def get_surface_residues(filename, my_acc_array, my_threshold):
"""
Given a pdb file, finds the residues exposed to the solvent (not buried)
according to the ASA (accessible surface area) value given by DSSP module.
The user can select a threshold of ASA. Default is 0.2.
"""
p = PDBParser(PERMISSIVE=1)
s = p.get_structure("code.pdb", filename)
model = s[0]
d = DSSP(model, filename, dssp='mkdssp', acc_array=my_acc_array)
sys.stderr.write("\nHandled %i residues\n" % len(d))
residue_number = set()
for element in sorted(d):
if type(element[3]) is not str: #Sometimes the element[3] is NA
if element[3] >= my_threshold:
# foreach aa in the surface (according to threshold) store residue_number
try:
residue_number.add(
str(list(d.keys())[element[0] - 1][1][1]) +
list(d.keys())[element[0] - 1][0])
except IndexError:
sys.stderr.write("Element " + str(d.keys()[0]) +
" index out of range\n")
return residue_number
def _run(self, pdbcode, chains):
pdb_file = os.path.join(PDB_PATH, 'pdb' + pdbcode + '.ent')
p = PDBParser()
structure = p.get_structure(pdbcode, pdb_file)
model = structure[0]
dssp = DSSP(model, pdb_file, dssp=dssp_route)
valid_keys = [key for key in dssp.keys() if key[0] in chains]
chain = dssp.keys()[0][0]
return [dssp[key] for key in valid_keys], chain
def add_dssp(pdb_file):
'This is a small function that makes a dictionary with secondary structure information'
from Bio.Seq import Seq
from Bio import SeqIO
from Bio.Alphabet import IUPAC
io = PDBIO()
import pandas as pd
from Bio.PDB.DSSP import DSSP
pdbl = PDBList()
parser = PDBParser()
ppb = PPBuilder()
'If the structure hasnt been downloaed then it will - else parse it'
structure = parser.get_structure(pdb_file,pdb_file)
model = structure[0]
chain = model['A']
'path to the dssp excecutable:'
dssp_exc = '/Users/thorn/dssp-2.2.1/mkdssp'
dssp = DSSP(model, pdb_file, dssp_exc)
sec_dict = {}
for i in range(len(dssp)):
a_key = dssp.keys()[i]
index = a_key[1][1]
sec_structure = dssp[a_key][1]
sec_dict[index] = sec_structure
return sec_dict
def calc_dssp(model, chain_sites: dict, pdb_name: str) -> None:
# DSSP
# ============ ===
# Tuple Index Value
# ============ ===
# 0 DSSP index
# 1 Amino acid
# 2 Secondary structure
# 3 Relative ASA
# 4 Phi
# 5 Psi
# 6 NH-->O_1_relidx
# 7 NH-->O_1_energy
# 8 O-->NH_1_relidx
# 9 O-->NH_1_energy
# 10 NH-->O_2_relidx
# 11 NH-->O_2_energy
# 12 O-->NH_2_relidx
# 13 O-->NH_2_energy
# ============ ===
try:
dssp = DSSP(model, pdb_name, dssp="mkdssp")
except:
dssp = {}
print("dssp failed!")
for residue in dssp.keys():
(
dssp_i,
aa,
sec_struct,
sasa_r,
phi,
psi,
nh_o1_relidx,
nh_o1_e,
o_nh1_relidx,
o_nh1_e,
nh_o2_relidx,
nh_o2_e,
o_nh2_relidx,
o_nh2_e,
) = dssp[residue]
chain_id, res_id = residue
_, resnumb, _ = res_id
if chain_id in chain_sites and resnumb in chain_sites[chain_id]:
resid = chain_sites[chain_id][resnumb][1]
new_res = chain_sites[chain_id][resnumb][2]
new_res.sec_struct = sec_struct
new_res.sasa_r = sasa_r
new_res.phi = phi
new_res.psi = psi
session.commit()
def calc_features(PATH, PDB_id, OUTPATH):
#Loading the files
parser = PDBParser(PERMISSIVE=1)
filename = os.path.join(PATH, PDB_id + ".pdb")
structure = parser.get_structure(PDB_id, filename)
model = structure[0]
#DSSP Analysis for SS, PHI, PSI
dssp = DSSP(model, filename)
#NACCESS Analysis for SASA
rsa, asa = run_naccess(model, filename)
rsa = process_rsa_data(rsa)
#Feature mapping to each atomic coordinate
dssp_present, dssp_not_present = 0, 0
feature = dict() #The feature dictionary
for model in structure:
for chain in model:
for residue in chain:
for atom in residue:
print(atom.get_full_id())
ID = (atom.get_full_id()[2], atom.get_full_id()[3])
if (ID in list(dssp.keys())):
if (rsa[ID]["all_atoms_abs"] > Threshold):
rsa_label = 1
else:
rsa_label = 0
feat = (SS_Labels[dssp[ID][2]], dssp[ID][4] / 360,
dssp[ID][5] / 360, rsa_label)
feature[tuple(atom.get_coord())] = feat
print(ID, atom.get_coord(), feat)
dssp_present += 1
else:
print("==> ID not present : ", atom.get_full_id())
dssp_not_present += 1
#Printing the Stats
print(
"==> STATS : PDBID : %s , DSSP PRESENT : %s , DSSP NOT PRESENT : %s" %
(PDB_id, dssp_present, dssp_not_present))
#Saving the feature to each PDB file
with open(os.path.join(OUTPATH, PDB_id + ".dat"), "wb+") as f:
pickle.dump(feature, f)
print("==> Dump completed")
def getSecondaryStructure(pdbID):
pdb.fetchInPDBFormat(pdbID)
p = PDBParser()
structure = p.get_structure(pdbID, pdb.getTemporaryPDBPath())
model = structure[0]
dssp = DSSP(model, pdb.getTemporaryPDBPath())
seq = ""
secStructure = ""
for key in dssp.keys():
seq = seq + dssp[key][1]
secStructure = secStructure + dssp[key][2]
# print("Secuencia: ", seq + " de longitud " + str(len(seq)))
# print("Estructura secundaria: ", secStructure + " de longitud " + str(len(secStructure)))
return secStructure
def accessible_surface_area(PDB_file):
# Calcul de la surface accessible au solvant pour chaque residus
ASA_dict = {}
parser = PDBParser()
structure_id = PDB_file.split(".")[0]
structure = parser.get_structure(structure_id, PDB_file)
model = structure[0]
dssp = DSSP(model, PDB_file, dssp='mkdssp')
id_CA = 0
for CA in list(dssp.keys()):
if dssp[CA][1] != 'X':
ASA_dict[id_CA] = dssp[CA][3]
id_CA += 1
return ASA_dict
def calculate_asa(model, filename, AROM_LIST, chain_list):
"""Returns a list of surface exposed residues as determined by relative solvent accessibility.
Only standard protein residues are currently supported. Non-protein and user specified custom residues cannot be
classified as surface exposed using this criteria.
Parameters
---------
model: :class:`Bio.PDB.Model.Model`
Model which contains chains and residues of protein strucutre
filename: str
Name of pdb file to be analyzed
AROM_LIST : list of str
List containing which standard residues are included in analysis
chain_list: list of str
Chains are included in analysis
Notes
-----
The relative accessible surface area (RSA) of each residue is calculated using the Bio.PDB.DSSP module. A residue
with an RSA value of 0.05 or higher is classified as surface exposed.
References
---------
Tien, M. Z.; Meyer, A. G.; Sydykova, D. K.; Spielman, S. J.; Wilke, C. O. PLoS ONE 2013, 8 (11).
Reference for relative solvent accessibility cutoff of 0.05, and for MaxASA values
"""
cutoff = .05
surface_exposed_res = []
letter_list = []
for res_name in AROM_LIST:
if res_name_to_char.get(res_name):
letter_list.append(res_name_to_char.get(res_name))
try:
dssp = DSSP(model, filename, acc_array="Wilke")
keys = list(dssp.keys())
for key in keys:
if key[0] in chain_list and dssp[key][3] >= cutoff and dssp[key][
1] in letter_list:
goal_str = dssp[key][1] + \
str(key[1][1]) + "(" + str(key[0]) + ")"
surface_exposed_res.append(goal_str)
except Exception as e:
warnings.warn(
"Unable to calculate solvent accessibility. Check that DSSP is installed.",
RuntimeWarning,
stacklevel=2)
return surface_exposed_res
def __init__(self, pdbfile, fastafile):
names = {'HIS':'H','ASP':'D','ARG':'R','PHE':'F','ALA':'A','CYS':'C','GLY':'G',\
'GLN':'Q','GLU':'E','LYS':'K','LEU':'L','MET':'M','ASN':'N','SER':'S',\
'TYR':'Y','THR':'T','ILE':'I','TRP':'W','PRO':'P','VAL':'V','SER':'S'}
# Load fasta residue sequence
f = open(fastafile)
ff = [line.rstrip("\n") for line in f]
f.close()
p_id = ff[0][1:]
self.seq = ff[1]
# Load pdb information
p = PDBParser(PERMISSIVE=1)
st = p.get_structure(p_id, pdbfile)
model = st[0]
tag = p_id[-1]
chain = model[tag]
residues = chain.get_residues()
self.residues = [res for res in residues]
## sequence info from pdb
self.pdbseq = "".join([names[res.get_resname()] for res in self.residues if \
names.has_key(res.get_resname())])
## 3-state sse info from pdb
dssp = DSSP(model, pdbfile)
to3_dict = {'-':'C', 'G':'H', 'H':'H', 'I':'H', 'E':'E', 'B':'E', 'T':'C', \
'S':'C', 'L':'C'}
keys = list(dssp.keys())
self.pdbss3seq = "".join([to3_dict[dssp[k][2]]for k in keys])
# Align the pdb sequence(always missing some residues) to fasta sequence
alignment = AlignNW(self.seq, self.pdbseq)
self.re_index = alignment['j']
self.re_index = [i-1 for i in alignment['j']] # minus 1 for indexing
# generate sequence alignment between pdb sequence and fasta sequence
self.alignment = "".join([self.pdbseq[i] if i > -1 else "-" for i in self.re_index])
self.alignment = "\n".join([self.seq, self.alignment])
# generate full lenght of 3-state SSE sequence according to re-index
self.ss3seq = "".join([self.pdbss3seq[i] if i > -1 else "C" for i in self.re_index])
# generate full lenght of distance matrix(distance=-1 when disappear in pdbseq)
self.dist_matrix = self.generate_dist_matrix()
#np.savetxt("test.txt", self.dist_matrix)
# generate full lenght of angle matrix(distance=None when disappear in pdbseq)
self.angle_matrix = self.generate_angle_matrix()
def getDSSP(
struct,
fname,
dsspPath=os.path.expanduser(
"~/Tesis/rriPredMethod/dependencies/bioinformaticTools/dssp/mkdssp")):
dssp = DSSP(struct[0], fname, dssp=dsspPath)
chains = struct[0].child_list
dsspDict = {
chain.get_id(): {symbol: []
for symbol in DSSP_SYMBOLS}
for chain in chains
}
for chainId, resId in dssp.keys():
secStruct = dssp[(chainId, resId)][2]
dsspDict[chainId][secStruct].append(resId)
return dsspDict
def __applyDssp(self):
import Bio.PDB as bio
print('PSU: applying dssp')
from Bio.PDB.DSSP import DSSP
p = bio.PDBParser()
pdbFile = self.pdbDataPath + 'pdb' + self.pdbCode + '.ent'
structure = p.get_structure(self.pdbCode, pdbFile)
model = structure[0]
dssp = DSSP(model, pdbFile)
for akey in list(dssp.keys()):
chain = akey[0]
res_no = akey[1][1]
row = dssp[akey]
ss = row[2]
for atom in self.atoms:
if atom.values['rid'] == res_no and atom.values[
'chain'] == chain:
atom.setDsspInfo(ss)
print('PSU: applied dssp successfully')
def Make_dssp():
ref = {
'A': 'ALA',
'R': 'ARG',
'N': 'ASN',
'D': 'ASP',
'B': 'ASX',
'C': 'CYS',
'E': 'GLU',
'Q': 'GLN',
'Z': 'GLX',
'G': 'GLY',
'H': 'HIS',
'I': 'ILE',
'L': 'LEU',
'K': 'LYS',
'M': 'MET',
'F': 'PHE',
'P': 'PRO',
'S': 'SER',
'T': 'THR',
'W': 'TRP',
'Y': 'TYR',
'V': 'VAL',
'X': '---'
}
dssp_dict = {}
dssp_dict['X'] = np.NaN
p = PDBParser()
structure = p.get_structure('model', './current_pdb.txt')
mod = structure[0]
dssp = DSSP(mod, './current_pdb.txt')
for i in range(len(dssp)):
a_key = list(dssp.keys())[i]
dssp_dict[str(a_key[0]) + str(a_key[1][1]) +
str(ref[dssp[a_key][1]])] = dssp[a_key][2]
return (dssp_dict)
print()
os.remove(filename)
continue
valid_aa = [
'A', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'K', 'L', 'M', 'N', 'P',
'Q', 'R', 'S', 'T', 'V', 'W', 'Y'
]
current_chain = ""
chain = ""
phi_psis = []
dihedrals = []
chains = []
chain_names = []
# we assume that the chains and residues are in order, i.e. A1,A2,A3,...,B1,B2,B3,...
for key in dssp.keys():
chain_id = key[0]
if not current_chain or chain_id != current_chain:
current_chain = chain_id
chain_names.append(chain_id)
if chain:
assert len(chain) == len(phi_psis), \
"the length of chain '%s' does not equal the number of dihedrals: %s" % (chain, len(phi_psis))
chains.append(chain)
dihedrals.append(phi_psis)
chain = ""
phi_psis = []
residue = dssp[key]
from Bio.PDB import PDBParser
from Bio.PDB.DSSP import DSSP
#Read in and Parse PDB file to obtain DSSP --> secondary structure determination
#follows the basic outline on biopython.org -- tutorial
parse = PDBParser()
struc = parse.get_structure('6hrc', "6hrc.pdb")
model = struc[0]
dssp = DSSP(model, '6hrc.pdb')
sec_struc = ''
a_helix = 0
b_sheet = 0
other = 0
none = 0
key = list(dssp.keys())[2]
dssp[key]
for c in range(len(dssp)):
key = list(dssp.keys())[c]
sec_struc += dssp[key][2]
if dssp[key][2] == "H" or dssp[key][2] == "G" or dssp[key][2] == "I":
a_helix += 1
if dssp[key][2] == "E" or dssp[key][2] == "B":
b_sheet += 1
if dssp[key][2] == "-":
none += 1
else:
other += 1
def calc_features(PATH, pdb_ligand_ID, OUTPATH):
#Loading the files
parser = PDBParser(PERMISSIVE=1)
PDB_id = pdb_ligand_ID[:4].lower() #+ '_pocket'
filename = os.path.join(PATH, PDB_id + ".pdb")
structure = parser.get_structure(PDB_id, filename)
model = structure[0]
#DSSP Analysis for SS, PHI, PSI
dssp = DSSP(model, filename)
#NACCESS Analysis for SASA
rsa, asa = run_naccess(model, filename)
rsa = process_rsa_data(rsa)
# print(rsa)
#Feature mapping to each atomic coordinate
dssp_present, dssp_not_present = 0, 0
feature = dict() #The feature dictionary
for model in structure:
for chain in model:
if (chain.get_full_id()[2] == pdb_ligand_ID.split('_')[2]):
pssm_ID = chain.get_full_id()[0][:4].upper(
) + '_' + chain.get_full_id()[2]
pssm = parse_PSSM(pssm_ID)
start = True
gap = 0
idx_prev = 0
for residue in chain:
# if(start):
# start_idx =residue.get_full_id()[3][1]
# idx_prev = 0
idx = residue.get_full_id()[3][1]
if (idx < 1):
print(idx)
a = 0
pass
elif (idx - idx_prev >= 1):
print(idx)
a = 1
gap += idx - idx_prev - 1
# elif(start):
# gap += -1
# start = False
for atom in residue:
# print(atom.get_full_id())
ID = (atom.get_full_id()[2], atom.get_full_id()[3])
if (ID in list(dssp.keys())):
if (rsa[ID]["all_atoms_abs"] > RSA_Threshold):
rsa_label = 1
else:
rsa_label = 0
print(gap, atom.get_full_id()[3][1], a)
feat = (SS_Labels[dssp[ID][2]],
dssp[ID][4] / 360, dssp[ID][5] / 360,
rsa_label) + tuple(
pssm[str(atom.get_full_id()[3][1] -
gap)])
feature[tuple(atom.get_coord())] = feat
print(pdb_ligand_ID[:4], ID, atom.get_coord(),
feat)
dssp_present += 1
else:
print(">>> ID not present : ",
atom.get_full_id())
dssp_not_present += 1
idx_prev = idx
#Printing the Stats
print(
"===> STATS : PDBID : %s , DSSP PRESENT : %s , DSSP NOT PRESENT : %s" %
(PDB_id, dssp_present, dssp_not_present))
#Saving the feature to each PDB file
with open(os.path.join(OUTPATH, pdb_ligand_ID + ".dat"), "wb+") as f:
pickle.dump(feature, f)
print("====> Dump completed")
args = parser.parse_args()
# args = parser.parse_args(['-p', '-c', 'A', '--pymol', 'ctc445/CTC-445.pdb'])
# args = parser.parse_args(['-p', '-c', 'A', '--pymol', 'ctc445/CTC-640.pdb'])
for pdb in args.pdbs:
init_time = time()
input_basename, _ = path.splitext(path.basename(pdb))
# parse input pdb and prepare for prediction
input_structure = PDB.PDBParser().get_structure('input_structure', pdb)
try:
dssp = DSSP(input_structure[0], pdb)
except Exception as e:
raise RuntimeError('DSSP failed', e)
ss = [
dssp[k][2] for k in filter(
lambda k: k[0] == args.chain if args.chain else True, dssp.keys())
]
if args.chain:
input_structure = next(
filter(lambda c: c.id == args.chain, input_structure.get_chains()))
# ca_atoms = list(filter(lambda a: a.name == 'CA', input_structure.get_atoms()))
n_atoms = list(filter(lambda a: a.name == 'N',
input_structure.get_atoms()))
c_atoms = list(filter(lambda a: a.name == 'C',
input_structure.get_atoms()))
ca_atoms = list(
filter(lambda a: a.name == 'CA', input_structure.get_atoms()))
cb_xyz = []
from Bio.PDB import *
from Bio.PDB.DSSP import DSSP
print("START")
p = PDBParser()
structure = p.get_structure("X", "3g8n.pdb")
model = structure[0]
#chain = model['A']
#print("RESIDES #:", len(list(chain)))
dssp = DSSP(model, "3g8n.pdb", dssp='mkdssp')
print(dssp.keys()[2][1][1]) #RESIDUE NUMBER
result = ""
for i in range(len(list(dssp.keys()))):
if dssp.keys()[i][0] != 'A':
continue
a_key = list(dssp.keys())[i]
result += dssp[a_key][2]
print("STRING", result)
print(len(result))
#a_key = list(dssp.keys())[4]
#print("RESULT", dssp[a_key])
#print("LEN", len(list(dssp.keys())))
def computeOneFile(self, pdbFname, struct):
'''
Computes DSSP for a given pdb file
:param pdbFname: str. fname to pdb file
:param struct: Bio.PDB.Structure
'''
allResidues = set([])
for chain in struct[0]:
residues = chain.get_residues()
allResidues = allResidues.union(set(residues))
prefixExtended = self.getExtendedPrefix(pdbFname)
prefix, chainType = self.splitExtendedPrefix(prefixExtended)[:2]
if self.checkAlreayComputed(prefixExtended):
print("Dssp already computed for %s" % prefixExtended)
return 0
print("launching Dssp over %s" % prefixExtended)
try:
featuresDict = {}
try:
dssp_out = DSSP(struct[0], pdbFname, dssp=self.dsspBinPath)
except Exception as e:
if "DSSP failed to produce an output" in e.message:
dssp_out = {}
else:
print(e)
raise e
for chainId, resId in dssp_out.keys():
secStruct = dssp_out[(chainId, resId)][2]
if secStruct == "-": secStruct = "Z"
try:
featuresDict[struct[0][chainId][resId]] = secStruct
except KeyError:
continue
dataDict = {}
for aa in allResidues:
chainId_resIdStr_resName = self.fromRes2ChainResIdAndName(aa)
chainId, resIdStr, resName = chainId_resIdStr_resName
if resName is None: continue
if aa in featuresDict:
values = [featuresDict[aa]]
else:
values = ["Z"]
record = [chainId, resIdStr, resName] + values
record = " ".join(record)
try:
dataDict[chainId].append(record)
except KeyError:
dataDict[chainId] = [record]
categoricalLevels = {
("H", "B", "E", "G", "I", "T", "S", "Z"): ("2ndStruct", )
}
self.writeResultsFromDataDictSingleChain(
dataDict,
outName=self.getFNames(prefixExtended)[0],
categoricalLevels=categoricalLevels)
except (Exception, KeyboardInterrupt):
self.tryToRemoveAllFnames(prefixExtended)
raise
def get_struc_feat(seq_len, model_file):
'''
Agrs:
seq_len (int): sequence length.
model_file (string): the path of model file.
'''
feature = {}
structure = pdb_parser.get_structure("tmp_stru", model_file)
model = structure.get_list()[0]
residues = model.get_list()[0].get_list()
# SS and RSA
dssp = DSSP(model, model_file, dssp='dssp')
SS3s, RSAs = np.zeros((seq_len, 3)), np.zeros((seq_len, 1))
for _key in dssp.keys():
res_index = _key[1][1]
if res_index >= 1 and res_index <= seq_len:
SS3s[res_index - 1, SS3_TYPES[dssp[_key][2]]] = 1
RSAs[res_index - 1] = [dssp[_key][3]]
feature['SS3'] = SS3s
feature['RSA'] = RSAs
atom_types = ['CA', 'CB', 'N', 'O']
# generate empty coordinates
coordinates = []
for _ in range(seq_len):
_dict = {}
for atom_type in atom_types:
_dict[atom_type] = None
coordinates.append(_dict)
# extract coordinates from pdb
for res in residues:
for atom in res:
if atom.name in atom_types:
coordinates[res.id[1] - 1][atom.name] = atom.coord
# copy CA coordinate to CB if CB is None (GLY)
if 'CB' in atom_types and coordinates[res.id[1] - 1]['CB'] is None:
coordinates[res.id[1] - 1]['CB'] = coordinates[res.id[1] - 1]['CA']
# distance map
atom_pairs = ['CaCa', 'CbCb', 'NO']
for atom_pair in atom_pairs:
atom1, atom2 = atom_pair[:int(len(atom_pair) / 2)].upper(
), atom_pair[int(len(atom_pair) / 2):].upper()
X = [
list(c[atom1]) if
(c is not None and c[atom1] is not None) else [0, 0, 0]
for c in coordinates
]
X_valid = [
0 if (c is None or c[atom1] is None) else 1 for c in coordinates
]
Y = [
list(c[atom2]) if
(c is not None and c[atom2] is not None) else [0, 0, 0]
for c in coordinates
]
Y_valid = [
0 if (c is None or c[atom2] is None) else 1 for c in coordinates
]
dist = scipy.spatial.distance_matrix(X, Y).astype(np.float16)
XY_valid = np.outer(X_valid, Y_valid)
np.putmask(dist, XY_valid == 0, -1)
if atom1 == atom2:
np.fill_diagonal(dist, 0) # set the self distance to 0
feature[atom_pair] = dist.reshape(
(dist.shape[0], dist.shape[1], -1)) * 0.1
return feature
vectors = []
phi = []
psi = []
exp_phi = np.array([60, -80])
exp_psi = np.array([-120, 0])
b_turns = []
for atom in structure[0]['A'].get_atoms():
vectors.append(atom.get_vector())
for i in range(len(vectors) - 2):
if i % 2 == 0:
phi.append(calc_angle(vectors[i], vectors[i + 1], vectors[i + 2]))
else:
psi.append(calc_angle(vectors[i], vectors[i + 1], vectors[i + 2]))
df = pd.DataFrame(list(zip(phi, psi)), columns=['phi', 'psi'])
for df_slice in df.rolling(window=2):
if np.allclose(df_slice['phi'].values, exp_phi,
atol=error) and np.allclose(
df_slice['psi'].values, exp_psi, atol=error):
b_turns.append(df_slice)
print('hello')
dssp = DSSP(structure[0], '1g60.pdb', dssp='mkdssp')
phi = list(dssp.keys())[4]
print(dssp[phi])
#Printing atomic types
#########################################
f_write.write("\n[")
#print("[")
for i in range(len(all_atom_types)):
#print("[")
f_write.write("[")
for j in range(len(all_atom_types[i])):
f_write.write(list_atoms_types[all_atom_types[i][j]] + " ")
#print(list_atoms_types[all_atom_types[i][j]])
#print("]")
f_write.write("]")
f_write.write("]")
#print("]")
#########################################
#Printing secondary structure
#########################################
model = structure[0]
dssp = DSSP(model, file_data_name, dssp='mkdssp')
secondary_struct = ""
for i in range(len(list(dssp.keys()))):
if dssp.keys()[i][0] != num_chain:
continue
a_key = list(dssp.keys())[i]
secondary_struct += dssp[a_key][2]
f_write.write("\n" + secondary_struct)
num_of_structures += 1
f_write.write("\n\n")
os.remove(new_name)
f_write.write("]")
f_write.write("]")
print("]")
f_write.write("\n[")
print("[")
for i in range(len(all_atom_types)):
print("[")
f_write.write("[")
for j in range(len(all_atom_types[i])):
f_write.write(list_atoms_types[all_atom_types[i][j]] + " ")
print(list_atoms_types[all_atom_types[i][j]])
print("]")
f_write.write("]")
f_write.write("]")
print("]\n")
#break
# dssp:
model = structure[0]
dssp = DSSP(model, file_data_name, dssp='mkdssp')
num_res_in_chain = 5 # TODO: change!
secondary_struct = ""
for i in range(num_res_in_chain):
a_key = list(dssp.keys())[i]
result = dssp[a_key][2] # returns DSSP secondary structure
secondary_struct += result
f_write.write("secondary_struct" + secondary_struct)
print("secondary_struct" + secondary_struct)
# delete current file
os.remove(file_name)
num_of_structures += 1
break
def main(pdb_file, dssp_exe, out_file):
# Read PDB structrue
p = PDBParser()
structure = p.get_structure('id', pdb_file)
model = structure[0]
# Run DSSP
dssp = DSSP(model, pdb_file, dssp=dssp_exe)
# keys:
# (dssp index, amino acid, secondary structure, relative ASA, phi, psi,
# NH_O_1_relidx, NH_O_1_energy, O_NH_1_relidx, O_NH_1_energy,
# NH_O_2_relidx, NH_O_2_energy, O_NH_2_relidx, O_NH_2_energy)
sec_struc = []
rel_asa = []
for a_key in list(dssp.keys()):
aux = dssp[a_key]
sec_struc.append(aux[2])
rel_asa.append(0 if aux[3] == 'NA' else aux[3])
rel_asa = np.expand_dims(rel_asa, axis=1)
# Get coordinates for N, CA and C atoms
coor_N = []
coor_CA = []
coor_C = []
for chain in model.get_list():
total_len = len(chain.get_list())
for residue in chain.get_list():
coor_N.append(residue['N'].get_coord())
coor_CA.append(residue['CA'].get_coord())
coor_C.append(residue['C'].get_coord())
assert len(coor_N) == len(coor_CA) == len(coor_C) == total_len
# Get dihedral angles and inter-residue angles
angles = []
for j in range(total_len):
# Initialize angles if not found
phi_angle, psi_angle, omega_angle = (2 * np.pi, 2 * np.pi, 2 * np.pi)
theta_angle, tau_angle = (2 * np.pi, 2 * np.pi)
vec_N2, vec_CA2, vec_C2 = (coor_N[j], coor_CA[j], coor_C[j])
# Phi
if j != 0:
vec_C1 = coor_C[j - 1]
phi_angle = calculate_dihedral(vec_C1, vec_N2, vec_CA2, vec_C2)
# Psi and Omega
if j != total_len - 1:
vec_N3 = coor_N[j + 1]
psi_angle = calculate_dihedral(vec_N2, vec_CA2, vec_C2, vec_N3)
vec_CA3 = coor_CA[j + 1]
omega_angle = calculate_dihedral(vec_CA2, vec_C2, vec_N3, vec_CA3)
# Theta
if np.logical_and(j != 0, j != total_len - 1):
vec_CA1 = coor_CA[j - 1]
vec_CA3 = coor_CA[j + 1]
theta_angle = calculate_angle(vec_CA1, vec_CA2, vec_CA3)
# Tau
if np.logical_and(j > 1, j != total_len - 1):
vec_CA0, vec_CA1, vec_CA3 = (coor_CA[j - 2], coor_CA[j - 1],
coor_CA[j + 1])
tau_angle = calculate_dihedral(vec_CA0, vec_CA1, vec_CA2, vec_CA3)
# Concatenate angles
angles.append([phi_angle, psi_angle, theta_angle, tau_angle])
# Calculate sine and cosine of each angle
angles = np.array(angles)
angles_sin = np.sin(angles)
angles_cos = np.cos(angles)
angles_sin_cos = np.stack([
angles_sin[:, 0], angles_cos[:, 0], angles_sin[:, 1], angles_cos[:, 1],
angles_sin[:, 2], angles_cos[:, 2], angles_sin[:, 3], angles_cos[:, 3]
]).T
angles_sin_cos[np.where(np.abs(angles_sin_cos) < 1e-10)] = 0
# One-hot encoding of secondary structure (8-state)
alphabet = 'HBEGITS-' # H Alpha helix (4-12)
# B Isolated beta-bridge residue
# E Strand
# G 3-10 helix
# I Pi helix
# T Turn
# S Bend
# - None
ohdict = dict((c, i) for i, c in enumerate(alphabet))
ss_onehot = np.zeros((total_len, len(ohdict)))
for i in range(total_len):
ss_onehot[i, ohdict[sec_struc[i]]] = 1
# Create feature matrix for the protein
features = np.hstack([ss_onehot, rel_asa, angles_sin_cos])
# Save dictionary of features
with open(out_file, 'wb') as f:
pickle.dump(features.astype('float32'), f, protocol=2)
