def load_dens(self):
""" Returns multi-dim numpy array containing all density profiles from file self.fname """
try:
open(self.fname, "r")
except:
raise IOError, "File %s could not be opened." % (self.fname, )
return np.array(rxvg.lines_to_list_of_lists(rxvg.read_xvg(self.fname)))
def load_x_q_as_nparray(self):
""" loads x and q profile from a xvg file self.fname into self.x,q in-place """
try: open(self.fname,"r")
except:
print "File %s could not be opened." % (self.fname,)
raise IOError
try: self.x, self.q_noisy = np.split(np.array( rxvg.lines_to_list_of_lists(rxvg.read_xvg(self.fname)) ).ravel(order="F") , 2 )
except: raise RuntimeError, "File %s could not be loaded, possibly because of wrong formatting." % (self.fname,)
def load_x_q_as_nparray(self):
""" loads x and q profile from a xvg file self.fname into self.x,q in-place """
try:
open(self.fname, "r")
except:
print "File %s could not be opened." % (self.fname, )
raise IOError
try:
self.x, self.q_noisy = np.split(
np.array(rxvg.lines_to_list_of_lists(rxvg.read_xvg(
self.fname))).ravel(order="F"), 2)
except:
raise RuntimeError, "File %s could not be loaded, possibly because of wrong formatting." % (
self.fname, )
def read_trajs(self, binwidth):
"""
Reads in Plumed's Pitch files (3columns) or Gromacs pullx files (8 columns)
File format plain (Plumed, also suits pitch-n-roll output with change bias->roll):
------------
3 columns : floats
Time -- PITCH -- bias (roll for PNR)
crashes or unexpected behaviour if different format (e.g. more cols)
File format pnr_[roll, pitch] (for roll/pitch coordinate from pitch-n-roll code):
------------
3 columns : floats
Time -- PITCH -- ROLL
crashes or unexpected behaviour if different format (e.g. more cols)
File format plain2 (Plumed):
------------
2 columns (otherwise same)
File format xvg (Gromacs pullx):
------------
8 columns : float
Time -- x,y,z -- length -- dx,dy,dz
Returns
-------
trajs : list of dictionaries containing arrays
with the key convention of pytram
dict-key 'm' -- Markov state -- 0-based indexing
"""
def discretize(in_data, offset=0.0):
"""
turns data into integers, i.e.
discretizes data based on the offset and bin width
Discretized data are 0-based indices
(shifted with local var smallest)
Parameters
----------
data : np.array
data to be discretized
offset : float
starting value for the discretization scheme
"""
data = np.copy(in_data)
if offset != 0.0:
data -= offset
# binwidth variable taken from the
# enclosing scope (at definition time) -- closure
if binwidth != 1.0:
data /= binwidth
return data.astype(int)
smallest = None
trajs = []
for sim in self.sims:
filename = sim['fname']
tmpdict = {}
if self.traj_file_format == "plain":
# should contain exactly 3 columns
tmpdict['time'], tmpdict['x'], tmpdict['b'] = np.hsplit(
np.loadtxt(filename), 3)
# convert from kJ/mol to kT units
tmpdict['b'] /= k_b * sim['temperature']
elif self.traj_file_format == "plain2":
# should contain exactly 2 columns
tmpdict['time'], tmpdict['x'] = np.hsplit(
np.loadtxt(filename), 2)
elif self.traj_file_format == "xvg":
# create a numpy array with the contents of .xvg file;
# should contain exactly 8 columns
tmp_arr = np.array(
rxvg.lines_to_list_of_lists(rxvg.read_xvg(filename)))
tmpdict['time'] = tmp_arr[:, 0]
tmpdict['x'] = tmp_arr[:, 4]
elif "pnr" in self.traj_file_format:
# should contain exactly 3 columns
tmpdict['time'], tmpdict['pitch'], tmpdict['roll'] = np.hsplit(
np.loadtxt(filename), 3)
# convert from kJ/mol to kT units
if "roll" in self.traj_file_format:
tmpdict['x'] = tmpdict['roll']
elif "pitch" in self.traj_file_format:
tmpdict['x'] = tmpdict['pitch']
else:
raise RuntimeError, "only pitch -or- roll defined in the file format!"
# setting bias to 0, as the file format does not contain it anyway
# for the case some other method/procedure might need it
tmpdict['b'] = tmpdict['x'] * 0.0
else:
print "Given file format undefined: ", self.traj_file_format
raise RuntimeError
tmpdict['m'] = discretize(tmpdict['x'])
# find the smallest state-no
if smallest == None or smallest > tmpdict['m'].min():
smallest = tmpdict['m'].min()
# thermodynamic state assumed constant during each simulation
tmpdict['t'] = np.ones(tmpdict['m'].shape, dtype=int) * sim['t']
trajs.append(tmpdict)
# shift the trajs by the smallest state-no (smallest var)
for traj in trajs:
traj['m'] -= smallest
return trajs
def load_dens(self):
""" Returns multi-dim numpy array containing all density profiles from file self.fname """
try: open(self.fname,"r")
except:
raise IOError, "File %s could not be opened." % (self.fname,)
return np.array( rxvg.lines_to_list_of_lists(rxvg.read_xvg(self.fname)) )
def get_potential(charge_file,
Xref=14.8,
crop=0,
fcorr_type="cent_plat_zero",
field_intensity=0.0,
qcorr=True,
plot=False,
**kwargs):
""" Function get_potential returns the voltage drop (in V) across the membrane from simulations with two membranes.
It uses function int_q for integration, which starts the integration from the centre,
Scales the box length to the reference value Xref (approximate average from the simulations),
allows cropping from left/right sides,
It returns the absolute (positive) average value of the voltage drop (in V) taken from drops centre-left and centre-right.
"""
# Xref is in nm, value taken from the average of the longest simulations
#create a numpy array with the contents of .xvg file; nested: file -> lines -> list-of-lists(floats) -> np.array -> flatten(ravel) -> split in 2
X, Charge = np.split(
np.array(rxvg.lines_to_list_of_lists(
rxvg.read_xvg(charge_file))).ravel(order="F"), 2)
# V--- list of lists <----------- list of lines
# numpy array ----------------------------------------------> flattened ...
# .... and split .... ....... in ............. .......... two!
#Scale potential to the average Z-length
box_fluct_thres = 0.2
if abs(Xref - X.max()) > box_fluct_thres:
print "Scaling box length from %4.2f to %4.2f nm (average length from the longest simulation)" % (
X.max(), Xref)
print "This is quite a large difference. \n"
X *= Xref / X.max()
#Crop box ends
if crop >= 0:
X = X[crop:len(X) - crop]
Charge = Charge[crop:len(Charge) - crop]
#shift Box-Z centre to 0.0
#assumes that Box starts at z=0 and ends at some z=Length
X -= X[int(len(X) / 2)]
# total charge and dipole of the system
#Qtot = np.trapz(Charge, x=X)
#print "Net charge: %5.4f q/nm^2" % (Qtot,)
#Ptot = np.trapz(np.multiply(Charge,X), x=X)
#print "Net dipole: %5.4f D/nm^2 \n" % (1/0.393456*1.889725989e1*Ptot,) # Magic constants by Stepan
(Field, Potential, V) = int_q_twice(Charge, X, fcorr_type, field_intensity,
qcorr)
if plot:
import matplotlib.pyplot as plt
#Read-in densities
densities_file = kwargs.get("densfile")
Xdens, phos, tail, wat = np.genfromtxt(densities_file,
comments="#",
skip_header=24,
unpack=True)
#Scaling to the same average Z-length (see above)
Xdens *= Xref / Xdens.max()
Numdens = len(Xdens)
cendens = int(Numdens / 2)
Xdens_shift = Xdens - Xdens[cendens]
#Plots
fig = plt.figure()
ax1 = fig.add_subplot(111)
ax1.plot(Xdens_shift, phos / 256, 'b', label="P")
ax1.plot(Xdens_shift, tail / 512, 'm', label="Terminal C")
ax1.plot(Xdens_shift, wat / 10128, 'c', label="Water")
ax1.set_xlabel('Z [nm]')
ax1.set_ylabel(r'Number density normalized to unity [nm$^{-3}$]')
plt.legend(loc="upper left")
ax2 = ax1.twinx()
ax2.plot(X, Potential, 'k', label="Electrostatic potential", lw=2)
ax2.plot([X[0], X[len(X) - 1]], [0, 0], '--', label="Zero", lw=2)
ax2.set_ylabel('Potential [V]')
plt.legend(loc="upper center")
plt.show()
fig = plt.figure()
ax1 = fig.add_subplot(111)
ax1.plot(Xdens_shift, phos / 256, 'b', label="P")
ax1.plot(Xdens_shift, tail / 512, 'm', label="Terminal C")
ax1.plot(Xdens_shift, wat / 10128, 'c', label="Water")
ax1.set_xlabel('Z [nm]')
ax1.set_ylabel(r'Number density normalized to unity [nm$^{-3}$]')
plt.legend(loc="upper left")
ax2 = ax1.twinx()
ax2.plot(X, Field, 'k', label="Intensity of electric field", lw=2)
ax2.set_ylabel('Field [V/nm]')
plt.legend(loc="upper center")
plt.show()
return V
def get_potential(charge_file, Xref=14.8, crop=0, fcorr_type="cent_plat_zero", field_intensity=0.0, qcorr=True, plot=False, **kwargs):
""" Function get_potential returns the voltage drop (in V) across the membrane from simulations with two membranes.
It uses function int_q for integration, which starts the integration from the centre,
Scales the box length to the reference value Xref (approximate average from the simulations),
allows cropping from left/right sides,
It returns the absolute (positive) average value of the voltage drop (in V) taken from drops centre-left and centre-right.
"""
# Xref is in nm, value taken from the average of the longest simulations
#create a numpy array with the contents of .xvg file; nested: file -> lines -> list-of-lists(floats) -> np.array -> flatten(ravel) -> split in 2
X, Charge = np.split(np.array( rxvg.lines_to_list_of_lists(rxvg.read_xvg(charge_file)) ).ravel(order="F") , 2 )
# V--- list of lists <----------- list of lines
# numpy array ----------------------------------------------> flattened ...
# .... and split .... ....... in ............. .......... two!
#Scale potential to the average Z-length
box_fluct_thres=0.2
if abs(Xref-X.max())>box_fluct_thres:
print "Scaling box length from %4.2f to %4.2f nm (average length from the longest simulation)" % (X.max(),Xref)
print "This is quite a large difference. \n"
X *= Xref/X.max()
#Crop box ends
if crop>=0 :
X = X[crop:len(X)-crop]
Charge = Charge[crop:len(Charge)-crop]
#shift Box-Z centre to 0.0
#assumes that Box starts at z=0 and ends at some z=Length
X -= X[int(len(X)/2)]
# total charge and dipole of the system
#Qtot = np.trapz(Charge, x=X)
#print "Net charge: %5.4f q/nm^2" % (Qtot,)
#Ptot = np.trapz(np.multiply(Charge,X), x=X)
#print "Net dipole: %5.4f D/nm^2 \n" % (1/0.393456*1.889725989e1*Ptot,) # Magic constants by Stepan
(Field, Potential, V) = int_q_twice(Charge, X, fcorr_type, field_intensity, qcorr)
if plot:
import matplotlib.pyplot as plt
#Read-in densities
densities_file = kwargs.get("densfile")
Xdens, phos, tail, wat = np.genfromtxt(densities_file, comments="#", skip_header=24, unpack=True)
#Scaling to the same average Z-length (see above)
Xdens *= Xref/Xdens.max()
Numdens = len(Xdens)
cendens = int(Numdens/2)
Xdens_shift = Xdens - Xdens[cendens]
#Plots
fig = plt.figure()
ax1 = fig.add_subplot(111)
ax1.plot(Xdens_shift, phos/256, 'b', label="P")
ax1.plot(Xdens_shift, tail/512, 'm', label="Terminal C")
ax1.plot(Xdens_shift, wat/10128, 'c', label="Water")
ax1.set_xlabel('Z [nm]')
ax1.set_ylabel(r'Number density normalized to unity [nm$^{-3}$]')
plt.legend(loc="upper left")
ax2 = ax1.twinx()
ax2.plot(X, Potential, 'k', label="Electrostatic potential", lw=2)
ax2.plot([X[0],X[len(X)-1]], [0,0], '--', label="Zero", lw=2)
ax2.set_ylabel('Potential [V]')
plt.legend(loc="upper center")
plt.show()
fig = plt.figure()
ax1 = fig.add_subplot(111)
ax1.plot(Xdens_shift, phos/256, 'b', label="P")
ax1.plot(Xdens_shift, tail/512, 'm', label="Terminal C")
ax1.plot(Xdens_shift, wat/10128, 'c', label="Water")
ax1.set_xlabel('Z [nm]')
ax1.set_ylabel(r'Number density normalized to unity [nm$^{-3}$]')
plt.legend(loc="upper left")
ax2 = ax1.twinx()
ax2.plot(X, Field, 'k', label="Intensity of electric field", lw=2)
ax2.set_ylabel('Field [V/nm]')
plt.legend(loc="upper center")
plt.show()
return V
def read_trajs(self, binwidth):
"""
Reads in Plumed's Pitch files (3columns) or Gromacs pullx files (8 columns)
File format plain (Plumed, also suits pitch-n-roll output with change bias->roll):
------------
3 columns : floats
Time -- PITCH -- bias (roll for PNR)
crashes or unexpected behaviour if different format (e.g. more cols)
File format pnr_[roll, pitch] (for roll/pitch coordinate from pitch-n-roll code):
------------
3 columns : floats
Time -- PITCH -- ROLL
crashes or unexpected behaviour if different format (e.g. more cols)
File format plain2 (Plumed):
------------
2 columns (otherwise same)
File format xvg (Gromacs pullx):
------------
8 columns : float
Time -- x,y,z -- length -- dx,dy,dz
Returns
-------
trajs : list of dictionaries containing arrays
with the key convention of pytram
dict-key 'm' -- Markov state -- 0-based indexing
"""
def discretize(in_data, offset=0.0):
"""
turns data into integers, i.e.
discretizes data based on the offset and bin width
Discretized data are 0-based indices
(shifted with local var smallest)
Parameters
----------
data : np.array
data to be discretized
offset : float
starting value for the discretization scheme
"""
data = np.copy(in_data)
if offset != 0.0:
data -= offset
# binwidth variable taken from the
# enclosing scope (at definition time) -- closure
if binwidth != 1.0:
data /= binwidth
return data.astype(int)
smallest = None
trajs = []
for sim in self.sims:
filename = sim['fname']
tmpdict = {}
if self.traj_file_format == "plain" :
# should contain exactly 3 columns
tmpdict['time'], tmpdict['x'], tmpdict['b'] = np.hsplit(np.loadtxt(filename), 3)
# convert from kJ/mol to kT units
tmpdict['b'] /= k_b*sim['temperature']
elif self.traj_file_format == "plain2" :
# should contain exactly 2 columns
tmpdict['time'], tmpdict['x'] = np.hsplit(np.loadtxt(filename), 2)
elif self.traj_file_format == "xvg" :
# create a numpy array with the contents of .xvg file;
# should contain exactly 8 columns
tmp_arr = np.array( rxvg.lines_to_list_of_lists(rxvg.read_xvg(filename)) )
tmpdict['time'] = tmp_arr[:,0]
tmpdict['x'] = tmp_arr[:,4]
elif "pnr" in self.traj_file_format :
# should contain exactly 3 columns
tmpdict['time'], tmpdict['pitch'], tmpdict['roll'] = np.hsplit(np.loadtxt(filename), 3)
# convert from kJ/mol to kT units
if "roll" in self.traj_file_format :
tmpdict['x'] = tmpdict['roll']
elif "pitch" in self.traj_file_format :
tmpdict['x'] = tmpdict['pitch']
else:
raise RuntimeError, "only pitch -or- roll defined in the file format!"
# setting bias to 0, as the file format does not contain it anyway
# for the case some other method/procedure might need it
tmpdict['b'] = tmpdict['x']*0.0
else:
print "Given file format undefined: ", self.traj_file_format
raise RuntimeError
tmpdict['m'] = discretize(tmpdict['x'])
# find the smallest state-no
if smallest == None or smallest > tmpdict['m'].min():
smallest = tmpdict['m'].min()
# thermodynamic state assumed constant during each simulation
tmpdict['t'] = np.ones(tmpdict['m'].shape, dtype=int) * sim['t']
trajs.append(tmpdict)
# shift the trajs by the smallest state-no (smallest var)
for traj in trajs:
traj['m'] -= smallest
return trajs
