def process_water_structures(initial_pdb, main_chains, ligand):
"""
Detects the waters we have to keep (important for the simulation) and returns
a structure holding them.
Important waters are the ones closer to Template residue 50 (Ile), the aa is not
but it is not guaranteed to be conserved, which means we have to rely into the
residue number to choose it, and take any offset into account if needed.
Extra: water molecules must be also close to the binding site. We will pick then the
water that has minimum distance to the binding site and residue 50
:param initial_pdb: The pdb (prody structure) we want to extract the chains.
:return: A dictionary indexed by the water id (res. num. + chain id) holding the prody pdb
structure of that water.
"""
hw = prody.HierView(initial_pdb.select("protein"))
water_structs = {}
for chain in hw.iterChains():
if chain.getChid() in main_chains:
# We cannot do a direct selection, instead we iterate
for i, residue in enumerate(chain.iterResidues()):
if i == 50: # 50th residue
break
residue_com = prody.calcCenter(residue)
if ligand is None:
ligand_com = prody.calcCenter(initial_pdb)
else:
ligand_com = prody.calcCenter(ligand)
# Identify closer water
waters = initial_pdb.select("name O and water")
if waters is not None:
distance_to_R50 = numpy.sqrt(
((residue_com - waters.getCoords())**2).sum(axis=1))
distance_to_BindSite = numpy.sqrt(
((ligand_com - waters.getCoords())**2).sum(axis=1))
distances = distance_to_R50 + distance_to_BindSite
min_dist = numpy.min(distances)
min_dist_index = numpy.where(distances == min_dist)
water_resnum = waters.getResnums()[min_dist_index]
water_chid = waters.getChids()[min_dist_index][0]
water_id = "%d:%s" % (water_resnum, water_chid)
# We use a dict in order to get rid of repeats
selection_string = "resnum %d and chain %s" % (water_resnum,
water_chid)
water_structs[water_id] = initial_pdb.water.select(
selection_string).copy()
return water_structs
def process_water_structures(initial_pdb, main_chains, ligand):
"""
Detects the waters we have to keep (important for the simulation) and returns
a structure holding them.
Important waters are the ones closer to Template residue 50 (Ile), the aa is not
but it is not guaranteed to be conserved, which means we have to rely into the
residue number to choose it, and take any offset into account if needed.
Extra: water molecules must be also close to the binding site. We will pick then the
water that has minimum distance to the binding site and residue 50
:param initial_pdb: The pdb (prody structure) we want to extract the chains.
:return: A dictionary indexed by the water id (res. num. + chain id) holding the prody pdb
structure of that water.
"""
hw = prody.HierView(initial_pdb.select("protein"))
water_structs = {}
for chain in hw.iterChains():
if chain.getChid() in main_chains:
# We cannot do a direct selection, instead we iterate
for i, residue in enumerate(chain.iterResidues()):
if i == 50: # 50th residue
break
residue_com = prody.calcCenter(residue)
if ligand is None:
ligand_com = prody.calcCenter(initial_pdb)
else:
ligand_com =prody.calcCenter(ligand)
# Identify closer water
waters = initial_pdb.select("name O and water")
if waters is not None:
distance_to_R50 = numpy.sqrt(((residue_com - waters.getCoords())**2).sum(axis=1))
distance_to_BindSite = numpy.sqrt(((ligand_com - waters.getCoords())**2).sum(axis=1))
distances = distance_to_R50 + distance_to_BindSite
min_dist = numpy.min(distances)
min_dist_index = numpy.where(distances == min_dist)
water_resnum = waters.getResnums()[min_dist_index]
water_chid = waters.getChids()[min_dist_index][0]
water_id = "%d:%s"%(water_resnum, water_chid)
# We use a dict in order to get rid of repeats
selection_string = "resnum %d and chain %s"%(water_resnum,
water_chid)
water_structs[water_id] = initial_pdb.water.select(selection_string).copy()
return water_structs
def _prepare_points(self):
"""Load structures and compute the location of the points of the
3D image to be generated.
"""
self.complex = parsePDB(self.path)
protein = self.complex.select("not (resname WER or water)")
ligand = self.complex.select("resname WER")
center = calcCenter(ligand.getCoords())
moveAtoms(self.complex, by=-center)
center = calcCenter(self.complex.select("resname WER").getCoords())
self.protein.structure = protein
self.ligand.structure = ligand
self.points = grid_around(center,
self.size,
spacing=24 / (self.size - 1))
def make_query_coords(self):
q1_coords = [
self.query.select('name ' + n).getCoords()[0]
for n in self.query_lig_corr[0]
]
if self.query_cyclic[0]:
len_coords = len(q1_coords)
q_sel1_coords = [[q1_coords[i - j] for i in range(len_coords)]
for j in range(len_coords)]
else:
q_sel1_coords = [q1_coords]
q2_coords = [
self.query.select('name ' + n).getCoords()[0]
for n in self.query_lig_corr[1]
]
if self.query_cyclic[1]:
len_coords = len(q2_coords)
q_sel2_coords = [[q2_coords[i - j] for i in range(len_coords)]
for j in range(len_coords)]
else:
q_sel2_coords = [q2_coords]
com = pr.calcCenter(
self.query.select('name ' + ' '.join(self.query_lig_corr[0])))
self.query_distance = np.max(cdist(
[com], q_sel2_coords[0])) + self.rmsd_threshold
superpose_list = []
for q1 in q_sel1_coords:
for q2 in q_sel2_coords:
superpose_list.append(np.vstack((q1, q2)))
self.query_coords = superpose_list
def orient(pdb, selection='all'):
act = pdb.select(selection)
adj = prody.calcCenter(act)
oldcoords = act.getCoords()
newcoords = np.subtract(oldcoords, adj)
nncoords = varimax(newcoords)
trans = prody.calcTransformation(oldcoords, nncoords)
trans.apply(pdb)
return pdb
@author: victor
"""
import sys
import os
import glob
import prody
from hivprotmut.structures.pdbcuration import CurationSelections
if __name__ == '__main__':
final_db_folder = sys.argv[1]
com_file = sys.argv[2]
com_handler = open(com_file, "w")
ligand_folders = os.listdir(final_db_folder) # first level are ligands
txt_root = os.path.split(final_db_folder)[1]
for path in ligand_folders:
files = glob.glob(os.path.join(final_db_folder, path, "*.pdb"))
for pdb_file in files:
pdb = prody.parsePDB(pdb_file)
txt_pdb = os.path.split(pdb_file)[1]
ligand = pdb.select(CurationSelections.HEAVY_LIGAND_SELECTION)
com = prody.calcCenter(ligand)
txt_pdb_file = os.path.join(path, txt_pdb)
com_handler.write("%s %.3f %.3f %.3f\n"%(
txt_pdb_file,
com[0],
com[1],
com[2]
))
com_handler.close()
def blast(pdb_path):
cdir = os.getcwd()
tdir = tempfile.mkdtemp()
os.chdir(tdir)
receptor = os.path.basename(os.path.splitext(pdb_path)[0])
pdbHead = prody.parsePDBHeader(pdb_path)
pdbFile = prody.parsePDB(pdb_path)
ligands = []
for chem in pdbHead['chemicals']:
ligands.append([chem.chain, str(chem.resnum), chem.resname, chem.name])
blast_result = []
for chain, resnum, resname, name in ligands:
rec = pdbFile.select('not (chain {} resnum {})'.format(chain, resnum))
ligand = pdbFile.select('chain {} resnum {}'.format(chain, resnum))
cen_ligand = prody.calcCenter(ligand)
res_coll = []
ligCoords = ligand.getCoords()
print('lig_size', len(ligCoords))
sequence = ''
i = 4
while len(sequence) < 100:
for center in ligCoords:
around_atoms = rec.select(
'same residue as within {} of center'.format(i),
center=center)
if around_atoms is None:
continue
res_coll.append(around_atoms)
#res_indices = around_atoms.getResindices()
#print(around_atoms.getHierView()['A'].getSequence())
#print (res_indices)
#res_coll = res_coll | set(res_indices)
resindices = reduce(lambda x, y: x | y, res_coll)
sequence = resindices.getHierView()['A'].getSequence()
print('sequence', i, len(sequence), sequence)
i += 1
with open('sequence.fasta', 'w') as fout:
fout.write(">receptor\n" + sequence + '\n')
cmd = 'blastp -db {} -query sequence.fasta -outfmt 5 -out result'.format(
BLASTDB)
#print(os.getcwd())
cl = subprocess.Popen(cmd, shell=True, stdout=subprocess.PIPE)
cl.wait()
#print(os.listdir(os.getcwd()))
dtree = xml.dom.minidom.parse("result")
collection = dtree.documentElement
hits = collection.getElementsByTagName("Hit")
hit_result = []
for hit in hits:
hit_id = hit.getElementsByTagName('Hit_id')[0].childNodes[0].data
hsps = hit.getElementsByTagName('Hit_hsps')[0]
identity = hsps.getElementsByTagName(
'Hsp_identity')[0].childNodes[0].data
align_len = hsps.getElementsByTagName(
'Hsp_align-len')[0].childNodes[0].data
qseq = hsps.getElementsByTagName('Hsp_qseq')[0].childNodes[0].data
hseq = hsps.getElementsByTagName('Hsp_hseq')[0].childNodes[0].data
midline = hsps.getElementsByTagName(
'Hsp_midline')[0].childNodes[0].data
blast_result.append([
receptor, hit_id,
str(identity),
str(align_len),
str(len(sequence)), midline, hseq, sequence
])
return blast_result
def compute_atom_distances(pdb_target, res_file, output_report, chain="L"):
"""
This function calculate atom-atom distances for ligand and residue atoms. The residue number and atom names
(for both, ligand and residue) must be specified in a file ('res_file').
:param pdb_target: input PDB file path
:param res_file: file with instructions. This file must have n rows with three format: RESNUM LATOMNAME/SLIGAND
ATOMNAMES/SRESIDUE. If you want to calculate a distance using a center of mass write [ATOM1,ATOM2,ATOMN]
:param output_report:
:param chain:
:return:
"""
# Load PDB files
target = pdb2prody(pdb_target)
ligand = target.select("chain {}".format(chain))
print(ligand.getNames())
# Reading instructions from file
list_of_instructions = read_selecteds_from_file(res_file)
# Select the input atoms
report = []
for line in list_of_instructions:
resnum, atom_name_ref, atom_name_tar = line.split()
# If the user wants to select more than one atom he has to put them in a string with this format: [atom1,atomN...]
if "[" in atom_name_ref or "]" in atom_name_ref:
print("Multiple atom selection for the ligand")
atom_string_with_comas = atom_name_ref.strip("[").strip("]")
atom_list = atom_string_with_comas.split(",")
atom_string = ' '.join(atom_list)
atom_ref_selected = ligand.select("name {}".format(atom_string))
print("Selected atoms: {}".format(atom_ref_selected.getNames()))
else:
print("Single atom selection for the ligand")
atom_ref_selected = ligand.select("name {}".format(atom_name_ref))
print("Selected atom: {}".format(atom_ref_selected.getNames()))
if "[" in atom_name_tar or "]" in atom_name_tar:
print("Multiple atom selection for the system")
atom_string_with_comas = atom_name_tar.strip("[").strip("]")
atom_list = atom_string_with_comas.split(",")
atom_string = ' '.join(atom_list)
atom_tar_selected = select_atom_given_name_type_and_num(
target, resnum, atom_string)
print("Selected atoms: {}".format(atom_tar_selected.getNames()))
else:
print("Single atom selection for the system")
atom_tar_selected = select_atom_given_name_type_and_num(
target, resnum, atom_name_tar)
print("Selected atom: {}".format(atom_tar_selected.getNames()))
try:
number_of_selected_atoms_ref = len(atom_ref_selected.getNames())
except AttributeError:
exit(
"None atoms where selected. Please, check if the selected atoms exists in the ligand in {}"
.format(pdb_target))
try:
number_of_selected_atoms_tar = len(atom_tar_selected.getNames())
except AttributeError:
exit(
"None atoms where selected. Please, check if the selected atoms exists in the residue {} in {}"
.format(resnum, pdb_target))
# Now there are four possibilities: len 1 in target and ref, len > 1 in one of both and len >1 in both.
# If the len is more than 1 we will use the center of mass as a point to compute the distance.
if number_of_selected_atoms_ref <= 1 and number_of_selected_atoms_tar <= 1:
distance = prody.calcDistance(atom_tar_selected, atom_ref_selected)
elif number_of_selected_atoms_ref <= 1 and number_of_selected_atoms_tar > 1:
center_tar = prody.calcCenter(atom_tar_selected)
atom_coords = atom_ref_selected.getCoords()[0]
distance = np.linalg.norm(center_tar - atom_coords)
elif number_of_selected_atoms_ref > 1 and number_of_selected_atoms_tar <= 1:
center_ref = prody.calcCenter(atom_ref_selected)
atom_coords = atom_tar_selected.getCoords()[0]
distance = np.linalg.norm(atom_coords - center_ref)
else:
center_tar = prody.calcCenter(atom_tar_selected)
center_ref = prody.calcCenter(atom_ref_selected)
distance = np.linalg.norm(center_tar - center_ref)
report_line = "{:4} {:10} {:10} {:6.3f}\n".format(
resnum, ''.join(atom_ref_selected.getNames()),
''.join(atom_tar_selected.getNames()), distance)
report.append(report_line)
report_final = ''.join(report)
if output_report:
with open(output_report, "w") as report_file:
report_file.write(report_final)
print(report_final)
return report_final
def bundle_ligand_data(self,
pick_one,
fake_ligand=True,
OUT=True,
compare_ResId_native='default',
Id_suffix='default',
filename=None,
benchmark=None):
'''
:param pick_one:
:param fake_ligand:
:param OUT:
:param compare_ResId_native:
:param Id_suffix:
:param filename:
:param benchmark:
:return:
'''
PDB = self.PDBname
if fake_ligand == False:
ResId = str(pick_one.getResindex())
else:
ResId = compare_ResId_native + '_' + str(Id_suffix)
pdb_store_dir = self.store_dir
other = self.receptor
# Extract this ligand from protein (as input for openbabel)
if filename is None:
filename = pdb_store_dir + '/{1}/{0}_{1}_ligand.pdb'.format(
PDB, ResId)
if not os.path.isfile(filename):
if not os.path.exists(pdb_store_dir + '/' + ResId):
os.mkdir(pdb_store_dir + '/' + ResId)
if OUT:
try:
pd.writePDB(filename, pick_one)
tar_filename = ''.join(filename.split('.')[:-1])
tar_filename += '.mol'
pdb_to_mol2(filename, tar_filename)
except:
print 'Unexpected Error!'
logging.error('Cannot convert {} to mol2 format!'.format(
filename.split('/')[-1]))
return
if not os.path.isfile(filename):
if not os.path.exists(pdb_store_dir + '/' + ResId):
os.mkdir(pdb_store_dir + '/' + ResId)
naming = '{}_{}'.format(PDB, ResId)
# Get coordinate of center
xyz = pick_one.getCoords()
middle = pd.calcCenter(pick_one)
# in pi degree , the rotation of the box (if needed)
rotation = [0, 0, 0]
scale = max(
max(xyz[:, 0]) - middle[0], middle[0] - min(xyz[:, 0]),
max(xyz[:, 1]) - middle[1], middle[1] - min(xyz[:, 1]),
max(xyz[:, 2]) - middle[2], middle[2] - min(xyz[:, 2]))
# assert scale <= 10
if scale > self.BOX_range / 2:
logging.warning(
'Warning! {} has a ligand out of box scale with {} atom distance to center'
.format(PDB, scale))
# Now shifting the boxes:
max_scale = max(
max(xyz[:, 0]) - min(xyz[:, 0]),
max(xyz[:, 1]) - min(xyz[:, 1]),
max(xyz[:, 2]) - min(xyz[:, 2]))
if max_scale > self.BOX_range:
logging.error(
'Assertion failed, {} has a ligand out of box completely with scale'
.format(PDB, scale))
return
# Try to move to the new center
temp_mid = [(max(xyz[:, 0]) + min(xyz[:, 0])) / 2,
(max(xyz[:, 1]) + min(xyz[:, 1])) / 2,
(max(xyz[:, 2]) + min(xyz[:, 2])) / 2]
temp_mid[0] = round(temp_mid[0], 6)
temp_mid[1] = round(temp_mid[1], 6)
temp_mid[2] = round(temp_mid[2], 6)
middle = np.array(temp_mid)
print middle
# print middle
scale_extension = (self.BOX_range - self.BOX_size) / 2
box_num = int(np.ceil(self.BOX_range / self.BOX_size))
xx, yy, zz = np.meshgrid(
np.linspace(middle[0] - scale_extension,
middle[0] + scale_extension, box_num),
np.linspace(middle[1] - scale_extension,
middle[1] + scale_extension, box_num),
np.linspace(middle[2] - scale_extension,
middle[2] + scale_extension, box_num))
# print xx
vector = np.c_[xx.ravel(), yy.ravel(), zz.ravel()]
num_vector = [0] * len(vector)
#print len(vector), box_num
for atom in pick_one.iterAtoms():
x, y, z = atom.getCoords()
x_pos = int(round(x - vector[0][0]))
# assert 0 <= x_pos <= 19
y_pos = int(round(y - vector[0][1]))
# assert 0 <= y_pos <= 19
z_pos = int(round(z - vector[0][2]))
# assert 0 <= z_pos <= 19
if 0 <= x_pos < box_num and 0 <= y_pos < box_num and 0 <= z_pos < box_num:
# Simply change here to fulfill the mark as 'H_1'
# note (z(y(x))) follows from atuogrid map file format , otherwise the coordinate system is not correspond coorectly
num_vector[z_pos * box_num * box_num + y_pos * box_num +
x_pos] = atom.getName() + '_' + str(HETERO_PART)
# quick,dirty way to find atoms of protein in cubic boxes
pd.defSelectionMacro(
'inbox',
'abs(x-{1}) <= {0} and abs(y-{2}) <= {0} and abs(z-{3}) <= {0}'.
format(self.BOX_size / 2, middle[0], middle[1], middle[2]))
residues = other.select(
'protein and same residue as within 18 of center', center=middle)
if residues is None:
logging.warning('{} in {} has no atoms nearby'.format(ResId, PDB))
return
# This place might have some potential problem
# for ADP or ATP , they might either be part of nucleic and the ligand
# This will cause a severe bug when calculating autovina score
# TODO fix this issue
nearby = residues.select('inbox')
if nearby is not None:
for atom in nearby.iterAtoms():
x, y, z = atom.getCoords()
x_pos = int(round(x - vector[0][0]))
# assert 0 <= x_pos <= 19
y_pos = int(round(y - vector[0][1]))
# assert 0 <= y_pos <= 19
z_pos = int(round(z - vector[0][2]))
# assert 0 <= z_pos <= 19
temp = z_pos * box_num * box_num + y_pos * box_num + x_pos
if 0 <= x_pos < box_num and 0 <= y_pos < box_num and 0 <= z_pos < box_num and num_vector[
temp] == 0:
# Simply change here to fulfill the mark as 'C_2'
num_vector[temp] = atom.getName() + '_' + str(PROTEIN_PART)
else:
# num_vector[temp] += '|'+atom.getName() + '_' + str(PROTEIN_PART)
print atom.getName()
logging.warning('Coorinate {} {} {} found at {}'.format(
x_pos, y_pos, z_pos, self.PDBname))
# Save into the dict for future locating
# naming = '{}_{}'.format(PDB, ResId)
# Do autogrid mapgeneration:
# ligand_filename = os.path.join(temp_pdb_PREFIX, PDB + '/' + naming + '_ligand.pdb')
# receptor_filename = os.path.join(temp_pdb_PREFIX, PDB + '/' + naming + '_receptor.pdb')
# complex_filename = os.path.join(temp_pdb_PREFIX, PDB + '/' + naming + '_complex.pdb')
# fake_ligand_filename = os.path.join(temp_pdb_PREFIX, 'fake-ligand.pdb')
self.heterodict[ResId] = {
'raw_vector': num_vector,
'center': middle,
'rotation': rotation,
'naming': '{}_{}'.format(PDB, ResId),
'chain': 'NA',
'filename': filename,
'id': ResId,
'Resname': 'NA',
'ligand': pick_one,
'protein': residues,
'vina_score': 'NA',
'original_one': True,
'file_generated': False,
'fake_ligand': True,
'RMSF': 0,
'Contact Similarity': 1,
'gridmap_protein': 'NA',
'gridmap_ligand': 'NA',
'gridmap_complex': 'NA'
}
if fake_ligand == True:
try:
dist = self._calcRMSD(
self.heterodict[compare_ResId_native]['ligand'],
pick_one,
benchmark=benchmark)
print dist
self.heterodict[ResId]['RMSF'] = dist
except:
print 'oops'
raise IOError
self.heterodict[ResId]['Contact Similarity'] = self._calcQ(
self.heterodict[compare_ResId_native]['ligand'],
pick_one,
benchmark=benchmark)
else:
self.heterodict[ResId]['Resname'] = pick_one.getResname()
self.heterodict[ResId]['chain'] = pick_one.getChid()
def process_residue_into_vector(self,
prody_ligand,
prody_residue,
distance_cutoff_polar=3.5,
distance_cutoff_greasy=4):
"""
Converting each residue into a representative vector
Metrics I care about for each residue:
* Contact distance
* Residue characteristics
* Amino acid identity or degenerate amino acid groups?
* Amino acid chemical characteristics e.g. [Sigma Amino Acid Reference Chart]
(http://www.sigmaaldrich.com/life-science/metabolomics/learning-center/amino-acid-reference-chart.html)
* Position of residue relative to fragment
* Vector from fragment centroid to {residue centroid | closest residue atom }
* Backbone or side chain
{Angstroms} - X component, vector from fragment centroid to closest residue atom
{Angstroms} - Y component, vector from fragment centroid to closest residue atom
{Angstroms} - Z component, vector from fragment centroid to closest residue atom
{ 0 | 1 } - Backbone contact OR Sidechain contact
{ 0 | 1 } - Ligand Polar Contact OR Ligand Non-polar Contact
{ 0 | 1 } - Side chain has hydrogen bond donor/acceptor (DEHKNQRSTY)
{ 0 | 1 } - Hydrophobic, aliphatic (AILV)
{ 0 | 1 } - Hydrophobic, aromatic (FWY)
{ 0 | 1 } - Polar (NCQMST)
{ 0 | 1 } - Charged, Acidic (DE)
{ 0 | 1 } - Charged, Basic (HKR)
{ 0 | 1 } - Glycine
{ 0 | 1 } - Proline
{ 0 | 1 } - Backbone carbonyl
{ 0 | 1 } - Backbone amino
{ 0 | 1 } - Backbone C/CA
:return:
"""
min_contact_distance, row_index_low, column_index_low = minimum_contact_distance(
prody_residue, prody_ligand, return_indices=True)
polar_residues = [
'ASP', 'GLU', 'HIS', 'LYS', 'ASN', 'GLN', 'ARG', 'SER', 'THR',
'TYR'
]
if all([
prody_residue.getResnames()[0] in polar_residues,
min_contact_distance > distance_cutoff_polar
]):
return None
elif min_contact_distance > distance_cutoff_greasy:
return None
else:
residue_contact_atom = prody_residue.copy().select(
'index {}'.format(row_index_low))
ligand_contact_atom = prody_ligand.copy().select(
'index {}'.format(column_index_low))
# Save min contact residue and ligand atom indicies for evaluating cluster quality later
# self.min_contact_distance = min_contact_distance
# Residue Contact Type
residue_contact_type = 0 if residue_contact_atom.getNames()[0] in [
'C', 'CA', 'N', 'O'
] else 1
# Ligand Contact Type
ligand_contact_type = 1 if ligand_contact_atom.getNames(
)[0][0] in ['C'] else 0
# Vector from fragment centroid to closest residue atom
contact_vector = (residue_contact_atom.getCoords() -
prody.calcCenter(prody_ligand))[0]
# contact_unit_vector = contact_vector / np.linalg.norm(contact_vector)
# Side chain has hydrogen bond donor/acceptor (DEHKNQRSTY)
h_bond_donor_acceptor = 1 if residue_contact_atom.getResnames(
)[0] in polar_residues else 0
# Polar atom on residue is contacting ligand
residue_polar_contact = 1 if residue_contact_atom.getNames(
)[0][0] in ['O', 'N'] else 0
# Residue characteristics
# todo: UPDATE so that only one of the below can hold value of 1 at any given time
# {0 | 1} - Hydrophobic, aliphatic(AILVC)
greasy_ali = 1 if residue_contact_atom.getResnames()[0] in [
'ALA', 'ILE', 'LEU', 'VAL', 'CYS'
] else 0
# {0 | 1} - Hydrophobic, aromatic(FWY)
greasy_aro = 1 if residue_contact_atom.getResnames()[0] in [
'PHE', 'TYR', 'TRP'
] else 0
# {0 | 1} - Polar(NCQMST)
polar = 1 if residue_contact_atom.getResnames()[0] in [
'ASN', 'CYS', 'GLN', 'MET', 'SER', 'THR'
] else 0
# {0 | 1} - Charged, Acidic(DE)
charged_acid = 1 if residue_contact_atom.getResnames()[0] in [
'ASP', 'GLU'
] else 0
# {0 | 1} - Charged, Basic(HKR)
charged_basic = 1 if residue_contact_atom.getResnames()[0] in [
'HIS', 'LYS', 'ARG'
] else 0
# {0 | 1} - Glycine
glycine = 1 if residue_contact_atom.getResnames()[0] in ['GLY'
] else 0
# {0 | 1} - Proline
proline = 1 if residue_contact_atom.getResnames()[0] in ['PRO'
] else 0
# {0 | 1} - Backbone carbonyl
bb_carbonyl = 1 if residue_contact_atom.getNames()[0] in ['O'
] else 0
# {0 | 1} - Backbone amino
bb_amino = 1 if residue_contact_atom.getNames()[0] in ['N'] else 0
# {0 | 1} - Backbone C / CA
bb_c_ca = 1 if residue_contact_atom.getNames()[0] in ['C', 'CA'
] else 0
categorical_vector = [
residue_contact_type, ligand_contact_type,
h_bond_donor_acceptor, residue_polar_contact, greasy_ali,
greasy_aro, polar, charged_acid, charged_basic, glycine,
proline, bb_carbonyl, bb_amino, bb_c_ca
]
self.categorical_array = np.asanyarray(categorical_vector)
self.contact_vector = contact_vector
return True
"""
Created on 1/9/2014
@author: victor
"""
import sys
import os
import glob
import prody
from hivprotmut.structures.pdbcuration import CurationSelections
if __name__ == '__main__':
final_db_folder = sys.argv[1]
com_file = sys.argv[2]
com_handler = open(com_file, "w")
ligand_folders = os.listdir(final_db_folder) # first level are ligands
txt_root = os.path.split(final_db_folder)[1]
for path in ligand_folders:
files = glob.glob(os.path.join(final_db_folder, path, "*.pdb"))
for pdb_file in files:
pdb = prody.parsePDB(pdb_file)
txt_pdb = os.path.split(pdb_file)[1]
ligand = pdb.select(CurationSelections.HEAVY_LIGAND_SELECTION)
com = prody.calcCenter(ligand)
txt_pdb_file = os.path.join(path, txt_pdb)
com_handler.write("%s %.3f %.3f %.3f\n" %
(txt_pdb_file, com[0], com[1], com[2]))
com_handler.close()
def __init__(self, parsed_pdb, comb):
"""instance of class IntFG has attributes including selection names, neighboring atoms, neighborhood density
of atoms, etc."""
self.sele = parsed_pdb.possible_ifgs.pop()
self.resindex, self._ind = np.unique(self.sele.getResindices(), return_index=True)
self.resname = self.sele.getResnames()[self._ind]
self.resnum = self.sele.getResnums()[self._ind]
self.atom_names = {resname: self.sele.select('resindex ' + str(resindex)).getNames() for resname, resindex in
zip(self.resname, self.resindex)}
self.chid = np.unique(self.sele.getChids())[0]
self.center_coords = pr.calcCenter(self.sele)
self.vdm_count = 1
self.count = comb.ifg_count
self.sasa = None
self.residue_sasa = None
self.dssp_sasa = None
self.sasa_3A_probe = None
self.sasa_4A_probe = None
self.sasa_5A_probe = None
self.contact_atoms_all = None
self.contact_atoms_protein = None
self.contact_resnums = None
self.contact_chids = None
self.contact_resindices = None
self.contact_segments = None
self.contact_atoms_water = None
self.contact_atoms_ligand = None
self.contact_atoms_metal = None
self.contact_info_water = []
self.contact_info_ligand = []
self.contact_info_metal = []
self.contact_info_protein = []
self.contact_dict = collections.defaultdict(set)
self.contact_pair_dict = collections.defaultdict(list)
self.probe_hbonds = []
self.rotamer = None
self.min_hull_dist_ifg = None
self.min_hull_dist_cb_ca = None
self.cbeta_density = None
self.heavy_atom_density_5A = None
self.heavy_atom_density_10A = None
if comb.ifg_seq_str != 'element':
self.ifg_frag = parsed_pdb.prody_pdb.select('segment A and chain ' + self.chid + ' and resnum `' + str(np.min(self.resnum)-1)
+ 'to' + str(np.max(self.resnum)+1) + '`')
else:
self.ifg_frag = self.sele
self.frag_length = len(np.unique(self.ifg_frag.getResindices()))
if comb.ifg_seq_str != 'element':
self.sequence = ''.join(one_letter_code[rn] for rn in self.resname)
else:
self.sequence = ''
self.sec_struct_dssp = None
self.sec_struct_phi_psi = None
self.contact_number_water = None
self.per_res_contact_number_water = None
self.contact_atom_names_water = None
self.contact_resnames_water = None
self.contact_resnums_water = None
self.contact_number_ligand = None
self.per_res_contact_number_ligand = None
self.contact_atom_names_ligand = None
self.contact_resnames_ligand = None
self.contact_resnums_ligand = None
self.contact_number_metal = None
self.per_res_contact_number_metal = None
self.contact_atom_names_metal = None
self.contact_resnames_metal = None
self.contact_resnums_metal = None
self.hbond_atom_names = []
self.hbond_resnames = []
self.hbond_resnums = []
self.hbond_angle = []
self.hbond_dist_acc_hyd = []
self.hbond_dist_heavy = []
self.hbond_atom_names_water = []
self.hbond_number_water = []
self.hbond_resnames_water = []
self.hbond_resnums_water = []
self.hbond_angle_water = []
self.hbond_dist_acc_hyd_water = []
self.hbond_dist_heavy_water = []
self.hbond_number_ligand = []
self.hbond_atom_names_ligand = []
self.hbond_resnames_ligand = []
self.hbond_resnums_ligand = []
self.hbond_angle_ligand = []
self.hbond_dist_acc_hyd_ligand = []
self.hbond_dist_heavy_ligand = []
self.ca_hbond_atom_names = []
self.ca_hbond_resnames = []
self.ca_hbond_resnums = []
self.ca_hbond_angle = []
self.ca_hbond_dist_acc_hyd = []
self.ca_hbond_dist_heavy = []
self.bb_cb_atom_ind = self.get_bb_cb_atom_indices(parsed_pdb)
def binding_pocket_selection(pose_store, p, ligand_name, selection_radius,
center):
'''
This function will find by default mass center
of the ligand using Prody.
If the -gc option is selected the spatial center of
the ligand is used by computing the mean distance between the
furthest x axis coordinates.
If the -ds option is selected a dobule sphere procedure is followed
by selecting the furthest x axis atoms.
'''
amino = [
'CYS', 'ASP', 'SER', 'GLN', 'LYS', 'ILE', 'PRO', 'THR', 'PHE', 'ASN',
'GLY', 'HIS', 'LEU', 'ARG', 'TRP', 'ALA', 'VAL', 'GLU', 'TYR', 'MET'
]
two_let_atom_code = ['Br', 'FE']
coord = []
x_coord = []
binding_pocket = []
ligand = []
min_coord = None
max_coord = None
for pose in pose_store:
structure = pose_store[pose].split('\n')
for line in structure:
if line[17:20] == ligand_name:
coord.append(float(line[30:38]))
coord.append(float(line[38:46]))
coord.append(float(line[46:54]))
ligand_atom = (line[17:20].strip(), line[12:16].strip(),
line[30:38].strip(), line[38:46].strip(),
line[46:54].strip(), line[-2:].strip())
ligand.append(ligand_atom)
ligand_atom = ()
for i in range(0, len(coord), 3):
x_coord.append(float(coord[i]))
if center == 'double':
print('\nDouble center of ligand selected')
x_out_left = coord[coord.index(min(x_coord))]
y_out_left = coord[coord.index(min(x_coord)) + 1]
z_out_left = coord[coord.index(min(x_coord)) + 2]
x_out_right = coord[coord.index(max(x_coord))]
y_out_right = coord[coord.index(max(x_coord)) + 1]
z_out_right = coord[coord.index(max(x_coord)) + 2]
print('\n')
print("Left sphere center coordinates: ", x_out_left, y_out_left,
z_out_left)
print("Right sphere center coordinates: ", x_out_right, y_out_right,
z_out_right)
print("Spheres radii: ", selection_radius)
for line in structure:
if (line[0:6].strip() == "ATOM" or line[0:6].strip()
== "HETATM") and line[17:20].strip() in amino:
x1 = math.pow((float(line[30:38]) - x_out_left), 2)
y1 = math.pow((float(line[38:46]) - y_out_left), 2)
z1 = math.pow((float(line[46:54]) - z_out_left), 2)
if (x1 + y1 + z1) <= selection_radius**2:
atom = (line[17:20].strip(), line[12:16].strip(),
line[30:38].strip(), line[38:46].strip(),
line[46:54].strip(), line[-2:].strip())
binding_pocket.append(atom)
atom = ()
for line in structure:
if (line[0:6].strip() == "ATOM" or line[0:6].strip()
== "HETATM") and line[17:20].strip() in amino:
x1 = math.pow((float(line[30:38]) - x_out_right), 2)
y1 = math.pow((float(line[38:46]) - y_out_right), 2)
z1 = math.pow((float(line[46:54]) - z_out_right), 2)
if (x1 + y1 + z1) <= selection_radius**2:
if line[-3:].strip() in two_let_atom_code:
atom = (line[17:20].strip(), line[12:16].strip(),
line[30:38].strip(), line[38:46].strip(),
line[46:54].strip(), line[-3:].strip())
else:
atom = (line[17:20].strip(), line[12:16].strip(),
line[30:38].strip(), line[38:46].strip(),
line[46:54].strip(), line[-3:].strip()[0])
binding_pocket.append(atom)
atom = ()
print("Number of protein atoms selected: {}".format(
len(binding_pocket)))
print('Number of ligand atoms selected: {}'.format(len(ligand)))
print('Total number of atoms selected: {}'.format(
len(binding_pocket) + len(ligand)))
elif center == 'geometric':
print("\nGeometric ligand center selected")
sphere_center_x = (coord[coord.index(max(x_coord))] +
coord[coord.index(min(x_coord))]) / 2
sphere_center_y = (coord[coord.index(max(x_coord)) + 1] +
coord[coord.index(min(x_coord)) + 1]) / 2
sphere_center_z = (coord[coord.index(max(x_coord)) + 2] +
coord[coord.index(min(x_coord)) + 2]) / 2
print('\n')
print("Sphere center coordinates: ", sphere_center_x, sphere_center_y,
sphere_center_z)
print("Sphere radius: ", selection_radius)
for line in structure:
if (line[0:6].strip() == "ATOM" or line[0:6].strip()
== "HETATM") and line[17:20].strip() in amino:
x1 = math.pow((float(line[30:38]) - sphere_center_x), 2)
y1 = math.pow((float(line[38:46]) - sphere_center_y), 2)
z1 = math.pow((float(line[46:54]) - sphere_center_z), 2)
if (x1 + y1 + z1) <= selection_radius**2:
if line[-3:].strip() in two_let_atom_code:
atom = (line[17:20].strip(), line[12:16].strip(),
line[30:38].strip(), line[38:46].strip(),
line[46:54].strip(), line[-3:].strip())
else:
atom = (line[17:20].strip(), line[12:16].strip(),
line[30:38].strip(), line[38:46].strip(),
line[46:54].strip(), line[-3:].strip()[0])
binding_pocket.append(atom)
atom = ()
print("Number of protein atoms selected: {}".format(
len(binding_pocket)))
print('Number of ligand atoms selected: {}'.format(len(ligand)))
print('Total number of atoms selected: {}'.format(
len(binding_pocket) + len(ligand)))
elif center == 'mass':
print('\nLigand mass center selected')
ligand_selection = p.select('not water and hetero')
weights = ligand_selection.getMasses()
mass_center = calcCenter(ligand_selection, weights)
sphere_center_x = mass_center[0]
sphere_center_y = mass_center[1]
sphere_center_z = mass_center[2]
print("\nSphere center coordinates: {}, {}, {}".format(
sphere_center_x, sphere_center_y, sphere_center_z))
print("Sphere radius: {}".format(selection_radius))
for line in structure:
if (line[0:6].strip() == "ATOM" or line[0:6].strip()
== "HETATM") and line[17:20].strip() in amino:
x1 = math.pow((float(line[30:38]) - sphere_center_x), 2)
y1 = math.pow((float(line[38:46]) - sphere_center_y), 2)
z1 = math.pow((float(line[46:54]) - sphere_center_z), 2)
if (x1 + y1 + z1) <= selection_radius**2:
if line[-3:].strip() in two_let_atom_code:
atom = (line[17:20].strip(), line[12:16].strip(),
line[30:38].strip(), line[38:46].strip(),
line[46:54].strip(), line[-3:].strip())
else:
atom = (line[17:20].strip(), line[12:16].strip(),
line[30:38].strip(), line[38:46].strip(),
line[46:54].strip(), line[-3:].strip()[0])
binding_pocket.append(atom)
atom = ()
print("Number of protein atoms selected: {}".format(
len(binding_pocket)))
print('Number of ligand atoms selected: {}'.format(len(ligand)))
print('Total number of atoms selected: {}'.format(
len(binding_pocket) + len(ligand)))
return binding_pocket, ligand
def append_vectors(self, hetero_file):
'''
Append each docked result as a vector to the dict
:param hetero_file: file position
:return: nothing , but will generate a vector into the dict
'''
#need to split the files
TEMP = 'temp.pdb'
o = open(hetero_file, 'r')
one_pdb = ''
for line in o:
one_pdb += line
if 'END' in line:
#write a temporial file
with open(TEMP, 'wb') as w:
w.write(one_pdb)
w.close()
one_pdb = ''
pdb = pd.parsePDB(TEMP)
if pdb.numAtoms() <= 3:
continue
# Get coordinate of center
xyz = pdb.getCoords()
middle = pd.calcCenter(pdb)
scale = max(
max(xyz[:, 0]) - middle[0], middle[0] - min(xyz[:, 0]),
max(xyz[:, 1]) - middle[1], middle[1] - min(xyz[:, 1]),
max(xyz[:, 2]) - middle[2], middle[2] - min(xyz[:, 2]))
# assert scale <= 10
if scale > 10:
logging.warning(
'Warning! {} has a ligand out of box scale with {} atom distance to center'
.format(self.PDBname, scale))
# Now shifting the boxes:
max_scale = max(
max(xyz[:, 0]) - min(xyz[:, 0]),
max(xyz[:, 1]) - min(xyz[:, 1]),
max(xyz[:, 2]) - min(xyz[:, 2]))
if max_scale > 20:
logging.error(
'Assertion failed, {} has a ligand out of box completely with scale'
.format(self.PDBname, scale))
continue
# Try to move to the new center
middle = [(max(xyz[:, 0]) + min(xyz[:, 0])) / 2,
(max(xyz[:, 1]) + min(xyz[:, 1])) / 2,
(max(xyz[:, 2]) + min(xyz[:, 2])) / 2]
xx, yy, zz = np.meshgrid(
np.linspace(middle[0] - 9.5, middle[0] + 9.5, 20),
np.linspace(middle[1] - 9.5, middle[1] + 9.5, 20),
np.linspace(middle[2] - 9.5, middle[2] + 9.5, 20))
# print xx
vector = np.c_[xx.ravel(), yy.ravel(), zz.ravel()]
num_vector = [0] * 8000
for atom in pdb.iterAtoms():
x, y, z = atom.getCoords()
x_pos = int(round(x - vector[0][0]))
# assert 0 <= x_pos <= 19
y_pos = int(round(y - vector[0][1]))
# assert 0 <= y_pos <= 19
z_pos = int(round(z - vector[0][2]))
# assert 0 <= z_pos <= 19
if 0 <= x_pos <= 19 and 0 <= y_pos <= 19 and 0 <= z_pos <= 19:
# Simply change here to fulfill the mark as 'H_1'
num_vector[
x_pos * 400 + y_pos * 20 +
z_pos] = atom.getName() + '_' + str(HETERO_PART)
# quick,dirty way to find atoms of protein in cubic boxes
pd.defSelectionMacro(
'inbox',
'abs(x-{}) < 10 and abs(y-{}) < 10 and abs(z-{}) < 10'.
format(middle[0], middle[1], middle[2]))
nearby = self.protein.select('inbox')
if nearby is not None:
for atom in nearby.iterAtoms():
x, y, z = atom.getCoords()
x_pos = int(round(x - vector[0][0]))
# assert 0 <= x_pos <= 19
y_pos = int(round(y - vector[0][1]))
# assert 0 <= y_pos <= 19
z_pos = int(round(z - vector[0][2]))
# assert 0 <= z_pos <= 19
if 0 <= x_pos <= 19 and 0 <= y_pos <= 19 and 0 <= z_pos <= 19 and num_vector[
x_pos * 400 + y_pos * 20 + z_pos] == 0:
# Simply change here to fulfill the mark as 'C_2'
num_vector[x_pos * 400 + y_pos * 20 +
z_pos] = atom.getName() + '_' + str(
PROTEIN_PART)
else:
logging.warning(
'Coorinate {} {} {} found at {}'.format(
x_pos, y_pos, z_pos, self.PDBname))
# This is for checking the correctness when we add atoms in proteins.
# filename2= 'data/{}_{}_2.pdb'.format(PDB, ResId)
# writePDB(filename2, pick_one+nearby)
# Save into the dict for future locating
self.heterodict[str(self.ct)] = {
'raw_vector':
num_vector,
'center':
middle,
'filename':
hetero_file,
'id':
hetero_file.split('/')[-1].split('.')[0] + '_' +
str(self.ct)
}
self.ct += 1
def find_possible_ifgs_rmsd(self, comb, rmsd_threshold=1.0):
"""uses iFG definitions in comb object to select iFGs in the parsed protein object that have all atoms
and occupancies = 1.
"""
possible_ifgs = []
if comb.num_res_ifg_query == 1:
poss_ifg_sel = self.prody_pdb.select('segment A and chain ' +
self.pdb_chain +
' sequence "' +
comb.ifg_seq_str_query + '"')
if poss_ifg_sel is not None:
ifg_resindices, indices = np.unique(
poss_ifg_sel.getResindices(), return_index=True)
ifg_resnames = poss_ifg_sel.getResnames()[indices]
for ifg_resindex, ifg_resname in zip(ifg_resindices,
ifg_resnames):
ifg_selection = self.prody_pdb.select(
'resindex ' + str(ifg_resindex) + ' and name ' +
comb.ifg_sele_dict_query[1][ifg_resname])
if ifg_selection is not None:
num_atoms = len(ifg_selection)
if num_atoms == len(comb.ifg_sele_dict_query[1]
[ifg_resname].split()):
if all(ifg_selection.getResnums() > 0):
possible_ifgs.append(ifg_selection)
comb.total_possible_ifgs += len(possible_ifgs)
else:
poss_ifg_sel = self.prody_pdb.select('segment A and chain ' +
self.pdb_chain +
' sequence "' +
comb.ifg_seq_str_query + '"')
if poss_ifg_sel is not None:
ifg_resindices_cat_list, indices = np.unique(
poss_ifg_sel.getResindices(), return_index=True)
ifg_resnames_cat_list = poss_ifg_sel.getResnames()[indices]
ifg_resindex_pairs = [
ifg_resindices_cat_list[i:i + 2]
for i in range(0, len(ifg_resindices_cat_list), 2)
]
ifg_resname_pairs = [
ifg_resnames_cat_list[i:i + 2]
for i in range(0, len(ifg_resnames_cat_list), 2)
]
for ifg_resindex_pair, ifg_resname_pair in zip(
ifg_resindex_pairs, ifg_resname_pairs):
resind1, resind2 = ifg_resindex_pair
resname1, resname2 = ifg_resname_pair
try:
ifg_selection = self.prody_pdb.select(
'(resindex ' + str(resind1) + ' and name ' +
comb.ifg_sele_dict_query[1][resname1] + ')' +
' or (resindex ' + str(resind2) + ' and name ' +
comb.ifg_sele_dict_query[2][resname2] + ')')
except KeyError:
print('Non-canonical residue in iFG, skipping.')
ifg_selection = None
if ifg_selection is not None:
num_atoms = len(ifg_selection)
names = comb.ifg_sele_dict_query[1][resname1].split()
names.extend(
comb.ifg_sele_dict_query[2][resname2].split())
if num_atoms == len(names):
if all(ifg_selection.getResnums() > 0):
possible_ifgs.append(ifg_selection)
comb.total_possible_ifgs += len(possible_ifgs)
passed_possible_ifgs = []
for pifg in possible_ifgs:
com = pr.calcCenter(
pifg.select('name ' + ' '.join(comb.query_names[0])))
q2_sel = self.prody_pdb.select(
'name ' + ' '.join(comb.query_names[1]) + ' within ' +
str(comb.query_distance) + ' of center',
center=com)
if q2_sel is not None:
resinds_query2s = np.unique(q2_sel.getResindices())
q_sel1_coords = [
pifg.select('name ' + n).getCoords()[0]
for n in comb.query_names[0]
]
for resind in resinds_query2s:
q_sel = self.prody_pdb.select(
'name ' + ' '.join(comb.query_names[1]) +
' and resindex ' + str(resind))
if len(q_sel) == len(comb.query_names[1]):
q_sel2_coords = [
q_sel.select('name ' + n).getCoords()[0]
for n in comb.query_names[1]
]
pifg_coords = np.vstack((q_sel1_coords, q_sel2_coords))
for coords in comb.query_coords:
R, m_com, t_com = get_rot_trans(
coords, pifg_coords)
coords_transformed = np.dot(
(coords - m_com), R) + t_com
rmsd = pr.calcRMSD(coords_transformed, pifg_coords)
if rmsd <= rmsd_threshold:
passed_possible_ifgs.append(
q_sel
) # This only takes the query2 selection as the iFG.
break
return passed_possible_ifgs
def compute_center_of_chain(pdb_object, chain="L"):
molecule_to_center = pdb_object.select("chain {}".format(chain))
center = prody.calcCenter(molecule_to_center)
return center