def _init(self):
from natsort import natsorted
folders = natsorted(glob(path.join(self.generatorspath, '*', ''))) # I need the extra '' to add a finishing /
if len(folders) == 0:
logger.info('Generators folder has no subdirectories, using folder itself')
folders.append(self.generatorspath)
numF = len(folders)
numCopies = np.ones(numF, dtype=int) * int(np.floor(self.nmax / numF))
numExtra = np.mod(self.nmax, numF)
extraChoices = np.random.choice(numF, numExtra, replace=False) # draw the extra
numCopies[extraChoices] += 1
# numCopies = numCopies + np.random.multinomial(numExtra, [1/numF]*numF) # draw the extra equally from a flat distribution
if not path.exists(self.inputpath):
makedirs(self.inputpath)
# Check if epoch 1 directories already exist in the input folder
existing = glob(path.join(self.inputpath, 'e1s*'))
if len(existing) != 0:
raise NameError('Epoch 1 directories already exist.')
k = 1
for i in range(numF):
for j in range(numCopies[i]):
name = _simName(folders[i])
inputdir = path.join(self.inputpath, 'e1s' + str(k) + '_' + name)
#src = path.join(self.generatorspath, name, '*')
src = folders[i]
copytree(src, inputdir, symlinks=True, ignore=ignore_patterns(*_IGNORE_EXTENSIONS))
k += 1
def _writeInputsFunction(i, f, epoch, inputpath, coorname):
regex = re.compile('(e\d+s\d+)_')
frameNum = f.frame
piece = f.piece
if f.sim.parent is None:
currSim = f.sim
else:
currSim = f.sim.parent
traj = currSim.trajectory[piece]
if currSim.input is None:
raise NameError('Could not find input folder in simulation lists. Cannot create new simulations.')
wuName = _simName(traj)
res = regex.search(wuName)
if res: # If we are running on top of adaptive, use the first name part for the next sim name
wuName = res.group(1)
# create new job directory
newName = 'e' + str(epoch) + 's' + str(i + 1) + '_' + wuName + 'p' + str(piece) + 'f' + str(frameNum)
newDir = path.join(inputpath, newName, '')
# copy previous input directory including input files
copytree(currSim.input, newDir, symlinks=False, ignore=ignore_patterns('*.coor', '*.rst', '*.out', *_IGNORE_EXTENSIONS))
# overwrite input file with new one. frameNum + 1 as catdcd does 1 based indexing
mol = Molecule(currSim.molfile) # Always read the mol file, otherwise it does not work if we need to save a PDB as coorname
mol.read(traj)
mol.dropFrames(keep=frameNum) # Making sure only specific frame to write is kept
mol.write(path.join(newDir, coorname))
def _writeInputsFunction(i, f, epoch, inputpath, coorname):
regex = re.compile('(e\d+s\d+)_')
frameNum = f.frame
piece = f.piece
if f.sim.parent is None:
currSim = f.sim
else:
currSim = f.sim.parent
traj = currSim.trajectory[piece]
if currSim.input is None:
raise NameError('Could not find input folder in simulation lists. Cannot create new simulations.')
wuName = _simName(traj)
res = regex.search(wuName)
if res: # If we are running on top of adaptive, use the first name part for the next sim name
wuName = res.group(1)
# create new job directory
newName = 'e' + str(epoch) + 's' + str(i + 1) + '_' + wuName + 'p' + str(piece) + 'f' + str(frameNum)
newDir = path.join(inputpath, newName, '')
# copy previous input directory including input files
copytree(currSim.input, newDir, symlinks=False, ignore=ignore_patterns('*.coor', '*.rst', '*.out', *_IGNORE_EXTENSIONS))
# overwrite input file with new one. frameNum + 1 as catdcd does 1 based indexing
mol = Molecule(currSim.molfile) # Always read the mol file, otherwise it does not work if we need to save a PDB as coorname
mol.read(traj)
mol.dropFrames(keep=frameNum) # Making sure only specific frame to write is kept
mol.write(path.join(newDir, coorname))
def _init(self):
from natsort import natsorted
folders = natsorted(glob(path.join(self.generatorspath, '*', ''))) # I need the extra '' to add a finishing /
if len(folders) == 0:
logger.info('Generators folder has no subdirectories, using folder itself')
folders.append(self.generatorspath)
numF = len(folders)
numCopies = np.ones(numF, dtype=int) * int(np.floor(self.nmax / numF))
numExtra = np.mod(self.nmax, numF)
extraChoices = np.random.choice(numF, numExtra, replace=False) # draw the extra
numCopies[extraChoices] += 1
# numCopies = numCopies + np.random.multinomial(numExtra, [1/numF]*numF) # draw the extra equally from a flat distribution
if not path.exists(self.inputpath):
makedirs(self.inputpath)
# Check if epoch 1 directories already exist in the input folder
existing = glob(path.join(self.inputpath, 'e1s*'))
if len(existing) != 0:
raise NameError('Epoch 1 directories already exist.')
k = 1
for i in range(numF):
for j in range(numCopies[i]):
name = _simName(folders[i])
inputdir = path.join(self.inputpath, 'e1s' + str(k) + '_' + name)
#src = path.join(self.generatorspath, name, '*')
src = folders[i]
copytree(src, inputdir, symlinks=True, ignore=ignore_patterns(*_IGNORE_EXTENSIONS))
k += 1
def _writeInputsFunction(i, f, epoch, inputpath):
regex = re.compile('(e\d+s\d+)_')
frameNum = f.frame
piece = f.piece
if f.sim.parent is None:
currSim = f.sim
else:
currSim = f.sim.parent
traj = currSim.trajectory[piece]
if currSim.input is None:
raise NameError(
'Could not find input folder in simulation lists. Cannot create new simulations.'
)
wuName = _simName(traj)
res = regex.search(wuName)
if res: # If we are running on top of adaptive, use the first name part for the next sim name
wuName = res.group(1)
# create new job directory
newName = 'e' + str(epoch) + 's' + str(i + 1) + '_' + wuName + 'p' + str(
piece) + 'f' + str(frameNum)
newDir = path.join(inputpath, newName, '')
# copy previous input directory including input files
copytree(currSim.input,
newDir,
symlinks=False,
ignore=ignore_patterns(_TRAJ_EXTENSIONS_COOR))
# overwrite input file with new one. frameNum + 1 as catdcd does 1 based indexing
mol = Molecule()
mol.read(traj)
mol.frame = frameNum
mol.write(path.join(newDir, 'input.coor'))
def _writeInputs(self, simsframes, epoch=None):
if epoch is None:
epoch = self._getEpoch() + 1
test = glob(path.join(self.inputpath, 'e' + str(epoch) + '*'))
if len(test) != 0:
raise NameError('Input dirs of epoch ' + str(epoch) +
' already exists.')
if path.exists(
path.join(self.inputpath,
'e' + str(epoch) + '_writeinputs.log')):
raise NameError('Epoch logfile already exists. Cant overwrite it.')
fid = open(
path.join(self.inputpath, 'e' + str(epoch) + '_writeinputs.log'),
'w')
regex = re.compile('(e\d+s\d+)_')
for i, f in enumerate(simsframes):
frameNum = f.frame
piece = f.piece
#print(frameNum)
if f.sim.parent is None:
currSim = f.sim
else:
currSim = f.sim.parent
traj = currSim.trajectory[piece]
if currSim.input is None:
raise NameError(
'Could not find input folder in simulation lists. Cannot create new simulations.'
)
wuName = _simName(traj)
res = regex.search(wuName)
if res: # If we are running on top of adaptive, use the first name part for the next sim name
wuName = res.group(1)
# create new job directory
newName = 'e' + str(epoch) + 's' + str(
i + 1) + '_' + wuName + 'p' + str(piece) + 'f' + str(frameNum)
newDir = path.join(self.inputpath, newName, '')
# copy previous input directory including input files
copytree(currSim.input,
newDir,
symlinks=False,
ignore=ignore_patterns('*.dcd', '*.xtc', '*.coor'))
# overwrite input file with new one. frameNum + 1 as catdcd does 1 based indexing
mol = Molecule()
mol.read(traj)
mol.frame = frameNum
mol.write(path.join(newDir, 'input.coor'))
# write nextInput file
fid.write('# {0} \n{1} {2}\n'.format(newName, traj, frameNum))
fid.close()
def _writeInputs(self, simsframes, epoch=None):
if epoch is None:
epoch = self._getEpoch() + 1
test = glob(path.join(self.inputpath, 'e' + str(epoch) + '*'))
if len(test) != 0:
raise NameError('Input dirs of epoch ' + str(epoch) + ' already exists.')
if path.exists(path.join(self.inputpath, 'e' + str(epoch) + '_writeinputs.log')):
raise NameError('Epoch logfile already exists. Cant overwrite it.')
fid = open(path.join(self.inputpath, 'e' + str(epoch) + '_writeinputs.log'), 'w')
regex = re.compile('(e\d+s\d+)_')
for i, f in enumerate(simsframes):
frameNum = f.frame
piece = f.piece
#print(frameNum)
if f.sim.parent is None:
currSim = f.sim
else:
currSim = f.sim.parent
traj = currSim.trajectory[piece]
if currSim.input is None:
raise NameError('Could not find input folder in simulation lists. Cannot create new simulations.')
wuName = _simName(traj)
res = regex.search(wuName)
if res: # If we are running on top of adaptive, use the first name part for the next sim name
wuName = res.group(1)
# create new job directory
newName = 'e' + str(epoch) + 's' + str(i+1) + '_' + wuName + 'p' + str(piece) + 'f' + str(frameNum)
newDir = path.join(self.inputpath, newName, '')
# copy previous input directory including input files
copytree(currSim.input, newDir, symlinks=False, ignore=ignore_patterns('*.dcd', '*.xtc', '*.coor'))
# overwrite input file with new one. frameNum + 1 as catdcd does 1 based indexing
mol = Molecule()
mol.read(traj)
mol.frame = frameNum
mol.write(path.join(newDir, 'input.coor'))
# write nextInput file
fid.write('# {0} \n{1} {2}\n'.format(newName, traj, frameNum))
fid.close()
def reconstructAdaptiveTraj(simlist, trajID):
""" Reconstructs a long trajectory out of short adaptive runs.
Parameters
----------
simlist : numpy.ndarray of :class:`Sim <htmd.simlist.Sim>` objects
A simulation list generated by the :func:`simlist <htmd.simlist.simlist>` function
trajID : int
The id of the trajectory from which to start going back.
Returns
-------
mol : :class:`Molecule <moleculekit.molecule.Molecule>` object
A Molecule object containing the reconstructed trajectory
chain : np.ndarray
The simulation IDs of all simulations involved
pathlist : np.ndarray of str
The names of all simulations involved.
Examples
--------
>>> mol, chain, pathlist = reconstructAdaptiveTraj(data.simlist, 52)
"""
sim = None
for s in simlist:
if s.simid == trajID:
sim = s
break
if sim is None:
raise NameError(
'Could not find sim with ID {} in the simlist.'.format(trajID))
pathlist = []
pathlist.append(sim.trajectory[0])
chain = []
chain.append((sim, -1, -1))
epo = None
while epo != 1:
[sim, piece, frame,
epo] = _findprevioustraj(simlist, _simName(sim.trajectory[0]))
pathlist.append(sim.trajectory[piece])
chain.append((sim, piece, frame))
pathlist = pathlist[::-1]
chain = chain[::-1]
mol = Molecule(sim.molfile)
mol.coords = np.zeros((mol.numAtoms, 3, 0), dtype=np.float32)
mol.fileloc = []
mol.box = np.zeros((3, 0))
for i, c in enumerate(chain):
tmpmol = Molecule(sim.molfile)
tmpmol.read(c[0].trajectory)
endpiece = c[1]
fileloc = np.vstack(tmpmol.fileloc)
filenames = fileloc[:, 0]
pieces = np.unique(filenames)
firstpieceframe = np.where(filenames == pieces[endpiece])[0][0]
endFrame = firstpieceframe + c[2]
if endFrame != -1:
tmpmol.coords = tmpmol.coords[:, :, 0:endFrame +
1] # Adding the actual respawned frame (+1) since the respawned sim doesn't include it in the xtc
tmpmol.fileloc = tmpmol.fileloc[0:endFrame + 1]
tmpmol.box = tmpmol.box[:, 0:endFrame + 1]
mol.coords = np.concatenate((mol.coords, tmpmol.coords), axis=2)
mol.box = np.concatenate((mol.box, tmpmol.box), axis=1)
mol.fileloc += tmpmol.fileloc
#mol.fileloc[:, 1] = range(np.size(mol.fileloc, 0))
return mol, chain, pathlist
def reconstructAdaptiveTraj(simlist, trajID):
""" Reconstructs a long trajectory out of short adaptive runs.
Parameters
----------
simlist : numpy.ndarray of :class:`Sim <htmd.simlist.Sim>` objects
A simulation list generated by the :func:`simlist <htmd.simlist.simlist>` function
trajID : int
The id of the trajectory from which to start going back.
Returns
-------
mol : :class:`Molecule <htmd.molecule.molecule.Molecule>` object
A Molecule object containing the reconstructed trajectory
chain : np.ndarray
The simulation IDs of all simulations involved
pathlist : np.ndarray of str
The names of all simulations involved.
Examples
--------
>>> mol, chain, pathlist = reconstructAdaptiveTraj(data.simlist, 52)
"""
sim = None
for s in simlist:
if s.simid == trajID:
sim = s
break
if sim is None:
raise NameError('Could not find sim with ID {} in the simlist.'.format(trajID))
pathlist = []
pathlist.append(sim.trajectory[0])
chain = []
chain.append((sim, -1, -1))
epo = None
while epo != 1:
[sim, piece, frame, epo] = _findprevioustraj(simlist, _simName(sim.trajectory[0]))
pathlist.append(sim.trajectory[piece])
chain.append((sim, piece, frame))
pathlist = pathlist[::-1]
chain = chain[::-1]
mol = Molecule(sim.molfile)
mol.coords = np.zeros((mol.numAtoms, 3, 0), dtype=np.float32)
mol.fileloc = []
mol.box = np.zeros((3, 0))
for i, c in enumerate(chain):
tmpmol = Molecule(sim.molfile)
tmpmol.read(c[0].trajectory)
endpiece = c[1]
fileloc = np.vstack(tmpmol.fileloc)
filenames = fileloc[:, 0]
pieces = np.unique(filenames)
firstpieceframe = np.where(filenames == pieces[endpiece])[0][0]
endFrame = firstpieceframe + c[2]
if endFrame != -1:
tmpmol.coords = tmpmol.coords[:, :, 0:endFrame + 1] # Adding the actual respawned frame (+1) since the respawned sim doesn't include it in the xtc
tmpmol.fileloc = tmpmol.fileloc[0:endFrame + 1]
tmpmol.box = tmpmol.box[:, 0:endFrame + 1]
mol.coords = np.concatenate((mol.coords, tmpmol.coords), axis=2)
mol.box = np.concatenate((mol.box, tmpmol.box), axis=1)
mol.fileloc += tmpmol.fileloc
#mol.fileloc[:, 1] = range(np.size(mol.fileloc, 0))
return mol, chain, pathlist
