def _parse(self, fname):
crdfile = open(fname, 'r')
try:
readingHeader = True
while readingHeader:
line = crdfile.readline()
if not len(line):
raise CharmmFileError('Premature end of file')
line = line.strip()
words = line.split()
if len(line) != 0:
if words[0] == 'ENERGIES' or words[0] == '!ENERGIES':
readingHeader = False
else:
self.header.append(line.strip())
else:
self.header.append(line.strip())
for row in xrange(len(self.header)):
if len(self.header[row].strip()) != 0:
line = self.header[row].strip().split()
if line[0][0:5] == 'NATOM' or line[0][0:6] == '!NATOM':
try:
line = self.header[row + 1].strip().split()
self.natom = int(line[0])
self.npriv = int(line[1]) # num. previous steps
self.nstep = int(line[2]) # num. steps in file
self.nsavc = int(line[3]) # coord save frequency
self.nsavv = int(line[4]) # velocities "
self.jhstrt = int(line[5]) # Num total steps?
break
except (ValueError, IndexError):
raise CharmmFileError(
'Problem parsing CHARMM restart')
self.scan(crdfile, '!XOLD')
self._get_formatted_crds(crdfile, self.coordsold)
self.scan(crdfile, '!VX')
self._get_formatted_crds(crdfile, self.vels)
self.scan(crdfile, '!X')
self._get_formatted_crds(crdfile, self.coords)
# Convert velocities to angstroms/ps
self.vels = [v * ONE_TIMESCALE for v in self.vels]
finally:
crdfile.close()
def _parse(self, fname):
crdfile = open(fname, 'r')
try:
line = crdfile.readline()
while len(line.strip()) == 0: # Skip whitespace, as a precaution
line = crdfile.readline()
intitle = True
while intitle:
self.title.append(line.strip())
line = crdfile.readline()
if len(line.strip()) == 0:
intitle = False
elif line.strip()[0] != '*':
intitle = False
else:
intitle = True
while len(line.strip()) == 0: # Skip whitespace
line = crdfile.readline()
try:
self.natom = int(line.strip().split()[0])
for row in xrange(self.natom):
line = crdfile.readline().strip().split()
self.atomno.append(int(line[0]))
self.resno.append(int(line[1]))
self.resname.append(line[2])
self.attype.append(line[3])
self.coords.append(float(line[4]))
self.coords.append(float(line[5]))
self.coords.append(float(line[6]))
self.segid.append(line[7])
self.resid.append(int(line[8]))
self.weighting.append(float(line[9]))
if 3 * self.natom != len(self.coords):
raise CharmmFileError(
"Error parsing CHARMM .crd file: %d "
"atoms requires %d coords (not %d)" %
(self.natom, 3 * self.natom, len(self.coords)))
except (ValueError, IndexError):
raise CharmmFileError('Error parsing CHARMM coordinate file')
finally:
crdfile.close()
def _convert(data, type, msg=''):
"""
Converts a data type to a desired type, raising CharmmFileError if it
fails
"""
try:
return type(data)
except ValueError:
raise CharmmFileError('Could not convert %s to %s' % (msg, type))
def _get_formatted_crds(self, crdfile, crds):
for row in xrange(self.natom):
line = crdfile.readline()
if not line:
raise CharmmFileError('Premature end of file')
if len(line) < 3 * charmlen:
raise CharmmFileError("Less than 3 coordinates present in "
"coordinate row or coords may be "
"truncated.")
line = line.replace('D', 'E') # CHARMM uses 'D' for exponentials
# CHARMM uses fixed format (len = charmlen = 22) for crds in .rst's
crds.append(float(line[0:charmlen]))
crds.append(float(line[charmlen:2 * charmlen]))
crds.append(float(line[2 * charmlen:3 * charmlen]))
def __init__(self, fname, mode='r'):
if mode not in ('r', 'w'):
raise ValueError('Cannot open CharmmFile with mode "%s"' % mode)
if mode == 'r':
self.status = 'OLD'
else:
self.status = 'NEW'
try:
self._handle = open(fname, mode)
except IOError, e:
raise CharmmFileError(str(e))
def scan(self, handle, str, r=0): # read lines in file till 'str' is found
scanning = True
if (r): handle.seek(0)
while scanning:
line = handle.readline()
if not line:
raise CharmmFileError('Premature end of file')
if len(line.strip()) != 0:
if line.strip().split()[0][0:len(str)] == str:
scanning = False
def read_topology_file(self, tfile):
"""
Reads _only_ the atom type definitions from a topology file. This is
unnecessary for versions 36 and later of the CHARMM force field.
Parameters
----------
tfile : str
Name of the CHARMM topology file to read
"""
conv = CharmmParameterSet._convert
if isinstance(tfile, str):
own_handle = True
f = iter(CharmmFile(tfile))
else:
own_handle = False
f = tfile
hpatch = tpatch = None # default Head and Tail patches
residues = dict()
patches = dict()
hpatches = dict()
tpatches = dict()
line = next(f)
try:
while line:
line = line.strip()
if line[:4] == 'MASS':
words = line.split()
try:
idx = conv(words[1], int, 'atom type')
name = words[2]
mass = conv(words[3], float, 'atom mass')
except IndexError:
raise CharmmFileError(
'Could not parse MASS section of '
'%s' % tfile)
# The parameter file might or might not have an element name
try:
elem = words[4]
if len(elem) == 2:
elem = elem[0] + elem[1].lower()
atomic_number = AtomicNum[elem]
except (IndexError, KeyError):
# Figure it out from the mass
atomic_number = AtomicNum[element_by_mass(mass)]
atype = AtomType(name=name,
number=idx,
mass=mass,
atomic_number=atomic_number)
self.atom_types_str[atype.name] = atype
self.atom_types_int[atype.number] = atype
self.atom_types_tuple[(atype.name, atype.number)] = atype
elif line[:4] == 'DECL':
pass # Not really sure what this means
elif line[:4] == 'DEFA':
words = line.split()
if len(words) < 5:
warnings.warn('DEFA line has %d tokens; expected 5' %
len(words))
else:
it = iter(words[1:5])
for tok, val in zip(it, it):
if val.upper() == 'NONE':
val = None
if tok.upper().startswith('FIRS'):
hpatch = val
elif tok.upper() == 'LAST':
tpatch = val
else:
warnings.warn('DEFA patch %s unknown' % val)
elif line[:4].upper() in ('RESI', 'PRES'):
restype = line[:4].upper()
# Get the residue definition
words = line.split()
resname = words[1]
# Assign default patches
hpatches[resname] = hpatch
tpatches[resname] = tpatch
try:
charge = float(words[2])
except (IndexError, ValueError):
warnings.warn('No charge for %s' % resname)
if restype == 'RESI':
res = ResidueTemplate(resname)
elif restype == 'PRES':
res = PatchTemplate(resname)
else:
assert False, 'restype != RESI or PRES'
line = next(f)
group = []
ictable = []
while line:
if line[:5].upper() == 'GROUP':
if group:
res.groups.append(group)
group = []
elif line[:4].upper() == 'ATOM':
words = line.split()
name, type, charge = words[1:4]
charge = float(charge)
atom = Atom(name=name, type=type, charge=charge)
group.append(atom)
res.add_atom(atom)
elif line.strip() and line.split()[0].upper() in \
('BOND', 'DOUBLE'):
words = line.split()[1:]
it = iter(words)
for a1, a2 in zip(it, it):
if a1.startswith('-'):
res.head = res[a2]
continue
if a2.startswith('-'):
res.head = res[a1]
continue
if a1.startswith('+'):
res.tail = res[a2]
continue
if a2.startswith('+'):
res.tail = res[a1]
continue
# Apparently PRES objects do not need to put +
# or - in front of atoms that belong to adjacent
# residues
if restype == 'PRES' and (a1 not in res
or a2 not in res):
continue
res.add_bond(a1, a2)
elif line[:4].upper() == 'CMAP':
pass
elif line[:5].upper() == 'DONOR':
pass
elif line[:6].upper() == 'ACCEPT':
pass
elif line[:2].upper() == 'IC':
w = line.split()[1:]
ictable.append((w[:4], [float(x) for x in w[4:]]))
elif line[:3].upper() == 'END':
break
elif line[:5].upper() == 'PATCH':
it = iter(line.split()[1:])
for tok, val in zip(it, it):
if val.upper() == 'NONE': val = None
if tok.upper().startswith('FIRS'):
hpatches[resname] = val
elif tok.upper().startswith('LAST'):
tpatches[resname] = val
elif line[:5].upper() == 'DELETE':
pass
elif line[:4].upper() == 'IMPR':
it = iter(line.split()[1:])
for a1, a2, a3, a4 in zip(it, it, it, it):
if a2[0] == '-' or a3[0] == '-' or a4 == '-':
res.head = res[a1]
elif line[:4].upper() in ('RESI', 'PRES', 'MASS'):
# Back up a line and bail
break
line = next(f)
if group: res.groups.append(group)
_fit_IC_table(res, ictable)
if restype == 'RESI':
residues[resname] = res
elif restype == 'PRES':
patches[resname] = res
else:
assert False, 'restype != RESI or PRES'
# We parsed a line we need to look at. So don't update the
# iterator
continue
# Get the next line and cycle through
line = next(f)
except StopIteration:
pass
# Go through the patches and add the appropriate one
for resname, res in residues.iteritems():
if hpatches[resname] is not None:
try:
res.first_patch = patches[hpatches[resname]]
except KeyError:
warnings.warn('Patch %s not found' % hpatches[resname])
if tpatches[resname] is not None:
try:
res.last_patch = patches[tpatches[resname]]
except KeyError:
warnings.warn('Patch %s not found' % tpatches[resname])
# Now update the residues and patches with the ones we parsed here
self.residues.update(residues)
self.patches.update(patches)
if own_handle: f.close()
def read_parameter_file(self, pfile):
"""
Reads all of the parameters from a parameter file. Versions 36 and
later of the CHARMM force field files have an ATOMS section defining
all of the atom types. Older versions need to load this information
from the RTF/TOP files.
Parameters
----------
pfile : str
Name of the CHARMM parameter file to read
Notes
-----
The atom types must all be loaded by the end of this routine. Either
supply a PAR file with atom definitions in them or read in a RTF/TOP
file first. Failure to do so will result in a raised RuntimeError.
"""
conv = CharmmParameterSet._convert
if isinstance(pfile, str):
own_handle = True
f = CharmmFile(pfile)
else:
own_handle = False
f = pfile
# What section are we parsing?
section = None
# The current cmap we are building (these span multiple lines)
current_cmap = None
current_cmap2 = None
current_cmap_data = []
current_cmap_res = 0
nonbonded_types = dict() # Holder
parameterset = None
read_first_nonbonded = False
for line in f:
line = line.strip()
if not line:
# This is a blank line
continue
if parameterset is None and line.strip().startswith('*>>'):
parameterset = line.strip()[1:78]
continue
# Set section if this is a section header
if line.startswith('ATOMS'):
section = 'ATOMS'
continue
if line.startswith('BONDS'):
section = 'BONDS'
continue
if line.startswith('ANGLES'):
section = 'ANGLES'
continue
if line.startswith('DIHEDRALS'):
section = 'DIHEDRALS'
continue
if line.startswith('IMPROPER'):
section = 'IMPROPER'
continue
if line.startswith('CMAP'):
section = 'CMAP'
continue
if line.startswith('NONBONDED'):
read_first_nonbonded = False
section = 'NONBONDED'
continue
if line.startswith('NBFIX'):
section = 'NBFIX'
continue
if line.startswith('HBOND'):
section = None
continue
# It seems like files? sections? can be terminated with 'END'
if line.startswith('END'): # should this be case-insensitive?
section = None
continue
# If we have no section, skip
if section is None: continue
# Now handle each section specifically
if section == 'ATOMS':
if not line.startswith('MASS'): continue # Should this happen?
words = line.split()
try:
idx = conv(words[1], int, 'atom type')
name = words[2]
mass = conv(words[3], float, 'atom mass')
except IndexError:
raise CharmmFileError('Could not parse MASS section.')
# The parameter file might or might not have an element name
try:
elem = words[4]
if len(elem) == 2:
elem = elem[0] + elem[1].lower()
atomic_number = AtomicNum[elem]
except (IndexError, KeyError):
# Figure it out from the mass
atomic_number = AtomicNum[element_by_mass(mass)]
atype = AtomType(name=name,
number=idx,
mass=mass,
atomic_number=atomic_number)
self.atom_types_str[atype.name] = atype
self.atom_types_int[atype.number] = atype
self.atom_types_tuple[(atype.name, atype.number)] = atype
continue
if section == 'BONDS':
words = line.split()
try:
type1 = words[0]
type2 = words[1]
k = conv(words[2], float, 'bond force constant')
req = conv(words[3], float, 'bond equilibrium dist')
except IndexError:
raise CharmmFileError('Could not parse bonds.')
key = (min(type1, type2), max(type1, type2))
bond_type = BondType(k, req)
self.bond_types[(type1, type2)] = bond_type
self.bond_types[(type2, type1)] = bond_type
continue
if section == 'ANGLES':
words = line.split()
try:
type1 = words[0]
type2 = words[1]
type3 = words[2]
k = conv(words[3], float, 'angle force constant')
theteq = conv(words[4], float, 'angle equilibrium value')
except IndexError:
raise CharmmFileError('Could not parse angles.')
angle_type = AngleType(k, theteq * DEG_TO_RAD)
self.angle_types[(type1, type2, type3)] = angle_type
self.angle_types[(type3, type2, type1)] = angle_type
# See if we have a urey-bradley
try:
ubk = conv(words[5], float, 'Urey-Bradley force constant')
ubeq = conv(words[6], float, 'Urey-Bradley equil. value')
ubtype = BondType(ubk, ubeq)
except IndexError:
ubtype = NoUreyBradley
self.urey_bradley_types[(type1, type2, type3)] = ubtype
self.urey_bradley_types[(type3, type2, type1)] = ubtype
continue
if section == 'DIHEDRALS':
words = line.split()
try:
type1 = words[0]
type2 = words[1]
type3 = words[2]
type4 = words[3]
k = conv(words[4], float, 'dihedral force constant')
n = conv(words[5], float, 'dihedral periodicity')
phase = conv(words[6], float, 'dihedral phase')
except IndexError:
raise CharmmFileError('Could not parse dihedrals.')
key = (type1, type2, type3, type4)
# See if this is a second (or more) term of the dihedral group
# that's already present.
dihedral = DihedralType(k, n, phase * DEG_TO_RAD)
if key in self.dihedral_types:
# See if the existing dihedral type list has a term with
# the same periodicity -- If so, replace it
replaced = False
for i, dtype in enumerate(self.dihedral_types[key]):
if dtype.per == dihedral.per:
# Replace. Warn if they are different
if dtype != dihedral:
warnings.warn('Replacing dihedral %r with %r' %
(dtype, dihedral))
self.dihedral_types[key][i] = dihedral
replaced = True
break
if not replaced:
self.dihedral_types[key].append(dihedral)
# Now do the other order
replaced = False
key = (type4, type3, type2, type1)
for i, dtype in enumerate(self.dihedral_types[key]):
if dtype.per == dihedral.per:
self.dihedral_types[key][i] = dihedral
replaced = True
break
if not replaced:
self.dihedral_types[key].append(dihedral)
else: # key not present
dtl = DihedralTypeList()
dtl.append(dihedral)
self.dihedral_types[(type1, type2, type3, type4)] = dtl
self.dihedral_types[(type4, type3, type2, type1)] = dtl
continue
if section == 'IMPROPER':
words = line.split()
try:
type1 = words[0]
type2 = words[1]
type3 = words[2]
type4 = words[3]
k = conv(words[4], float, 'improper force constant')
theteq = conv(words[5], float, 'improper equil. value')
except IndexError:
raise CharmmFileError('Could not parse dihedrals.')
# If we have a 7th column, that is the real psi0 (and the 6th
# is just a dummy 0)
try:
tmp = conv(words[6], float, 'improper equil. value')
theteq = tmp
except IndexError:
pass # Do nothing
# Improper types seem not to have the central atom defined in
# the first place, so just have the key a fully sorted list. We
# still depend on the PSF having properly ordered improper atoms
key = tuple(sorted([type1, type2, type3, type4]))
self.improper_types[key] = ImproperType(k, theteq * DEG_TO_RAD)
continue
if section == 'CMAP':
# This is the most complicated part, since cmap parameters span
# many lines. We won't do much error catching here.
words = line.split()
try:
holder = [float(w) for w in words]
current_cmap_data.extend(holder)
except ValueError:
# We assume this is a definition of a new CMAP, so
# terminate the last CMAP if applicable
if current_cmap is not None:
# We have a map to terminate
ty = CmapType(current_cmap_res, current_cmap_data)
self.cmap_types[current_cmap] = ty
self.cmap_types[current_cmap2] = ty
try:
type1 = words[0]
type2 = words[1]
type3 = words[2]
type4 = words[3]
type5 = words[4]
type6 = words[5]
type7 = words[6]
type8 = words[7]
res = conv(words[8], int, 'CMAP resolution')
except IndexError:
raise CharmmFileError('Could not parse CMAP data.')
# order the torsions independently
k1 = [
type1, type2, type3, type4, type5, type6, type7, type8
]
k2 = [
type8, type7, type6, type5, type4, type3, type2, type1
]
current_cmap = tuple(min(k1, k2))
current_cmap2 = tuple(max(k1, k2))
current_cmap_res = res
current_cmap_data = []
continue
if section == 'NONBONDED':
# Now get the nonbonded values
words = line.split()
try:
atype = words[0]
# 1st column is ignored
epsilon = conv(words[2], float, 'vdW epsilon term')
rmin = conv(words[3], float, 'vdW Rmin/2 term')
except IndexError:
# If we haven't read our first nonbonded term yet, we may
# just be parsing the settings that should be used. So
# soldier on
if not read_first_nonbonded: continue
raise CharmmFileError('Could not parse nonbonded terms.')
except CharmmFileError, e:
if not read_first_nonbonded: continue
raise CharmmFileError(str(e))
else:
# OK, we've read our first nonbonded section for sure now
read_first_nonbonded = True
# See if we have 1-4 parameters
try:
# 4th column is ignored
eps14 = conv(words[5], float, '1-4 vdW epsilon term')
rmin14 = conv(words[6], float, '1-4 vdW Rmin/2 term')
except IndexError:
eps14 = rmin14 = None
nonbonded_types[atype] = [epsilon, rmin, eps14, rmin14]
continue
if section == 'NBFIX':
words = line.split()
try:
at1 = words[0]
at2 = words[1]
emin = abs(conv(words[2], float, 'NBFIX Emin'))
rmin = conv(words[3], float, 'NBFIX Rmin')
try:
emin14 = abs(conv(words[4], float, 'NBFIX Emin 1-4'))
rmin14 = conv(words[5], float, 'NBFIX Rmin 1-4')
except IndexError:
emin14 = rmin14 = None
try:
self.atom_types_str[at1].add_nbfix(
at2, rmin, emin, rmin14, emin14)
self.atom_types_str[at2].add_nbfix(
at1, rmin, emin, rmin14, emin14)
except KeyError:
# Some stream files define NBFIX terms with an atom that
# is defined in another toppar file that does not
# necessarily have to be loaded. As a result, not every
# NBFIX found here will necessarily need to be applied.
# If we can't find a particular atom type, don't bother
# adding that nbfix and press on
pass
except IndexError:
raise CharmmFileError('Could not parse NBFIX terms.')
self.nbfix_types[(min(at1, at2), max(at1, at2))] = (emin, rmin)