def extract_st(cms_file, trj_dir, asl='all', frames=1):
"""
Extract N frames from the trajectory and save them in structure.Structure
:param cms_file:
:param trj_dir:
:param asl:
:param frames: if int: number of frames uniformly distributed, if iterable: list fo frames
:return structures: list of schrodinger.structure.Structure
:type structures: list
"""
structures = []
msys_model, cms_model = topo.read_cms(cms_file)
frame_arch = traj.read_traj(trj_dir)
if type(frames) == int:
for f in np.linspace(0, len(frame_arch) - 1, frames).astype(int):
st = topo.update_cms(cms_model, frame_arch[f])
st = st.extract(evaluate_asl(cms_model, asl))
st.title = 'Frame {}'.format(f)
structures.append(st)
else:
try:
for f in frames:
st = topo.update_cms(cms_model, frame_arch[f])
st = st.extract(asl)
st.title = 'Frame {}'.format(f)
structures.append(st)
except Exception as e:
raise RuntimeError(e)
return structures
def __init__(self, cms_file, trj_dir, queue, frames=None, params=None):
"""
Docstring
:param cms_file:
:param trj_dir:
:param frames:
:param params:
"""
multiprocessing.Process.__init__(self)
self.queue = queue
# load cms_model
self.msys_model, self.cms_model = topo.read_cms(str(cms_file))
# Get framelist
if frames is not None:
self.frame_list = [
frame for (i, frame) in enumerate(traj.read_traj(str(trj_dir)))
if i in frames
]
else:
self.frame_list = traj.read_traj(str(trj_dir))
self.total_frames = len(self.frame_list)
self.align = Superimposer()
# Calculation parameters
self.calculation_parameters = []
if params is not None:
for param in params:
self.add_param(**param)
def get_desmond_cms(api, stid, dir=None):
"""
:param api:
:param stid:
:param dir:
:return:
"""
if dir is None:
tempdir = tempfile.TemporaryDirectory()
atexit.register(tempdir.cleanup)
dir = tempdir.name
filename = os.path.join(dir, 'desmond.cms')
try:
resp = api.get_structure_file(stid, file_type='desmond_cms')
resp.raise_for_status()
except Exception as e:
resp.close()
raise HTTPError(e)
with open(filename, 'wb') as fh:
fh.write(resp.content)
resp.close()
return topo.read_cms(filename)
def preprocess_hbond(api, dataset, data_type=None, no_ligand=False):
"""
:param api:
:param dataset:
:param data_type: Placeholder
:param no_ligand: do not rename ligand residues
:return:
"""
hbonds_df = pd.DataFrame()
if api is None:
dataset.dropna(axis=0,
subset=['desmond_cms', 'trj_hbonds'],
inplace=True)
for index, stid in zip(dataset.index, dataset.loc[:, 'structure_id']):
# Get data
if api is None:
msys_model, cms_model = topo.read_cms(dataset.loc[index,
'desmond_cms'])
data = pd.read_csv(dataset.loc[index, 'trj_hbonds'],
sep=',',
index_col=0)
data.index = np.arange(data.shape[0])
else:
msys_model, cms_model = get_desmond_cms(
api,
stid,
)
data = get_trj_hbonds(api, stid)
# Get ligand
if no_ligand:
ligand_resid = None
elif 'ligand_resnum' not in dataset.columns and 'ligand_chain' not in dataset.columns:
ligands = find_ligands(st=cms_model)
assert (len(ligands) == 1)
ligand_resid = None
for atm in ligands[0].st.atom:
if ligand_resid is None:
ligand_resid = (atm.resnum, atm.chain.strip())
else:
if (atm.resnum, atm.chain.strip()) != ligand_resid:
raise ValueError(
'Structure {}\nFound multiple ligand residues!'.
format(stid))
elif 'ligand_resnum' in dataset.columns and 'ligand_chain' in dataset.columns:
ligand_resid = (int(dataset.loc[index, 'ligand_resnum']),
dataset.loc[index, 'ligand_chain'])
frequency_df = _load_hbonds(cms_model, data, ligand_resid=ligand_resid)
for fi in frequency_df.index:
hbonds_df.loc[stid, fi] = frequency_df.loc[fi, 'frequency']
hbonds_df.fillna(0, inplace=True)
return hbonds_df
def preprocess_rms(api, dataset, data_type=None, no_ligand=False):
"""
:param api:
:type api: pldbclient.api_client.Api
:param dataset:
:type dataset: pandas.Dataframe
:param data_type: Placeholder
:param no_lignad: Placeholder
:return:
"""
if api is None:
dataset.dropna(axis=0, subset=['desmond_cms', 'trj_rms'], inplace=True)
df_rms = pd.DataFrame(index=dataset.structure_id)
for index, stid in zip(dataset.index, dataset.structure_id):
with tempfile.TemporaryDirectory() as tempdir:
if api is None:
msys_model, cms_model = topo.read_cms(
dataset.loc[index, 'desmond_cms'])
with open(dataset.loc[index, 'trj_rms'], 'r') as fh:
trj_rms = json.load(fh)
else:
# Load desomd cms
msys_model, cms_model = get_desmond_cms(api, stid, dir=tempdir)
# Load trj_rms
trj_rms = get_trj_rms(api, stid, dir=tempdir)
# Preprocess data
for data in trj_rms:
if data['name'] == 'calpha rmsd':
df_rms.loc[stid, 'mean_rmsd'] = np.mean(data['results'])
if data['name'] == 'calpha rmsf':
for aid, x in zip(data['atom_ids'], data['results']):
df_rms.loc[stid, '{}:{}'.format(*aid[:-1])] = x
if data['name'] == 'ligand_rmsf':
aid2an = {}
for n in evaluate_asl(cms_model, 'ligand and not a.element H'):
aid2an['{}:{}'.format(
str(cms_model.atom[n].resnum),
cms_model.atom[n].pdbname.strip())] = n
for aid, x in zip(data['atom_ids'], data['results']):
an = aid2an.get('{}:{}'.format(*aid[1:]))
if an is None: # Legacy check, this is true if calculation included pseudoatoms
continue
label = '#{}'.format(an)
df_rms.loc[stid, label] = x
return df_rms
def preprocess_torsion(api, dataset, data_type=None, no_ligand=False):
"""
:param api:
:param dataset:
:param data_type: Placeholder
:param no_ligand: Placeholder
:return:
"""
if api is None: # Only process structures which contain all input files
dataset.dropna(axis=0,
subset=['desmond_cms', 'trj_torsion'],
inplace=True)
torsion_entropy = pd.DataFrame(index=dataset.structure_id)
bins = np.arange(-180, 198, 18) # Bin size for calculating entropy
cwd = os.path.abspath('./')
for index, stid in zip(dataset.index, dataset.structure_id):
with tempfile.TemporaryDirectory() as tempdir:
os.chdir(tempdir)
if api is None:
msys_model, cms_model = topo.read_cms(
dataset.loc[index, 'desmond_cms'])
trj_torsion = dataset.loc[index, 'trj_torsion']
else:
# Load cms_model
msys_model, cms_model = get_desmond_cms(api, stid, dir=tempdir)
trj_torsion = get_trj_torsion(api, stid, dir=tempdir)
with tarfile.open(trj_torsion, 'r:gz') as tar:
tar.extractall(path=tempdir)
torsion_ids = pd.read_csv('torsion_ids.csv', sep=',', index_col=0)
for tid in torsion_ids.index:
torsion_label = label_torsion(
cms_model, *list(map(int, torsion_ids.loc[tid, :])))
series = np.genfromtxt('torsion_{}.csv'.format(int(tid)),
delimiter=',')
counts, bins = np.histogram(series, bins=bins)
freq = counts / np.sum(counts)
torsion_entropy.loc[stid, torsion_label] = entropy(freq)
os.chdir(cwd)
return torsion_entropy
def from_files(cls, cms_fname, traj_fname):
from schrodinger.application.desmond.packages import topo, traj
_, model = topo.read_cms(cms_fname)
traj = traj.read_traj(traj_fname)
return cls(model, traj)
def merge_torsion(data, dataset=None, api=None):
"""
:param data:
:type data: pd.DataFrame
:param dataset:
:type dataset: pd.DataFrame
:param api:
:type api: pldbclient.api_client.Api
:return:
"""
if len(data) == 1:
df_raw = data[0] # type: pd.DataFrame
else:
df_raw = pd.concat(data, sort=False) # type: pd.DataFrame
ligand_col = [c for c in df_raw.columns if c[0] == '#'
] # ligand_torsion columns are indicated by '#<a1:a2:a3:a4>'
# Get structures
cms_models = []
for stid in df_raw.index:
with tempfile.TemporaryDirectory() as tempdir:
try:
resp = api.get_structure_file(structure_id=stid,
file_type='desmond_cms')
resp.raise_for_status()
except HTTPError as e:
resp.close()
logger.error(e)
with open(os.path.join(tempdir, 'desmond.cms'), 'wb') as fh:
fh.write(resp.content)
resp.close()
msys_model, cms_model = topo.read_cms(
os.path.join(tempdir, 'desmond.cms'))
cms_models.append(cms_model)
common_atoms, ligand_st = get_min_common_substructure(
cms_models, return_st=True, return_common_atoms=True)
if not all(common_atoms
): # translate: if common atoms is a list of empty lists
logger.warning('Ligands do not share a common substructure')
df_raw.drop(ligand_col, axis=1, inplace=True)
return df_raw, None
else:
df_merged = df_raw.drop(ligand_col, axis=1) # type: pd.DataFrame
# Assign common atom ids
cid_map = dict([(stid, {}) for stid in df_merged.index])
for i, atom_array in enumerate(zip(*common_atoms)):
for stid, aid in zip(df_merged.index, atom_array):
cid_map[stid][aid] = i + 1 # Because atom indices start at 1
for stid in df_merged.index:
for torsion_atoms in ligand_col:
if not pd.isna(df_raw.loc[stid, torsion_atoms]):
x = df_raw.loc[stid, torsion_atoms]
torsion_atoms = list(map(int, torsion_atoms[1:].split(
':'))) # torsion ids == '#a1:a2:a3:a4'
if all([aid in cid_map[stid] for aid in torsion_atoms
]): # If aid has a common atom id
torsion_bond = sorted([
cid_map[stid][torsion_atoms[1]],
cid_map[stid][torsion_atoms[2]]
])
label = 'LIG:{}:{}'.format(*torsion_bond)
df_merged.loc[stid, label] = x
return df_merged, ligand_st
def merge_rms(data, dataset=None, api=None):
"""
Merge the trj_rms data.
Merger can also be called only one data file.
:param data:
:type data: pd.DataFrame
:param dataset:
:type dataset: pd.DataFrame
:param api:
:type api: pldbclient.api_client.Api
:return:
"""
if len(data) == 1:
df_raw = data[0]
else:
df_raw = pd.concat(data, sort=False)
if api is None:
dataset.dropna(axis=0, subset=['desmond_cms', 'trj_rms'], inplace=True)
ligand_ids = dataset.loc[:, ['ligand_resnum', 'ligand_chain']]
else:
ligand_ids = None
ligand_col = [c for c in df_raw.columns if c[0] == '#'
] # ligand_rmsf columns are indicated by '#<atom_id>'
# Get structures
cms_models = []
for index in df_raw.index:
if api is None:
msys_model, cms_model = topo.read_cms(dataset.loc[index,
'desmond_cms'])
else:
msys_model, cms_model = get_desmond_cms(api, index)
cms_models.append(cms_model)
# Get a list (of lists) of common atoms
common_atoms, ligand_st = get_min_common_substructure(
cms_models,
ligand_ids=ligand_ids,
return_st=True,
return_common_atoms=True)
if not all(common_atoms
): # translate: if common atoms is a list of empty lists
logger.warning('Ligands do not share a common substructure')
df_raw.drop(ligand_col, axis=1, inplace=True)
return df_raw, None
else:
df_merged = df_raw.drop(ligand_col, axis=1)
# Assign common atom ids
cid_map = dict([(stid, {}) for stid in df_merged.index])
for i, atom_array in enumerate(zip(*common_atoms)):
for stid, aid in zip(df_merged.index, atom_array):
cid_map[stid][aid] = i + 1 # Because atom indices start at 1
for stid in df_merged.index:
for aid in ligand_col:
if not pd.isna(df_raw.loc[stid, aid]):
x = df_raw.loc[stid, aid]
aid = int(aid[1:]) # Drop the preceeding '#'
if aid in cid_map[stid]: # If aid has a common atom id
label = 'LIG:{}'.format(cid_map[stid][aid])
df_merged.loc[stid, label] = x
return df_merged, ligand_st
def _process(structure_dict):
"""
Docstring
:param structure_dict:
:return:
"""
fork = None
# Check if transformers is called as part of a pipeline
if 'pipeline' in structure_dict['custom']:
pipeline = structure_dict['custom']['pipeline']
fork = [
pipeline[0],
]
if len(pipeline) == 1:
del (structure_dict['custom']['pipeline'])
else:
structure_dict['custom']['pipeline'] = pipeline[1:]
outname = '{}_trj_rms'.format(structure_dict['structure']['code'])
outfile = outname + '.json'
results = []
# Load simulation files
cms_file = structure_dict['files']['desmond_cms']
trjtar = structure_dict['files']['desmond_trjtar']
# If run from command line it does not make sense to provide a tarfile
if os.path.isdir(trjtar):
trj_dir = trjtar
elif tarfile.is_tarfile(trjtar):
with tarfile.open(name=trjtar, mode='r:gz') as tfile:
tfile.extractall()
logger.info('extracting frameset')
trj_dir = tfile.getnames()[0]
else:
raise RuntimeError('trjtar is neither a directory nor a tarfile')
# Check whether a ligand exists and add ligand rmsd to set of calculations
calculation_param = copy.copy(CALCULATION_PARAM)
ligand_mae = structure_dict['files'].get('ligand_mae')
if ligand_mae is None:
msys_model, cms_model = topo.read_cms(str(cms_file))
ligands = find_ligands(cms_model)
if len(ligands) != 0:
for ligand in ligands:
ligand_asl = ' or '.join([
'(r. {} and c. "{}" and not a.element H )'.format(
res.resnum, res.chain) for res in ligand.st.residue
])
calculation_param.append({
'name': 'ligand_rmsf',
'ref': None,
'align_asl': ligand_asl,
'calc_asl': None,
'calculation_type': 'rmsf'
})
else:
ligand_st = structure.Structure.read(str(ligand_mae))
ligand_asl = '( ' + ' or '.join([
'(r.ptype {} and c. "{}")'.format(res.pdbres, res.chain)
for res in ligand_st.residue
]) + ' ) and not a.element H'
calculation_param.append({
'name': 'ligand_rmsf',
'ref': None,
'align_asl': ligand_asl,
'calc_asl': None,
'calculation_type': 'rmsf'
})
out_queue = multiprocessing.Queue()
if NPROC != len(calculation_param):
nproc = len(calculation_param)
else:
nproc = NPROC
logger.info('Performing {} rms calculations, using {} cores.'.format(
len(calculation_param), nproc))
calculation_param = np.array_split(calculation_param, nproc)
workers = []
for i, params in enumerate(calculation_param):
workers.append(RMS(cms_file, trj_dir, out_queue, params=params))
logger.info('starting subprocess {}'.format(i))
workers[-1].start()
for i in range(nproc):
results.append(out_queue.get())
for w in workers:
w.join()
with open(outfile, 'w') as f:
json.dump(results, f)
transformer_dict = {
'structure': {
'parent_structure_id': structure_dict['structure']['structure_id']
},
'files': {
'trj_rms': outfile
},
'custom': structure_dict['custom']
}
if fork is not None:
logger.info('Forking pipeline: ' + ' '.join(fork))
transformer_dict['control'] = {'forks': fork}
yield transformer_dict
def _process(structure_dict):
"""
:param structure_dict:
:return:
"""
t = time.time()
fork = None
# Check if transformers is called as part of a pipeline
if 'pipeline' in structure_dict['custom']:
pipeline = structure_dict['custom']['pipeline']
fork = [pipeline[0], ]
if len(pipeline) == 1:
del (structure_dict['custom']['pipeline'])
else:
structure_dict['custom']['pipeline'] = pipeline[1:]
structure_code = structure_dict['structure']['code']
logger.info('Load simulation files')
cms_file = structure_dict['files']['desmond_cms']
msys_model, cms_model = topo.read_cms(str(cms_file))
trjtar = structure_dict['files']['desmond_trjtar']
# If run from command line it does not make sense to provide a tarfile
if os.path.isdir(trjtar):
trj_dir = trjtar
elif tarfile.is_tarfile(trjtar):
with tarfile.open(name=trjtar, mode='r:gz') as tfile:
tfile.extractall()
trj_dir = tfile.getnames()[0]
else:
raise RuntimeError('trjtar is neither a directory nor a tarfile')
frame_list = traj.read_traj(str(trj_dir))
frame_list = [frame_list[i] for i in range(0, len(frame_list), STEP)]
logger.info('Calculating torsion angles')
cms_model = set_original_atom_index(cms_model)
ligand_ct = cms_model.extract(evaluate_asl(cms_model, 'ligand'), copy_props=True)
torsion_list = get_hetero_torsion_atoms(ligand_ct, element_priority=ELEMENT_PRIORITY)
torsion_list.extend(get_protein_torsion_atoms(cms_model))
analyzers = []
torsion_ids = pd.DataFrame(columns=['index', 'aid1', 'aid2', 'aid3', 'aid4'])
torsion_ids.set_index('index', inplace=True)
for i, atom_set in enumerate(torsion_list):
atom_set = list(map(get_original_atom_index, atom_set))
analyzers.append(Torsion(msys_model, cms_model, *atom_set))
torsion_ids.loc[i, ['aid1', 'aid2', 'aid3', 'aid4']] = atom_set
results = analyze(frame_list, *analyzers,
**{"progress_feedback": functools.partial(print_iframe, logger=logger)})
out_arch = '{}_torsion.tar.gz'.format(structure_code)
with tarfile.open(out_arch, 'w:gz') as tar:
torsion_ids.to_csv('torsion_ids.csv', sep=',')
tar.add('torsion_ids.csv')
for i, timeseries in enumerate(results):
fname = 'torsion_{}.csv'.format(i)
np.savetxt(fname, timeseries, delimiter=',')
tar.add(fname)
logger.info('Calculated torsion angles in {:.0f} seconds'.format(time.time() - t))
# Return structure dict
transformer_dict = {
'structure': {
'parent_structure_id':
structure_dict['structure']['structure_id'],
'searchable': False
},
'files': {'trj_torsion': out_arch},
'custom': structure_dict['custom']
}
if fork is not None:
logger.info('Forking pipeline: ' + ' '.join(fork))
transformer_dict['control'] = {'forks': fork}
yield transformer_dict
def _process(structure_dict):
"""
DocString
:param structure_dict:
:return:
"""
fork = None
# Check if transformers is called as part of a pipeline
if 'pipeline' in structure_dict['custom']:
pipeline = structure_dict['custom']['pipeline']
fork = [
pipeline[0],
]
if len(pipeline) == 1:
del (structure_dict['custom']['pipeline'])
else:
structure_dict['custom']['pipeline'] = pipeline[1:]
structure_code = structure_dict['structure']['code']
outname = '{}_trj_nonbonded'.format(structure_code)
outfile = outname + '.json'
outfile_raw = outname + '.tar.bz2'
# desmond_cms file
cmsfile = structure_dict['files']['desmond_cms']
msys_model, cms_model = topo.read_cms(str(cmsfile))
# desmond_cfg file
cfgfile = structure_dict['files']['desmond_cfg']
# desmond frame archive
trjtar = structure_dict['files']['desmond_trjtar']
# If run from command line it does not make sense to provide a tarfile
if os.path.isdir(trjtar):
trj_dir = trjtar
elif tarfile.is_tarfile(trjtar):
with tarfile.open(name=trjtar, mode='r:gz') as tfile:
tfile.extractall()
logger.info('extracting frameset')
trj_dir = tfile.getnames()[0]
else:
raise RuntimeError('trjtar is neither a directory nor a tarfile')
logger.info('creating atomgroups')
_id, atom_groups, nonbonded_dict, structure_dict, fork = assign_atomgroups(
structure_dict, cms_model, fork)
logger.info('Finding free host')
logger.debug('Hosts: ' + ', '.join(HOSTS))
host = get_gpu_host(HOSTS)
logger.info('Running desmond job on: {}'.format(host))
vrun_obj = VRUN(cms_model, trj_dir, cfgfile, atom_groups)
energy_group_file = vrun_obj.calculate_energy(outname, host=host)
if RAW:
# Write raw output
with open('{}_atom_groups.json'.format(_id), 'w') as f:
json.dump(atom_groups, f)
nonbonded_raw = structure_dict['files'].get('desmond_nonbonded_raw')
if nonbonded_raw is None:
with tarfile.open(outfile_raw, 'w:bz2') as tar:
for fn in [
energy_group_file, '{}_atom_groups.json'.format(_id)
]:
tar.add(fn)
else:
with tarfile.open(nonbonded_raw, 'r:bz2') as tar:
tar.extractall()
members = tar.getmembers()
with tarfile.open(outfile_raw, 'w:bz2') as tar:
for member in members:
filename = member.name
if filename.split('_')[0] not in (
'custom', 'default'): # Backward Compatibility
logger.warning('Updating file names to new version')
os.rename(filename, 'default_' + filename)
filename = 'default_' + filename
tar.add(filename)
for fn in [
energy_group_file, '{}_atom_groups.json'.format(_id)
]:
tar.add(fn)
# Get time and energy components
sim_time, component_dict = parse_output(energy_group_file,
len(atom_groups),
ENERGY_COMPONENTS)
# calculate mean and error
results = {}
logger.info('Calculating average potential')
for comp in ENERGY_COMPONENTS:
pair_dict = component_dict[comp]
data_dict = {}
results[comp] = {}
results[comp]['keys'] = []
results[comp]['mean_potential'] = []
results[comp]['error'] = []
for pair, energy in pair_dict.items():
mean_potential = np.mean(energy)
# skip zero and near 0 potential (The latter sometimes causes overflow issues during error calc.)
if np.abs(mean_potential) < 1e-6:
continue
results[comp]['keys'].append(list(pair))
results[comp]['mean_potential'].append(mean_potential)
data_dict[pair] = energy # Only store non-zero energies
# Calculate the error separately over multiple processes
nproc = dynamic_cpu_assignment(NPROC)
logger.debug('Calculating error for {}'.format(comp))
error_dict = get_error(data_dict, nproc)
for k in results[comp]['keys']:
results[comp]['error'].append(error_dict[tuple(k)])
nonbonded_dict[_id]['energy_components'] = ENERGY_COMPONENTS
nonbonded_dict[_id]['time'] = sim_time
nonbonded_dict[_id]['results'] = results
with open(outfile, 'w') as f:
json.dump(nonbonded_dict, f)
transformer_dict = {
'structure': {
'parent_structure_id': structure_dict['structure']['structure_id']
},
'files': {
'desmond_nonbonded': outfile,
'desmond_nonbonded_raw': outfile_raw
},
'custom': structure_dict['custom'],
}
if fork is not None:
logger.info('Forking pipeline: ' + ' '.join(fork))
transformer_dict['control'] = {'forks': fork}
yield transformer_dict
def preprocess_elec(api, dataset, data_type='default', no_ligand=False):
"""
Get pairwise electrostatic interactions
:param api:
:param dataset:
:param data_type:
:param no_ligand: If true do not rename the ligand residue
:return:
"""
if data_type is None:
raise ValueError(
'elec data_type not set. Specify elec data_type or use default data_types'
)
energy_component = 'nonbonded_elec'
if api is None:
dataset.dropna(axis=0,
subset=['desmond_cms', 'desmond_nonbonded'],
inplace=True)
elec_dict = dict((index, {}) for index in dataset.index)
for index, stid in zip(dataset.index, dataset.structure_id):
# Load data
if api is None:
msys_model, cms_model = topo.read_cms(dataset.loc[index,
'desmond_cms'])
with open(dataset.loc[index, 'desmond_nonbonded'], 'r') as fh:
nonbonded_dict = json.load(fh)
else:
# Load desmond_cms
msys_model, cms_model = get_desmond_cms(api, stid)
# Load desmond_nonbonded
nonbonded_dict = get_desmond_nonbonded(api, stid)
# Check if data_type in nonbonded_dict
if data_type not in nonbonded_dict:
if data_type != 'default':
raise ValueError('{} data type not calculated for {}'.format(
data_type, stid))
elif 'group_ids' in nonbonded_dict:
logger.warning('{} uses outdated file format'.format(stid))
else:
raise ValueError('{} data type not claculated for {}'.format(
data_type, stid))
else:
nonbonded_dict = nonbonded_dict[data_type]
# Determine Ligand residue
if no_ligand:
ligand_resid = None
elif 'ligand_resnum' not in dataset.columns and 'ligand_chain' not in dataset.columns:
ligands = find_ligands(st=cms_model)
assert (len(ligands) == 1)
ligand_resid = None
for atm in ligands[0].st.atom:
if ligand_resid is None:
ligand_resid = (atm.resnum, atm.chain.strip())
else:
if (atm.resnum, atm.chain.strip()) != ligand_resid:
raise ValueError(
'Structure {}\nFound multiple ligand residues.'.
format(stid))
else:
ligand_resid = (int(dataset.loc[index, 'ligand_resnum']),
dataset.loc[index, 'ligand_chain'])
# Map group ids
group_ids = nonbonded_dict['group_ids']
id2resid = {}
if all([type(gid) == list for gid in group_ids]):
for i, resid in enumerate(map(tuple, group_ids)):
if resid == ligand_resid:
id2resid[i] = 'LIG'
else:
id2resid[i] = '{}:{}'.format(*resid)
else:
logger.warning('Custom group_ids found')
for i, group_id in enumerate(group_ids):
id2resid[i] = group_id
elec_keys = list(
map(tuple, nonbonded_dict['results'][energy_component]['keys']))
elec_means = nonbonded_dict['results'][energy_component][
'mean_potential']
for key, mean in zip(elec_keys, elec_means):
ri, rj = list(map(int, key))
pair = sorted([
id2resid[ri], id2resid[rj]
]) # Sort ensure common pairs are always assigned the same pair_id
pair_id = '{} - {}'.format(*pair)
if pair_id not in elec_dict[index]:
elec_dict[index][pair_id] = mean
df_elec = pd.DataFrame(elec_dict).T
# replace NaNs with 0
df_elec.fillna(0, inplace=True)
df_elec.index = dataset.structure_id
return df_elec