def generate_library(pharmacophore_path, output_format, library_dict,
library_path, pyrod_pharmacophore, directory, debugging):
""" This function writes a combinatorial pharmacophore library. """
logger = setup_logger('library', directory, debugging)
update_user('Starting library generation.', logger)
template_pharmacophore = pharmacophore_reader(pharmacophore_path,
pyrod_pharmacophore, logger)
pharmacophore_library = []
essential_hb, essential_hi, essential_ai, essential_ii = [], [], [], []
optional_hb, optional_hi, optional_ai, optional_ii = [], [], [], []
exclusion_volumes = []
# analyzing pharmacophore
for index, feature in enumerate(template_pharmacophore):
if feature[1] == 'ev':
exclusion_volumes.append(feature)
else:
if feature[1] in ['ha', 'hd', 'ha2', 'hd2', 'hda']:
if feature[2] == 'O':
optional_hb.append(index)
else:
essential_hb.append(index)
elif feature[1] == 'hi':
if feature[2] == 'O':
optional_hi.append(index)
else:
essential_hi.append(index)
elif feature[1] in ['pi', 'ni']:
if feature[2] == 'O':
optional_ii.append(index)
else:
essential_ii.append(index)
elif feature[1] == 'ai':
if feature[2] == 'O':
optional_ai.append(index)
else:
essential_ai.append(index)
essential_features = essential_hb + essential_hi + essential_ai + essential_ii
for hb_combination in combine_features(
optional_hb,
library_dict['minimal hydrogen bonds'] - len(essential_hb),
library_dict['maximal hydrogen bonds'] - len(essential_hb) + 1):
for hi_combination in combine_features(
optional_hi, library_dict['minimal hydrophobic interactions'] -
len(essential_hi),
library_dict['maximal hydrophobic interactions'] -
len(essential_hi) + 1):
for ai_combination in combine_features(
optional_ai,
library_dict['minimal aromatic interactions'] -
len(essential_ai),
library_dict['maximal aromatic interactions'] -
len(essential_ai) + 1):
for ii_combination in combine_features(
optional_ii,
library_dict['minimal ionizable interactions'] -
len(essential_ii),
library_dict['maximal ionizable interactions'] -
len(essential_ii) + 1):
pharmacophore = (essential_features + hb_combination +
hi_combination + ai_combination +
ii_combination)
if evaluate_pharmacophore(pharmacophore,
template_pharmacophore,
library_dict,
pyrod_pharmacophore):
pharmacophore_library.append(pharmacophore)
# estimate maximal library size and ask user if number and space of pharmacophores is okay
pharmacophore_writer(template_pharmacophore, [output_format],
'template_pharmacophore', library_path, logger)
pharmacophore_library_size = bytes_to_text(
os.path.getsize('{}/{}.{}'.format(
library_path, 'template_pharmacophore', output_format)) *
len(pharmacophore_library))
user_prompt = ''
while user_prompt not in ['yes', 'no']:
user_prompt = input(
'{} pharmacophores will be written taking about {} of space.\n'
'Do you want to continue? [yes/no]: '.format(
len(pharmacophore_library), pharmacophore_library_size))
if user_prompt == 'no':
sys.exit()
start = time.time()
# write pharmacophores
maximal_exclusion_volume_id = max(
[exclusion_volume[0] for exclusion_volume in exclusion_volumes])
for counter, index_pharmacophore in enumerate(pharmacophore_library):
extra_exclusion_volumes = []
extra_ev_counter = 1
pharmacophore = []
for index_feature in index_pharmacophore:
feature = template_pharmacophore[index_feature]
feature[2] = 'M'
pharmacophore.append(feature)
if feature[1] in ['ha', 'hd', 'ha2', 'hd2', 'hda']:
extra_exclusion_volumes.append([
maximal_exclusion_volume_id + extra_ev_counter, 'ev', 'M',
feature[5][0], 1.0, [], 0.0, 0.0
])
extra_ev_counter += 1
if feature[1] in ['ha2', 'hd2', 'hda']:
extra_exclusion_volumes.append([
maximal_exclusion_volume_id + extra_ev_counter, 'ev',
'M', feature[5][1], 1.0, [], 0.0, 0.0
])
extra_ev_counter += 1
pharmacophore_writer(
pharmacophore + exclusion_volumes + extra_exclusion_volumes,
[output_format], str(counter), library_path, logger)
update_progress(
(counter + 1) / len(pharmacophore_library),
'Writing {} pharmacophores'.format(len(pharmacophore_library)),
((time.time() - start) /
(counter + 1)) * (len(pharmacophore_library) - (counter + 1)))
update_user('Wrote pharmacophores to {}.'.format(library_path), logger)
return
def generate_exclusion_volumes(dmif, directory, debugging, shape_cutoff,
restrictive):
""" This function generates exclusion volumes. The exclusion volumes are described with a list of properties as
follows.
Format:
------------------------------------------------------------------------
0 1 2 3 4 5 6 7
------------------------------------------------------------------------
0 ev M [0.0,0.0,0.0] 1.0 [] 0.0 1.0
1 ev M [2.0,0.0,0.0] 1.0 [] 0.0 1.0
------------------------------------------------------------------------
Legend:
0 - index
1 - type
2 - flag (O - optional, M - mandatory)
3 - core position
4 - core tolerance [not needed for exclusion volumes]
5 - partner positions [not needed for exclusion volumes]
6 - partner tolerance [not needed for exclusion volumes]
7 - weight
"""
logger = setup_logger('exclusion_volumes', directory, debugging)
update_user('Generating exclusion volumes.', logger)
grid_space = 0.5
exclusion_volume_space = 4
if restrictive:
exclusion_volume_space = 2
grid_tree = cKDTree([[x, y, z]
for x, y, z in zip(dmif['x'], dmif['y'], dmif['z'])])
dtype = [('x', float), ('y', float), ('z', float), ('shape', int),
('count', int)]
dmif_shape = np.array(
[(x, y, z, shape, 0)
for x, y, z, shape in zip(dmif['x'], dmif['y'], dmif['z'],
dmif['shape']) if shape < shape_cutoff],
dtype=dtype)
positions = np.array([[
x, y, z
] for x, y, z in zip(dmif_shape['x'], dmif_shape['y'], dmif_shape['z'])])
shape_tree = cKDTree(positions)
shape_grid_size = len(dmif_shape)
# store number of neighbors with shape score smaller than shape_cutoff for grid points
for index in range(shape_grid_size):
dmif_shape['count'][index] = len(
shape_tree.query_ball_point(positions[index], grid_space * 4))
# sort for neighbor count
dmif_shape = np.sort(dmif_shape, order='count')
# rebuild positions and shape_tree
positions = np.array([[
x, y, z
] for x, y, z in zip(dmif_shape['x'], dmif_shape['y'], dmif_shape['z'])])
shape_tree = cKDTree(positions)
used = []
exclusion_volumes = []
counter = 1
start = time.time()
for index in range(shape_grid_size):
# grid_point index should not be in used list
if index not in used:
neighbor_list = shape_tree.query_ball_point(
positions[index], exclusion_volume_space / 2)
# elements of neighbor_list should not be in used list
if len(set(neighbor_list +
used)) == len(neighbor_list) + len(used):
# grid_point should not be at border of grid
if len(
grid_tree.query_ball_point(positions[index],
r=grid_space * 2)) == 33:
# grid_point should not be directly at border of binding pocket
if len(
shape_tree.query_ball_point(positions[index],
r=grid_space)) == 7:
# grid_point should not be surrounded by grid_points outside the binding pocket
if len(
shape_tree.query_ball_point(
positions[index], r=grid_space * 2)) < 33:
exclusion_volumes.append([
counter, 'ev', 'M', positions[index], 1.0, [],
0.0, 1.0
])
counter += 1
used += neighbor_list
eta = ((time.time() - start) / (index + 1)) * (shape_grid_size -
(index + 1))
update_progress(
float(index + 1) / shape_grid_size,
'Progress of exclusion volume generation', eta)
logger.debug('Passed grid index {}.'.format(index))
update_user(
'Finished with generation of {} exclusion volumes.'.format(
len(exclusion_volumes)), logger)
return exclusion_volumes
def generate_features(positions, feature_scores, feature_type,
features_per_feature_type, directory, partner_path,
debugging, total_number_of_features, start,
feature_counter, results):
""" This function generates features with variable tolerance based on a global maximum search algorithm. The
features are described with a list of properties as follows.
Format:
------------------------------------------------------------------------
0 1 2 3 4 5 6 7
------------------------------------------------------------------------
0 hi M [0.0,0.0,0.0] 1.5 [] 0.0 1.0
1 pi M [0.0,0.0,0.0] 1.5 [] 0.0 1.0
2 ni M [0.0,0.0,0.0] 1.5 [] 0.0 1.0
3 hd M [0.0,0.0,0.0] 1.5 [[3.0,0.0,0.0]] 1.9499999 1.0
4 ha M [0.0,0.0,0.0] 1.5 [[3.0,0.0,0.0]] 1.9499999 1.0
5 hd2 M [0.0,0.0,0.0] 1.5 [[3.0,0.0,0.0],[0.0,3.0,0.0]] 1.9499999 1.0
6 ha2 M [0.0,0.0,0.0] 1.5 [[3.0,0.0,0.0],[0.0,3.0,0.0]] 1.9499999 1.0
7 hda M [0.0,0.0,0.0] 1.5 [[3.0,0.0,0.0],[0.0,3.0,0.0]] 1.9499999 1.0
8 ai M [0.0,0.0,0.0] 1.5 [[1.0,0.0,0.0]] 0.43633232 1.0
------------------------------------------------------------------------
Legend:
0 - index
1 - type
2 - flag (O - optional, M - mandatory)
3 - core position
4 - core tolerance
5 - partner positions (hda feature with coordinates for first donor than acceptor)
6 - partner tolerance
7 - weight
"""
logger = setup_logger('_'.join(['features', feature_type]), directory,
debugging)
if partner_path is None:
partner_path = directory + '/data'
if feature_type in grid_list_dict.keys():
partners = pickle_reader(partner_path + '/' + feature_type + '.pkl',
feature_type + '.pkl', logger)
else:
partners = [[]] * len(positions)
score_minimum = 1
tree = cKDTree(positions)
generated_features = []
not_used = range(len(feature_scores))
used = []
while feature_scores[not_used].max() >= score_minimum:
feature_maximum = feature_scores[not_used].max()
logger.debug(
'Feature {} maximum of remaining grid points at {}.'.format(
feature_type, feature_maximum))
indices_not_checked = np.where(
abs(feature_scores - feature_maximum) < 1e-8)[0]
indices = []
# check if grid points within minimum tolerance already used for features
for index_not_checked in indices_not_checked:
feature_indices = tree.query_ball_point(
positions[index_not_checked], r=1.5)
if len(feature_indices) + len(used) == len(
set(feature_indices + used)):
indices.append(index_not_checked)
else:
not_used = [x for x in not_used if x != index_not_checked]
if len(indices) > 0:
# check if only one grid point
if len(indices) == 1:
index = indices[0]
core_tolerance, feature_indices = get_core_tolerance(
positions[index], tree, feature_scores, feature_maximum)
# if more than one grid point, search for the ones with the biggest tolerance
else:
core_tolerance, indices_maximal_tolerance, feature_indices_list = \
get_maximal_core_tolerance(indices, positions, tree, feature_scores, feature_maximum)
# if more than one grid point with biggest tolerance, search for the one with the biggest score
if len(indices_maximal_tolerance) > 1:
index, feature_indices = get_maximal_sum_of_scores(
feature_scores, indices_maximal_tolerance,
feature_indices_list)
else:
index = indices_maximal_tolerance[0]
feature_indices = feature_indices_list[0]
if len(feature_indices) + len(used) > len(
set(feature_indices + used)):
not_used = [x for x in not_used if x != index]
used.append(index)
else:
generated_features.append([
index, feature_type, 'M', positions[index], core_tolerance,
get_partner_positions(feature_type, partners[index]),
get_partner_tolerance(feature_type, core_tolerance), 1.0
])
not_used = [x for x in not_used if x not in feature_indices]
used += feature_indices
with feature_counter.get_lock():
feature_counter.value += 1
update_progress(
feature_counter.value / total_number_of_features,
'Progress of feature generation',
((time.time() - start) / feature_counter.value) *
(total_number_of_features - feature_counter.value))
if len(generated_features) >= features_per_feature_type:
break
if len(generated_features) < features_per_feature_type:
with feature_counter.get_lock():
feature_counter.value += features_per_feature_type - len(
generated_features)
update_progress(feature_counter.value / total_number_of_features,
'Progress of feature generation',
((time.time() - start) / feature_counter.value) *
(total_number_of_features - feature_counter.value))
results += generated_features
return
prog='PyRod',
description='\n'.join(logo),
formatter_class=argparse.RawTextHelpFormatter)
parser.add_argument('conf', help='path to configuration file')
parser.add_argument('--verbose',
dest='debugging',
action='store_true',
help='verbose logging for debugging')
conf = parser.parse_args().conf
debugging = parser.parse_args().debugging
config = configparser.ConfigParser()
config.read(conf)
directory = config.get('directory', 'directory')
if len(directory) == 0:
directory = os.getcwd() + '/pyrod'
logger = setup_logger('main', directory, debugging)
update_user('\n'.join(logo), logger)
logger.debug('\n'.join(
[': '.join(list(_)) for _ in config.items('directory')]))
# defining grid
if config.has_section('test grid parameters'):
logger.debug('\n'.join([
': '.join(list(_)) for _ in config.items('test grid parameters')
]))
center, edge_lengths, name = test_grid_parameters(config)
# determine space resulting in less than 100000 grid points
space = 0.5
space_found = False
while not space_found:
grid = generate_grid(center, edge_lengths, space)
if len(grid) < 100000:
def ensemble_to_centroid(topology, trajectories, output_name, directory,
debugging):
logger = setup_logger('ensembles_to_centroid', directory, debugging)
output_directory = '/'.join([directory, output_name])
protein_topology = '/'.join([output_directory, 'protein.pdb'])
protein_trajectories = []
frames = []
# check if frames in trajectory files, delete empty trajectory files, collect frames in list
for x in range(len(trajectories)):
with open('{}/frames_{}.csv'.format(output_directory, x), 'r') as csv:
frames += csv.readlines()
os.remove('{}/frames_{}.csv'.format(output_directory, x))
try:
mda.Universe(
protein_topology,
'/'.join([output_directory, 'ensemble_{}.dcd'.format(x)]))
protein_trajectories.append(x)
except OSError:
os.remove('/'.join([output_directory,
'ensemble_{}.dcd'.format(x)]))
if len(protein_trajectories) > 0:
# info to user
if len(frames) > 1:
update_user(
'Getting centroid from {} protein conformations.'.format(
len(frames)), logger)
else:
update_user('Found only 1 protein conformations.', logger)
# merge trajectories into one file
protein_trajectories = [
'/'.join([output_directory, 'ensemble_{}.dcd'.format(x)])
for x in protein_trajectories
]
u = mda.Universe(protein_topology, protein_trajectories)
with mda.Writer('/'.join([output_directory, 'ensemble.dcd']),
n_atoms=u.atoms.n_atoms) as DCD:
for _ in u.trajectory:
DCD.write(u.atoms)
# remove sub trajectories
for path in protein_trajectories:
os.remove(path)
# find centroid of frames
if len(frames) > 1:
u = mda.Universe(protein_topology,
'/'.join([output_directory, 'ensemble.dcd']))
conf_dist_matrix = encore.confdistmatrix.get_distance_matrix(
u,
selection='all',
superimpose=True,
n_job=1,
weights='mass',
metadata=False,
verbose=False)
centroid = conf_dist_matrix.as_array().sum(axis=1).argmin()
else:
centroid = 0
# write centroid
u = mda.Universe(topology,
trajectories[int(frames[centroid].split()[0])])
file_path('centroid.pdb', output_directory)
with mda.Writer('/'.join([output_directory, 'centroid.pdb']),
bonds=None,
n_atoms=u.atoms.n_atoms) as PDB:
for _ in u.trajectory[int(frames[centroid].split()[1]
):int(frames[centroid].split()[1]) + 1:]:
PDB.write(u.atoms)
# write csv with frame references
file_path('frames.csv', output_directory)
frames[centroid] = '{}\t{}\t{}\n'.format(frames[centroid].split()[0],
frames[centroid].split()[1],
'centroid')
with open('{}/frames.csv'.format(output_directory), 'w') as csv:
csv.write(''.join(['trajectory\tframe\tcentroid\n'] + frames))
else:
update_user('No protein conformations found.', logger)
sys.exit()
return
def trajectory_analysis(topology, trajectory, grid_score, grid_partners,
frame_counter, total_number_of_frames, first_frame,
last_frame, step_size, metal_names, counter, directory,
debugging, get_partners, trajectory_time, results):
logger = setup_logger('_'.join(['dmif_trajectory',
str(counter)]), directory, debugging)
logger.info('Started analysis of trajectory {}.'.format(counter))
if debugging:
u = mda.Universe(topology, trajectory)
else:
with warnings.catch_warnings():
warnings.simplefilter('ignore')
u = mda.Universe(topology, trajectory)
dtype = [('atomid', int), ('resname', 'U10'), ('resid', int),
('name', 'U10'), ('type', 'U10')]
topology = np.array([(a, b, c, d, e) for a, b, c, d, e in zip(
range(len(u.atoms.resnames)), u.atoms.resnames, u.atoms.resids,
u.atoms.names, u.atoms.types)],
dtype=dtype)
positions = np.array([[
x, y, z
] for x, y, z in zip(grid_score['x'], grid_score['y'], grid_score['z'])])
x_minimum, x_maximum, y_minimum, y_maximum, z_minimum, z_maximum = grid_characteristics(
positions)[:-1]
tree = cKDTree(positions)
main_atoms = main_selection(topology)
hd_atomids, hd_types, hd_hydrogen_atomid_lists = hd_selection(main_atoms)
ha_atomids, ha_types = ha_selection(main_atoms)
hi_atomids = hi_selection(main_atoms)
ni_atomids = ni_selection(main_atoms)
pi_atomids = pi_selection(main_atoms)
ai_atomids = ai_selection(main_atoms)
metal_atomids = metal_selection(topology, metal_names)
for frame, _ in enumerate(u.trajectory[first_frame:last_frame:step_size]):
# create index collectors
shape_inds = []
ha_inds = []
ha2_inds = []
hd_inds = []
hd2_inds = []
hda_inds = []
tw_inds = []
h2o_inds = []
positions = u.atoms.positions
h2os_os_box_inds = topology[((topology['resname'] == 'HOH') &
(topology['name'] == 'O') &
(positions[:, 0] >= x_minimum) &
(positions[:, 0] <= x_maximum) &
(positions[:, 1] >= y_minimum) &
(positions[:, 1] <= y_maximum) &
(positions[:, 2] >= z_minimum) &
(positions[:, 2] <= z_maximum))]['atomid']
if len(h2os_os_box_inds) > 0:
tree_h2os = cKDTree(positions[h2os_os_box_inds])
if len(hd_atomids) > 0:
hd_positions = positions[hd_atomids]
hd_lists = tree_h2os.query_ball_tree(cKDTree(hd_positions),
sel_cutoff_dict['hb'])
else:
hd_positions = []
hd_lists = [[]] * len(h2os_os_box_inds)
if len(ha_atomids) > 0:
ha_positions = positions[ha_atomids]
ha_lists = tree_h2os.query_ball_tree(cKDTree(ha_positions),
sel_cutoff_dict['hb'])
else:
ha_positions = []
ha_lists = [[]] * len(h2os_os_box_inds)
if len(hi_atomids) > 0:
hi_positions = positions[hi_atomids]
hi_lists = tree_h2os.query_ball_tree(cKDTree(hi_positions),
sel_cutoff_dict['hi'])
else:
hi_positions = []
hi_lists = [[]] * len(h2os_os_box_inds)
if len(ni_atomids) > 0:
ni_positions = [((x + y) / 2) for x, y in zip(
positions[ni_atomids[::2]], positions[ni_atomids[1::2]])]
ni_lists = tree_h2os.query_ball_tree(cKDTree(ni_positions),
sel_cutoff_dict['ii'])
else:
ni_positions = []
ni_lists = [[]] * len(h2os_os_box_inds)
if len(pi_atomids) > 0:
pi_positions = [((x + y) / 2) for x, y in zip(
positions[pi_atomids[::2]], positions[pi_atomids[1::2]])]
pi_lists = tree_h2os.query_ball_tree(cKDTree(pi_positions),
sel_cutoff_dict['ii'])
else:
pi_positions = []
pi_lists = [[]] * len(h2os_os_box_inds)
if len(ai_atomids) > 0:
ai_positions = [
((x + y + z) / 3)
for x, y, z in zip(positions[ai_atomids[::3]], positions[
ai_atomids[1::3]], positions[ai_atomids[2::3]])
]
ai_normals = [
normal(a, b, c) for a, b, c in zip(
positions[ai_atomids[::3]], ai_positions, positions[
ai_atomids[2::3]])
]
ai_lists = tree_h2os.query_ball_tree(cKDTree(ai_positions),
sel_cutoff_dict['ai'])
else:
ai_positions = []
ai_normals = []
ai_lists = [[]] * len(h2os_os_box_inds)
if len(metal_atomids) > 0:
metal_positions = positions[metal_atomids]
metal_lists = tree_h2os.query_ball_tree(
cKDTree(metal_positions), sel_cutoff_dict['metal'])
else:
metal_positions = []
metal_lists = [[]] * len(h2os_os_box_inds)
else:
h2os_os_box_inds = []
hd_positions, ha_positions, hi_positions, ni_positions = [], [], [], []
pi_positions, ai_positions, ai_normals, metal_positions = [], [], [], []
hd_lists, ha_lists, hi_lists, ni_lists, pi_lists, ai_lists, metal_lists = [], [], [], [], [], [], []
for o_ind, hd_list, ha_list, hi_list, ni_list, pi_list, ai_list, metal_list in \
zip(h2os_os_box_inds, hd_lists, ha_lists, hi_lists, ni_lists, pi_lists, ai_lists, metal_lists):
ha, ha_i, hd, hd_i, hi, pi, ni, ai, ai_i, ai_n = 0, [], 0, [], 0, 0, 0, 0, [], []
o_coor, h1_coor, h2_coor = positions[o_ind], positions[
o_ind + 1], positions[o_ind + 2]
# hydrogen bond acceptor feature
for hd_ind in hd_list:
hd_coor, hd_type, hd_hydrogen_coors = [
hd_positions[hd_ind], hd_types[hd_ind],
positions[hd_hydrogen_atomid_lists[hd_ind]]
]
if distance(o_coor, hd_coor) <= hb_dist_dict[hd_type]:
for hd_hydrogen_coor in hd_hydrogen_coors:
if angle(o_coor, hd_hydrogen_coor,
hd_coor) >= hb_angl_dict[hd_type]:
ha += 1
ha_i += [float(x) for x in hd_coor]
# hydrogen bond donor feature
for ha_ind in ha_list:
ha_coor, ha_type = ha_positions[ha_ind], ha_types[ha_ind]
if distance(o_coor, ha_coor) <= hb_dist_dict[ha_type]:
for h_coor in [h1_coor, h2_coor]:
if angle(ha_coor, h_coor,
o_coor) >= hb_angl_dict[ha_type]:
hd += 1
hd_i += [float(x) for x in ha_coor]
# metals
for metal_ind in metal_list:
metal_position = metal_positions[metal_ind]
ha += 1
ha_i += [float(x) for x in metal_position]
ni += 2.6 / distance(o_coor, metal_position)
# indices of points close to water
inds = tree.query_ball_point(o_coor, r=1.41)
h2o_inds += inds
# trapped water molecules
if hd + ha > 2:
tw_inds += inds
# water molecule is replaceable/displaceable
else:
# shape
shape_inds += inds
# hydrogen bond features
if hd == 0:
# single
if ha == 1:
ha_inds += inds
if get_partners:
for ind in inds:
grid_partners[ind][
grid_list_dict['ha']] += ha_i
# double
elif ha == 2:
ha2_inds += inds
if get_partners:
for ind in inds:
grid_partners[ind][
grid_list_dict['ha2']] += ha_i
# single hydrogen bond donors
elif hd == 1:
# single donor
if ha == 0:
hd_inds += inds
if get_partners:
for ind in inds:
grid_partners[ind][
grid_list_dict['hd']] += hd_i
# mixed donor acceptor
elif ha == 1:
hda_inds += inds
if get_partners:
for ind in inds:
grid_partners[ind][
grid_list_dict['hda']][0] += hd_i
grid_partners[ind][
grid_list_dict['hda']][1] += ha_i
else:
# double hydrogen bond donor
hd2_inds += inds
if get_partners:
for ind in inds:
grid_partners[ind][grid_list_dict['hd2']] += hd_i
# ionizable interactions and cation-pi interactions
# negative ionizable and cation-pi interactions
for pi_ind in pi_list:
pi_i = pi_positions[pi_ind]
# negative ionizable interaction
ni += 2.6 / distance(o_coor, pi_i)
# cation-pi interaction
for ind in inds:
grid_point = [
grid_score['x'][ind], grid_score['y'][ind],
grid_score['z'][ind]
]
pi_distance = distance(grid_point, pi_i)
if 3.1 <= pi_distance <= 6.0:
grid_score['ai'][
ind] += cation_pi_distance_score_dict[round(
pi_distance, 1)]
if get_partners:
grid_partners[ind][grid_list_dict['ai']] += [
float(x) for x in pi_i
]
# positive ionizable
for ni_ind in ni_list:
pi += 2.6 / distance(o_coor, ni_positions[ni_ind])
# add ionizable interaction score
if pi > 0:
grid_score['pi'][inds] += pi
grid_score['ni'][inds] -= pi
if ni > 0:
grid_score['ni'][inds] += ni
grid_score['pi'][inds] -= ni
# hydrophobic interactions
if len(hi_list) > 0:
hi += 1
if len(hi_list) > 1:
hi += buriedness(o_coor, hi_positions[hi_list])
if hi > 0:
grid_score['hi_norm'][inds] += hi
# no charged environment
if ni < 0.65 > pi:
grid_score['hi'][inds] += hi
# aromatic interactions grid point wise
for ai_ind in ai_list:
ai_i = ai_positions[ai_ind]
ai_n = ai_normals[ai_ind]
for ind in inds:
grid_point = [
grid_score['x'][ind], grid_score['y'][ind],
grid_score['z'][ind]
]
ai_distance = distance(grid_point, ai_i)
if 3.1 <= ai_distance <= 6.0:
ai_vector = vector(ai_i, grid_point)
ai_n, alpha = ai_geometry(ai_vector, ai_n)
# cation-pi interactions
if alpha <= CATION_PI_ANGLE_CUTOFF:
grid_score['pi'][
ind] += cation_pi_distance_score_dict[
round(ai_distance, 1)]
# pi- and t-stacking
if ai_distance >= 3.3:
# pi- and t-stacking with pi-system of protein aromatic center
if alpha < 45:
offset = opposite(alpha, ai_distance)
# pi-stacking
if ai_distance <= 4.7:
# check offset between grid point and aromatic center
if offset <= 2.0:
grid_score['ai'][
ind] += pi_stacking_distance_score_dict[
round(ai_distance, 1)]
if get_partners:
grid_partners[ind][grid_list_dict['ai']] += \
pi_stacking_partner_position(grid_point, ai_n, ai_distance, alpha)
# t-stacking
else:
# check offset between grid point and aromatic center
if offset <= 0.5:
grid_score['ai'][
ind] += t_stacking_distance_score_dict[
round(ai_distance, 1)]
if get_partners:
grid_partners[ind][grid_list_dict['ai']] += \
t_stacking_partner_position(ai_i, grid_point, ai_n, offset,
ai_distance, alpha, True)
# t-stacking with hydrogen of protein aromatic center
else:
if ai_distance >= 4.6:
# check offset between grid point and aromatic center
offset = adjacent(alpha, ai_distance)
if offset <= 0.5:
grid_score['ai'][
ind] += t_stacking_distance_score_dict[
round(ai_distance, 1)]
if get_partners:
ai_n2 = cross_product(
ai_n,
cross_product(
ai_n, ai_vector))
ai_n2, alpha = ai_geometry(
ai_vector, ai_n2)
grid_partners[ind][grid_list_dict['ai']] += \
t_stacking_partner_position(ai_i, grid_point, ai_n2, offset,
ai_distance, alpha)
# adding scores to grid
grid_score['shape'][shape_inds] += 1
grid_score['ha'][ha_inds] += 1
grid_score['ha2'][ha2_inds] += 1
grid_score['hd'][hd_inds] += 1
grid_score['hd2'][hd2_inds] += 1
grid_score['hda'][hda_inds] += 1
grid_score['tw'][tw_inds] += 1
grid_score['h2o'][h2o_inds] += 1
# grid partners to numpy array
with frame_counter.get_lock():
frame_counter.value += 1
update_progress(
frame_counter.value / total_number_of_frames,
'Progress of trajectory analysis',
((time.time() - trajectory_time) / frame_counter.value) *
(total_number_of_frames - frame_counter.value))
logger.debug('Trajectory {} finished with frame {}.'.format(
counter, frame))
logger.info('Finished analysis of trajectory {}.'.format(counter))
grid_partners = grid_partners_to_array(grid_partners)
results.append([grid_score, grid_partners])
return
def screen_protein_conformations(topology, trajectory, pharmacophore_path,
ligand_path, counter, first_frame, last_frame,
step_size, metal_names, directory,
output_name, debugging,
total_number_of_frames, frame_counter,
trajectory_time):
dcd_name = 'ensemble_' + str(counter) + '.dcd'
output_directory = '/'.join([directory, output_name])
file_path(dcd_name, output_directory)
logger = setup_logger(
'_'.join(['screen_protein_conformations',
str(counter)]), directory, debugging)
logger.info(
'Started screening of protein conformations in trajectory {}.'.format(
counter))
ligand_positions = None
if debugging:
u = mda.Universe(topology, trajectory)
if ligand_path:
ligand_positions = mda.Universe(ligand_path).atoms.positions
else:
with warnings.catch_warnings():
warnings.simplefilter('ignore')
u = mda.Universe(topology, trajectory)
if ligand_path:
ligand_positions = mda.Universe(ligand_path).atoms.positions
protein = u.select_atoms('protein')
dtype = [('atomid', int), ('resname', 'U10'), ('resid', int),
('name', 'U10'), ('type', 'U10')]
topology = np.array([(a, b, c, d, e) for a, b, c, d, e in zip(
range(len(u.atoms.resnames)), u.atoms.resnames, u.atoms.resids,
u.atoms.names, u.atoms.types)],
dtype=dtype)
main_atoms = main_selection(topology)
main_atomids = main_atoms['atomid']
heavy_atomids = heavy_atom_selection(main_atoms)
hd_atomids, hd_types, hydrogen_atomid_lists = hd_selection(main_atoms)
ha_atomids = ha_selection(main_atoms)[0]
hi_atomids = hi_selection(main_atoms)
ni_atomids = ni_selection(main_atoms)
pi_atomids = pi_selection(main_atoms)
ai_atomids = ai_selection(main_atoms)
metal_atomids = metal_selection(topology, metal_names)
features = [
feature
for feature in pharmacophore_reader(pharmacophore_path, False, logger)
if feature[1] != 'ev'
]
if counter == 0:
file_path('protein.pdb', output_directory)
if debugging:
with mda.Writer('/'.join([output_directory, 'protein.pdb']),
bonds=None,
n_atoms=protein.n_atoms) as PDB:
PDB.write(protein)
else:
with warnings.catch_warnings():
warnings.simplefilter('ignore')
with mda.Writer('/'.join([output_directory, 'protein.pdb']), bonds=None, n_atoms=protein.n_atoms) as \
PDB:
PDB.write(protein)
frame_collector = []
with mda.Writer('/'.join([output_directory, dcd_name]),
n_atoms=protein.n_atoms) as DCD:
for frame, _ in enumerate(
u.trajectory[first_frame:last_frame:step_size]):
positions = u.atoms.positions
matched_features = 0
for feature in features:
ai, ha, hd, pi, ni, cation_pi, hi = 0, 0, 0, 0, 0, 0, 0
feature_type = feature[1]
feature_position = np.array(feature[3])
partner_position = []
if feature_type in ['ha', 'hd', 'ha2', 'hd2', 'ai']:
partner_position = np.array(feature[5][0])
partner_tolerance = feature[6]
feature_score = feature[7]
# hydrogen bonds and metal interaction
if feature_type == 'hd':
if len(ha_atomids) > 0:
ha_positions = positions[ha_atomids]
ha += np.sum((cdist(partner_position.reshape(
1, 3), ha_positions) <= partner_tolerance)[0])
if ha == 0:
break
else:
matched_features += 1
elif feature_type == 'ha':
if len(hd_atomids) > 0:
hd_positions = positions[hd_atomids]
hd_bools = (cdist(partner_position.reshape(
1, 3), hd_positions) <= partner_tolerance)[0]
matched_hd_positions = hd_positions[hd_bools]
matched_hd_types = hd_types[hd_bools]
matched_hydrogen_atomid_lists = hydrogen_atomid_lists[
hd_bools]
for matched_hd_position, matched_hd_type, matched_hydrogen_atomid_list in \
zip(matched_hd_positions, matched_hd_types, matched_hydrogen_atomid_lists):
for matched_hydrogen_atomid in matched_hydrogen_atomid_list:
if angle(feature_position, positions[matched_hydrogen_atomid], matched_hd_position) >= \
hb_angl_dict[matched_hd_type]:
hd = 1
if len(metal_atomids) > 0:
metal_positions = positions[metal_atomids]
hd += np.sum((cdist(partner_position.reshape(
1, 3), metal_positions) <= partner_tolerance)[0])
if hd == 0:
break
else:
matched_features += 1
elif feature_type in ['hi', 'pi', 'ni']:
if len(ni_atomids) > 0:
ni_positions = np.array([
((x + y) / 2)
for x, y in zip(positions[ni_atomids[::2]],
positions[ni_atomids[1::2]])
])
ni_positions = ni_positions[(cdist(
feature_position.reshape(1, 3), ni_positions) <=
sel_cutoff_dict['ii'])[0]]
for ni_position in ni_positions:
pi += 2.6 / distance(feature_position, ni_position)
if len(pi_atomids) > 0:
pi_positions = np.array([
((x + y) / 2)
for x, y in zip(positions[pi_atomids[::2]],
positions[pi_atomids[1::2]])
])
pi_positions = pi_positions[(cdist(
feature_position.reshape(1, 3), pi_positions) <=
sel_cutoff_dict['ii'])[0]]
for pi_position in pi_positions:
ni += 2.6 / distance(feature_position, pi_position)
if len(metal_atomids) > 0:
metal_positions = positions[metal_atomids]
metal_booleans = (cdist(feature_position.reshape(1, 3),
metal_positions) <=
sel_cutoff_dict['metal'])[0]
metal_positions = metal_positions[metal_booleans]
for metal_position in metal_positions:
ni += 2.6 / distance(feature_position,
metal_position)
if feature_type == 'hi':
if len(hi_atomids) > 0:
# no charged protein environment
if ni >= 0.65 <= pi:
break
hi_positions = positions[hi_atomids]
hi_positions = hi_positions[(cdist(
feature_position.reshape(1, 3),
hi_positions) <= sel_cutoff_dict['hi'])[0]]
if len(hi_positions) > 0:
hi += 1
if len(hi_positions) > 1:
hi += buriedness(feature_position,
hi_positions)
# check if pi and ni > 0.65
if hi < feature_score:
break
else:
matched_features += 1
elif feature_type == 'pi':
if len(ai_atomids) > 0:
# cation-pi interactions
ai_positions = np.array([
((x + y + z) / 3)
for x, y, z in zip(positions[ai_atomids[::3]],
positions[ai_atomids[1::3]],
positions[ai_atomids[2::3]])
])
ai_normals = np.array([
normal(a, b, c) for a, b, c in zip(
positions[ai_atomids[::3]], ai_positions,
positions[ai_atomids[2::3]])
])
ai_booleans = (cdist(
feature_position.reshape(1, 3), ai_positions)
<= sel_cutoff_dict['ai'])[0]
ai_positions = ai_positions[ai_booleans]
ai_normals = ai_normals[ai_booleans]
for ai_i, ai_n in zip(ai_positions, ai_normals):
ai_distance = distance(ai_i, feature_position)
if 3.1 <= ai_distance <= 6.0:
ai_n, alpha = ai_geometry(
vector(ai_i, feature_position), ai_n)
if alpha <= CATION_PI_ANGLE_CUTOFF:
cation_pi += cation_pi_distance_score_dict[
round(ai_distance, 1)]
if pi + cation_pi - ni < feature_score:
break
else:
matched_features += 1
elif feature_type == 'ni':
if ni - pi < feature_score:
break
else:
matched_features += 1
elif feature_type == 'ai':
if len(pi_atomids) > 0:
# cation-pi interaction
pi_positions = np.array([
((x + y) / 2)
for x, y in zip(positions[pi_atomids[::2]],
positions[pi_atomids[1::2]])
])
pi_positions = pi_positions[(cdist(
feature_position.reshape(1, 3), pi_positions) <=
sel_cutoff_dict['ii'])[0]]
for pi_position in pi_positions:
pi_distance = distance(pi_position,
feature_position)
if 3.1 <= pi_distance <= 6.0:
alpha = ai_geometry(
vector(pi_position, feature_position),
partner_position)[1]
if alpha <= CATION_PI_ANGLE_CUTOFF:
ai += cation_pi_distance_score_dict[round(
pi_distance, 1)]
if len(ai_atomids) > 0:
# aromatic interactions
ai_positions = np.array([
((x + y + z) / 3)
for x, y, z in zip(positions[
ai_atomids[::3]], positions[ai_atomids[1::3]],
positions[ai_atomids[2::3]])
])
ai_normals = np.array([
normal(a, b, c) for a, b, c in zip(
positions[ai_atomids[::3]], ai_positions,
positions[ai_atomids[2::3]])
])
ai_booleans = (cdist(feature_position.reshape(
1, 3), ai_positions) <= sel_cutoff_dict['ai'])[0]
ai_positions = ai_positions[ai_booleans]
ai_normals = ai_normals[ai_booleans]
for ai_i, ai_n in zip(ai_positions, ai_normals):
ai_distance = distance(ai_i, feature_position)
if 3.3 <= ai_distance <= 6.0:
ai_vector = vector(ai_i, feature_position)
ai_n, alpha = ai_geometry(ai_vector, ai_n)
angle_tolerance = math.degrees(
partner_tolerance)
# pi- and t-stacking with pi-system of protein aromatic center
if alpha < 45:
offset = opposite(alpha, ai_distance)
# pi-stacking
if ai_distance <= 4.7:
# check offset between grid point and aromatic center
if offset <= 2.0:
# check angle between normals
if vector_angle(
ai_n, partner_position
) <= angle_tolerance:
ai += pi_stacking_distance_score_dict[
round(ai_distance, 1)]
# t-stacking
else:
# check offset between grid point and aromatic center
if offset <= 0.5:
# check angle between normals
if (90 - angle_tolerance <=
vector_angle(
ai_n, partner_position)
>= 90 + angle_tolerance):
ai += t_stacking_distance_score_dict[
round(ai_distance, 1)]
# t-stacking with hydrogen of protein aromatic center
else:
if ai_distance >= 4.6:
offset = adjacent(alpha, ai_distance)
# check offset between grid point and aromatic center
if offset <= 0.5:
if (90 - angle_tolerance <=
vector_angle(
ai_n, partner_position)
>= 90 + angle_tolerance):
ai += t_stacking_distance_score_dict[
round(ai_distance, 1)]
if ai < feature_score:
break
else:
matched_features += 1
if matched_features == len(features):
clash = False
heavy_atom_positions = positions[heavy_atomids]
for feature in features:
core_tolerance = feature[4]
if cdist(np.array([feature[3]]),
heavy_atom_positions).min() < core_tolerance:
clash = True
if not clash:
if ligand_path:
main_positions = positions[main_atomids]
if cdist(main_positions,
ligand_positions).min() < CLASH_CUTOFF:
clash = True
if not clash:
DCD.write(protein)
frame_collector.append(frame + first_frame)
logger.debug('Trajectory {} finished with frame {}.'.format(
counter, frame))
with frame_counter.get_lock():
frame_counter.value += 1
update_progress(
frame_counter.value / total_number_of_frames,
'Progress of trajectory analysis',
((time.time() - trajectory_time) / frame_counter.value) *
(total_number_of_frames - frame_counter.value))
logger.info('Finished screening of trajectory {}.'.format(counter))
with open('{}/frames_{}.csv'.format(output_directory, counter),
'w') as csv:
for frame in frame_collector:
csv.write('{}\t{}\n'.format(counter, frame))
return
