psi_diangle = []
omega_diangle = []
for res in chain:
if res.get_resname() in resdict.keys():
seq += resdict[res.get_resname()]
if len(seq) == 1:
N_prev = res["N"]
CA_prev = res["CA"]
C_prev = res["C"]
else:
n1 = N_prev.get_vector()
ca1 = CA_prev.get_vector()
c1 = C_prev.get_vector()
C_curr = res["C"]
N_curr = res["N"]
CA_curr = res["CA"]
c = C_curr.get_vector()
n = N_curr.get_vector()
ca = CA_curr.get_vector()
psi = calc_dihedral(n1, ca1, c1, n) ##goes to current res
omega = calc_dihedral(ca1, c1, n, ca)
phi = calc_dihedral(c1, n, ca, c) ##goes to current res
phi_diangle.append(phi * 180.0 / math.pi)
psi_diangle.append(psi * 180.0 / math.pi)
omega_diangle.append(omega * 180.0 / math.pi)
N_prev = res["N"]
CA_prev = res["CA"]
C_prev = res["C"]
model_structure_omega = PeptideBuilder.make_structure(
seq, phi_diangle, psi_diangle, omega_diangle
)
model_structure_phi_psi = PeptideBuilder.make_structure(
seq, phi_diangle, psi_diangle
)
return model_structure_omega, model_structure_phi_psi
def get_structure_from_angles(aa_list, phi_list, psi_list, omega_list):
assert len(aa_list) == len(phi_list) + 1 == len(psi_list) + 1 == len(
omega_list) + 1
structure = PeptideBuilder.make_structure(
aa_list, list(map(lambda x: math.degrees(x), phi_list)),
list(map(lambda x: math.degrees(x), psi_list)),
list(map(lambda x: math.degrees(x), omega_list)))
return structure
def build_phi_psi_model(pdb_filename):
parser=PDBParser()
structure=parser.get_structure('sample', \
path.join(PDBdir, pdb_filename))
model=structure[0]
chain=model['A']
seq=""
phi_diangle=[]
psi_diangle=[]
omega_diangle=[]
for res in chain:
if(res.get_resname() in resdict.keys()):
seq+=resdict[res.get_resname()]
if(len(seq)==1):
N_prev=res['N']
CA_prev=res['CA']
C_prev=res['C']
else:
n1=N_prev.get_vector()
ca1=CA_prev.get_vector()
c1=C_prev.get_vector()
C_curr=res['C']
N_curr=res['N']
CA_curr=res['CA']
c=C_curr.get_vector()
n=N_curr.get_vector()
ca=CA_curr.get_vector()
psi= calc_dihedral(n1, ca1, c1, n) ##goes to current res
omega= calc_dihedral(ca1, c1, n, ca)
phi= calc_dihedral(c1, n, ca, c) ##goes to current res
phi_diangle.append(phi*180.0/math.pi)
psi_diangle.append(psi*180.0/math.pi)
omega_diangle.append(omega*180.0/math.pi)
N_prev=res['N']
CA_prev=res['CA']
C_prev=res['C']
model_structure_omega= PeptideBuilder.make_structure( seq, phi_diangle, psi_diangle, omega_diangle )
model_structure_phi_psi= PeptideBuilder.make_structure( seq, phi_diangle, psi_diangle)
return model_structure_omega, model_structure_phi_psi
def torsion_to_3d(aa, phi, psi):
phi, psi = phi[1:], psi[:-1]
recon_struct = PeptideBuilder.make_structure(aa, phi, psi)
atoms = recon_struct.get_atoms()
coords = list()
for atom in atoms:
coords.append(atom.get_coord())
return np.array(coords)
def get_pdb_from_torsion(aa_seq, phis, psis, save_path):
#phis is of length n, disregard first one
#psis is of length n, disregard the last one
#aa_seq is a string of amino acids encoded by capital letters
phis = phis[1:]
psis = psis[:-1]
my_reconstruct = PeptideBuilder.make_structure(aa_seq, phis, psis)
out = Bio.PDB.PDBIO()
out.set_structure(my_reconstruct)
out.save(save_path)
def test_make_structure():
phi_list = []
psi_im1_list = []
for i in range(1, 20):
phi_list.append(-60 + i)
psi_im1_list.append(-40 - i)
structure = PeptideBuilder.make_structure("ACDEFGHIKLMNPQRSTVWY", phi_list,
psi_im1_list)
assert compare_to_reference(structure, "helix.pdb")
def get_structure_from_angles(aa_list_encoded, angles):
aa_list = protein_id_to_str(aa_list_encoded)
omega_list = angles[1:, 0]
phi_list = angles[1:, 1]
psi_list = angles[:-1, 2]
assert len(aa_list) == len(phi_list) + 1 == len(psi_list) + 1 == len(omega_list) + 1
structure = PeptideBuilder.make_structure(aa_list,
list(map(lambda x: math.degrees(x), phi_list)),
list(map(lambda x: math.degrees(x), psi_list)),
list(map(lambda x: math.degrees(x), omega_list)))
return structure
def find_max(length):
global seq, file, seq_to_pdb, pdb_to_seq
seq = 'W' * length
out = Bio.PDB.PDBIO()
i = pb.make_structure(seq, [180] * len(seq), [180] * len(seq))
out.set_structure(i)
out.save("ignore.pdb")
cleaning.cleanATOM("ignore.pdb", out_file=d + "\\ignore", ext='.pdb')
file = pd.parsePDB("ignore.pdb")
seq, seq_to_pdb, pdb_to_seq = find_seq(file)
final = main_calc(seq)
atom_length = len(final['primary'])
eucl_length = final['secondary'].max()
return atom_length, eucl_length
def test_make_structure():
"""
Build a helix containing all 20 amino acids, with slowly varying
backbone angles, using make_structure().
The resulting structure should be identical to `helix.pdb`
"""
phi_list = []
psi_im1_list = []
for i in range(1, 20):
phi_list.append(-60 + i)
psi_im1_list.append(-40 - i)
structure = PeptideBuilder.make_structure(
"ACDEFGHIKLMNPQRSTVWY", phi_list, psi_im1_list
)
assert compare_to_reference(structure, "helix.pdb")
def test_make_structure2():
phi_list = []
psi_im1_list = []
omega_list = []
for i in range(1, 20):
phi_list.append(-60 + i)
psi_im1_list.append(-40 - i)
omega_list.append(180)
for i in range(9, 19):
omega_list[i] = -178
structure = PeptideBuilder.make_structure("ACDEFGHIKLMNPQRSTVWY", phi_list,
psi_im1_list, omega_list)
assert compare_to_reference(structure, "helix2.pdb")
def main():
pdb_dir = 'data/pdb/'
pdb_prefix = 'pdb'
pdb_suffix = '.ent'
datafile = 'testing.pkl'
data = load_data(datafile)
pdbs, lengths, chains, aas, q8s, dcalphas, coords, phis, psis, pssms = data
# reconstruct from phis, psi
for index in range(4, 5): #len(pdbs)):
pdb = pdbs[index]
chain_index = chains[index]
aa = aas[index]
phi = phis[index][1:] # omit first None
psi = psis[index][:-1] # omit last None
coord = coords[index]
angles = compute_dihedral_list(torch.from_numpy(np.array(coord)))
set_trace()
out = Bio.PDB.PDBIO()
#pdb_path = pdb_dir + pdb_prefix + pdb + pdb_suffix
#structure = Bio.PDB.PDBParser(QUIET=True).get_structure(pdb_path[:-4], pdb_path)
#ppb = PPBuilder()
#model = ppb.build_peptides(structure)
#for chain in model
#for chain in structure.get_chains():
# print(chain.get_id())
#print(structure[0])
out.set_structure(structure[0]['A'])
out.save("chain_a.pdb")
out.set_structure(structure[0]['B'])
out.save("chain_b.pdb")
print('pdb', pdb)
#print('aa', aa)
#print('phi', phi)
#print('psi', psi)
#print('coord',coord)
recon_struct = PeptideBuilder.make_structure(aa, phi, psi)
recon_pdb = Bio.PDB.PDBIO()
recon_pdb.set_structure(recon_struct)
recon_pdb.save('reconstruct.pdb')
def get_structure_from_angles(aa_list_encoded, angles):
aa_list = protein_id_to_str(aa_list_encoded)
omega_list = angles[1:, 0]
phi_list = angles[1:, 1]
psi_list = angles[:-1, 2]
assert len(aa_list) == len(phi_list) + 1 == len(psi_list) + 1 == len(
omega_list) + 1
print("PREDICTION RESULTS")
print("AALIST", aa_list)
print("PHI", list(map(lambda x: math.degrees(x), phi_list)))
print("PHI len", len(list(map(lambda x: math.degrees(x), phi_list))))
time.sleep(100000)
structure = PeptideBuilder.make_structure(
aa_list, list(map(lambda x: math.degrees(x), phi_list)),
list(map(lambda x: math.degrees(x), psi_list)),
list(map(lambda x: math.degrees(x), omega_list)))
return structure
def first_structure_helix(seq_query, seq_template, structure_file,
output_file):
""" This function writes a linear 3D structure of a query protein sequence
in a pdb file. All positions are first set with phi and psi angle from
beta strand structure. Template structural information in used to set
phi and psi angles of positions matching a position in template, with
an alpha helix conformation, to alpha helix psi and phi angle values.
Parameters:
- seq_query : a string representing aligned query sequence.
- seq_template : a string representing aligned template sequence.
- structure_file : a string representing a path to a file
containing structural information extracted from a dssp file of
template.
- output_file : a string representing a path where to write output
pdb.
"""
#Create a temprary file with template secondary structural information
align_dict = mk_align_dict(seq_query, seq_template)
seq = seq_query.replace("-", "")
#angles default values
phi_list = [-120 for i in range(len(seq))]
psi_list = [120 for i in range(len(seq))]
# Building the first structure with alpha helix
# Phi/Psi angles detection from template structure
with open(structure_file, "r") as filin:
print("Looking for alpha helix...")
pos = 1
for line in filin:
# on extrait le premier et le deuxieme champ de chaque ligne (index?)
columns = line[:-1].split(";")
if (pos in align_dict) and (columns[2] == "aH"
or columns[2] == "aI"):
#angles value for alpha helix
phi_list[align_dict[pos] - 1] = -57.8
psi_list[align_dict[pos] - 1] = -47.0
pos = pos + 1
#building the pdb file
struct = PeptideBuilder.make_structure(seq, phi_list, psi_list)
opt = Bio.PDB.PDBIO()
opt.set_structure(struct)
opt.save(output_file)
def write_to_pdb_strcture(atomic_coords, aaSequence, prot_id):
_aa_dict_inverse = {v: k for k, v in AA_ID_DICT.items()}
atomic_coords = list([Vector(v) for v in atomic_coords.numpy()])
aa_list = []
phi_list = []
psi_list = []
omega_list = []
for i, coord in enumerate(atomic_coords):
if int(aaSequence[int(i / 3)]) == 0:
print("Reached end of protein, stopping")
break
if i % 3 == 0:
aa_symbol = _aa_dict_inverse[int(aaSequence[int(i / 3)])]
aa_list.append(aa_symbol)
if i != 0:
phi_list.append(
math.degrees(
Bio.PDB.calc_dihedral(atomic_coords[i - 1],
atomic_coords[i],
atomic_coords[i + 1],
atomic_coords[i + 2])))
if i + 3 < len(atomic_coords) and int(
aaSequence[int(i / 3) + 1]) != 0:
psi_list.append(
math.degrees(
Bio.PDB.calc_dihedral(atomic_coords[i],
atomic_coords[i + 1],
atomic_coords[i + 2],
atomic_coords[i + 3])))
omega_list.append(
math.degrees(
Bio.PDB.calc_dihedral(atomic_coords[i + 1],
atomic_coords[i + 2],
atomic_coords[i + 3],
atomic_coords[i + 4])))
out = Bio.PDB.PDBIO()
structure = PeptideBuilder.make_structure(aa_list, phi_list, psi_list,
omega_list)
out.set_structure(structure)
out.save("output/protein_" + str(prot_id) + ".pdb")
def test_make_structure2():
"""
Build a helix containing all 20 amino acids, with slowly varying
backbone angles, using make_structure(). Now we're changing omega also.
The first half of the resulting structure should be identical to
`helix.pdb`, while the second half should be slightly different.
"""
phi_list = []
psi_im1_list = []
omega_list = []
for i in range(1, 20):
phi_list.append(-60 + i)
psi_im1_list.append(-40 - i)
omega_list.append(180)
for i in range(9, 19):
omega_list[i] = -178
structure = PeptideBuilder.make_structure(
"ACDEFGHIKLMNPQRSTVWY", phi_list, psi_im1_list, omega_list
)
assert compare_to_reference(structure, "helix2.pdb")
def constructStr(seq, Phi, Psi, NumofStructGen):
""""Construct structures based on given Phi and Psi angle ensembles."""
SeqLen = len(Phi[:, 0])
iCount = 0
PhiOut = np.zeros((int(SeqLen), int(NumofStructGen)))
PsiOut = np.zeros((int(SeqLen), int(NumofStructGen)))
for j in range(10 * NumofStructGen):
model_structure_phi_psi = PeptideBuilder.make_structure(
seq, Phi[:, j], Psi[:, j])
Clashes = disCheck(model_structure_phi_psi)
if Clashes == False:
PhiOut[:, iCount] = Phi[:, j]
PsiOut[:, iCount] = Psi[:, j]
iCount = iCount + 1
if iCount % 100 == 0:
print(iCount)
out = PDBIO()
out.set_structure(model_structure_phi_psi)
name = 'output' + str(iCount) + '.pdb'
out.save('./PDBOUT/' + name)
if iCount > (NumofStructGen - 1):
break
return PhiOut, PsiOut
def to_pdb(aa, phi, psi, path):
phi, psi = phi[1:], psi[:-1]
struct = PeptideBuilder.make_structure(aa, phi, psi)
out = Bio.PDB.PDBIO()
out.set_structure(struct)
out.save(path)
geos.append(Geometry.geometry(aa))
structure = PeptideBuilder.make_structure_from_geos(geos)
out.set_structure(structure)
out.save("test5.pdb")
# Build a helix containing all 20 amino acids, with slowly varying
# backbone angles, using make_structure().
# The resulting structure should be identical to test2.pdb
phi_list = []
psi_im1_list = []
for i in range(1,20):
phi_list.append(-60+i)
psi_im1_list.append(-40-i)
structure = PeptideBuilder.make_structure("ACDEFGHIKLMNPQRSTVWY",\
phi_list, psi_im1_list)
out.set_structure(structure)
out.save("test6.pdb")
# Build a helix containing all 20 amino acids, with slowly varying
# backbone angles, using make_structure().
# The first half of the resulting structure should be identical to
# test6.pdb, while the second half should be slightly different.
phi_list = []
psi_im1_list = []
omega_list = []
for i in range(1,20):
phi_list.append(-60+i)
psi_im1_list.append(-40-i)
omega_list.append(180)
