def __atom2residueMatrix(self, m):
"""
Reduce binary matrix of n x k atoms to binary matrix of i x j residues.
@param m: atom contact matrix,
array n x k with 1(contact) or 0(no contact)
@type m: array
@return: residue contact matrix,
2-D numpy array(residues_receptor x residues_ligand)
@rtype: array
"""
recInd = N0.concatenate(
(self.rec().resIndex(), [self.rec().lenAtoms()]))
ligInd = N0.concatenate(
(self.lig_model.resIndex(), [self.lig_model.lenAtoms()]))
residueMatrix = N0.zeros((len(recInd) - 1, len(ligInd) - 1), N0.Int)
for r in range(len(recInd) - 1):
for l in range(len(ligInd) - 1):
res2res = m[int(recInd[r]):int(recInd[r + 1]),
int(ligInd[l]):int(ligInd[l + 1])]
if N0.any(res2res):
residueMatrix[r, l] = 1
return residueMatrix
def concat( self, *traj ):
"""
Concatenate this with other trajectories. The ref model of the
new Trajectory is a 'semi-deep' copy of this trajectorie's model.
(see L{PDBModel.take()} )::
concat( traj [, traj2, traj3, ..] ) -> Trajectory
@param traj: one or more Trajectory with identical atoms as this one
@type traj: Trajectories
@return: concatenated trajecties
@rtype: Trajectory
"""
if len( traj ) == 0:
return self
r = self.__class__()
r.frames = N0.concatenate( (self.frames, traj[0].frames), 0 )
r.setRef( self.ref.clone())
if self.frameNames and traj[0].frameNames:
r.frameNames = self.frameNames + traj[0].frameNames
try:
if self.pc is not None and traj[0].pc is not None:
r.pc['p'] = N0.concatenate( (self.pc['p'], traj[0].pc['p']),0)
r.pc['u'] = N0.concatenate( (self.pc['u'], traj[0].pc['u']),0)
except TypeError, why:
EHandler.error('cannot concat PC '+str(why) )
def concatAtoms( self, *traj ):
"""
Concatenate 2 trajectories of same (frame) length 'horizontally', i.e.
for each frame the coordinates of one are appended to the coordinates
of the other. The ref model of the new trajectory is a 'semi-deep' copy
of this trajectory's model (see L{PDBModel.take()} )::
concatAtoms( traj1 [traj2, traj3..]) -> Trajectory
@param traj: one or more Trajectory of the same number of frames
@type traj: Trajectories
@return: trajectory with concatenated atoms
@rtype: Trajectory
"""
if len( traj ) == 0:
return self
r = self.__class__()
r.frames = N0.concatenate( (self.frames, traj[0].frames), 1 )
r.setRef( self.ref.concat( traj[0].getRef() ) )
r.profiles = self.profiles.clone()
r.frameNames = self.frameNames
return r.concatAtoms( *traj[1:] )
def __exposedResidues( self, ASA_values, sidechainCut=0.0,
backboneCut=0.0, totalCut=0.0 ):
"""
Decide what is a surface exposed residue and what is not.
sidechainCut, backboneCut, totalCut - float, cutoff value
for what will be considered as a exposed residue. All
three values have to pass the test.
@param ASA_values: array with ASA values for side chains, backbone
and total calculated in L{__read_residueASA}.
@type ASA_values: array
@param sidechainCut: cutoff ASA value for considering the side chain
to consider thew residue being exposed
(default: 0.0)
@type sidechainCut: float
@param backboneCut: cutoffvalue for back bone ASA
@type backboneCut: float
@param totalCut: cutoff for total ASA
@type totalCut: float
@return: residue mask, where 0 = burried
@rtype: [1|0]
"""
col_0 = N0.greater( N0.transpose(ASA_values)[0], totalCut )
col_1 = N0.greater( N0.transpose(ASA_values)[1], backboneCut )
col_2 = N0.greater( N0.transpose(ASA_values)[2], sidechainCut )
col_012 = N0.concatenate( ([col_0],[col_1],[col_2]) )
exposedList = N0.greater(N0.sum(col_012), 0)
return exposedList
def castHmmDic( self, hmmDic, repete, hmmGap, key ):
"""
Blow up hmmDic to the number of repetes of the profile used.
Correct scores for possible deletions in the search sequence.
@param hmmDic: dictionary from L{getHmmProfile}
@type hmmDic: dict
@param repete: repete information from L{align}
@type repete: int
@param hmmGap: information about gaps from L{align}
@type hmmGap: [int]
@param key: name of scoring method to adjust for gaps and repetes
@type key: str
@return: dictionary with information about the profile
@rtype: dict
"""
s = hmmDic[key]
for i in range( repete ):
mask = N0.ones( len(s) )
N0.put( mask, hmmGap[i], 0 )
if i == 0:
score = N0.compress( mask, s, 0 )
if i > 0:
score = N0.concatenate( ( N0.compress( mask, s, 0 ), score ) )
hmmDic[key] = score
return hmmDic
def test_ComplexTraj(self):
"""Dock.ComplexTraj test"""
import Biskit.tools as T
## there is no complex trajectory in the test folder so will have
## to create a fake trajectory with a complex
f = [T.testRoot() + '/com/1BGS.pdb'] * 5
t = Trajectory(f, verbose=self.local)
t = ComplexTraj(t, recChains=[0])
#if self.local:
#print 'plotting contact density...'
#t.plotContactDensity( step=2 )
## create a fake second chain in the ligand
for i in range(1093 + 98, 1968):
t.ref.atoms['chain_id'][i] = 'B'
t.ref.chainIndex(force=1, cache=1)
t.cl = [1, 2]
r = N0.concatenate((range(1093, 1191), range(0,
1093), range(1191, 1968)))
tt = t.takeAtoms(r)
contactMat = tt.atomContacts(1)
if self.local:
print 'Receptor chains: %s Ligand chains: %s' % (t.cr, t.cl)
self.assertEqual(N0.sum(N0.ravel(contactMat)), 308)
def castHmmDic(self, hmmDic, repete, hmmGap, key):
"""
Blow up hmmDic to the number of repetes of the profile used.
Correct scores for possible deletions in the search sequence.
@param hmmDic: dictionary from L{getHmmProfile}
@type hmmDic: dict
@param repete: repete information from L{align}
@type repete: int
@param hmmGap: information about gaps from L{align}
@type hmmGap: [int]
@param key: name of scoring method to adjust for gaps and repetes
@type key: str
@return: dictionary with information about the profile
@rtype: dict
"""
s = hmmDic[key]
for i in range(repete):
mask = N0.ones(len(s))
N0.put(mask, hmmGap[i], 0)
if i == 0:
score = N0.compress(mask, s, 0)
if i > 0:
score = N0.concatenate((N0.compress(mask, s, 0), score))
hmmDic[key] = score
return hmmDic
def __parseBiomt(self, pdbFile, firstLine):
"""
Extract BIOMT (biological unit) information from REMARK 350 lines
Creates a 'BIOMT' dictionary.
"""
line = firstLine
biomtDict = {}
moleculeNum = -1
while line[0] == 'REMARK' and line[1].startswith(' 350'):
# 5 = len(' 350 ')
biomtLine = line[1][5:].lstrip()
if biomtLine.startswith('BIOMOLECULE:'): # start a new molecule
if moleculeNum != -1:
# lets update the dictionary with what we've got
biomtDict[moleculeNum] = (targetChains, rtList)
#12 = len('BIOMOLECULE:')
moleculeNum = int(biomtLine[12:].strip())
targetChains = []
rotation = []
translation = []
rtList = []
matrixLine = 0
if biomtLine.startswith('APPLY THE FOLLOWING TO CHAINS:'):
# parse targeted chains, we assume this comes after BIOMOLECULE line
# 30 = len('APPLY THE FOLLOWING TO CHAINS:')
targetChains.extend(c.strip()
for c in biomtLine[30:].split(','))
if biomtLine.startswith('AND CHAINS:'):
# 11 = len('AND CHAINS:')
targetChains.extend(c.strip()
for c in biomtLine[11:].split(','))
if biomtLine.startswith('BIOMT'):
# parse rotate-translate matri{x/ces}, we assume this comes after BIOMOLECULE line
matrixLine += 1
# 6 = len('BIOMT#')
rawCoords = biomtLine[6:].split()
rotation.append([float(x) for x in rawCoords[1:4]])
translation.append(float(rawCoords[4]))
if matrixLine % 3 == 0:
rotation = N0.array(rotation)
translation = N0.transpose([translation])
rotation = N0.concatenate((rotation, translation), axis=1)
rtList.append(N0.array(rotation))
## rtList.append((rotation,translation))
rotation = []
translation = []
try:
line = pdbFile.readLine()
except ValueError, what:
self.log.add('Warning: Error parsing line %i of %s' %
(i, T.stripFilename(fname)))
self.log.add('\tError: ' + str(what))
continue
def __exposedResidues(self,
ASA_values,
sidechainCut=0.0,
backboneCut=0.0,
totalCut=0.0):
"""
Decide what is a surface exposed residue and what is not.
sidechainCut, backboneCut, totalCut - float, cutoff value
for what will be considered as a exposed residue. All
three values have to pass the test.
@param ASA_values: array with ASA values for side chains, backbone
and total calculated in L{__read_residueASA}.
@type ASA_values: array
@param sidechainCut: cutoff ASA value for considering the side chain
to consider thew residue being exposed
(default: 0.0)
@type sidechainCut: float
@param backboneCut: cutoffvalue for back bone ASA
@type backboneCut: float
@param totalCut: cutoff for total ASA
@type totalCut: float
@return: residue mask, where 0 = burried
@rtype: [1|0]
"""
col_0 = N0.greater(N0.transpose(ASA_values)[0], totalCut)
col_1 = N0.greater(N0.transpose(ASA_values)[1], backboneCut)
col_2 = N0.greater(N0.transpose(ASA_values)[2], sidechainCut)
col_012 = N0.concatenate(([col_0], [col_1], [col_2]))
exposedList = N0.greater(N0.sum(col_012), 0)
return exposedList
def __random_matrix( self ):
"""
Random rotation matrix.
@return: 4 x 4 array of float, random rotation and translation matrix
@rtype: array
"""
r = ma.randomRotation()
## r = N0.array([[1,0,0],[0,1,0],[0,0,1]],'f')
t = self.__random_translation()
## create 3 x 4 matrix: 0:3, 0:3 contains rot; 3,0:3 contains trans
result = N0.concatenate( (r, N0.transpose( [ t.tolist() ] )), 1)
## make it square
result = N0.concatenate( (result, N0.array([[0,0,0,1]], N0.Float32)), 0 )
return result
def rtTuple2matrix(self, r, t):
"""
Put rotation and translation matrix into single 4x4 matrix.
@param r: rotation matric, array 3x3 of float
@type r: array
@param t: translation vector, array 1x3 of float
@type t: vector
@return: rotation/translation matrix, array 4x4 of float
@rtype: array
"""
## create 3 x 4 matrix: 0:3, 0:3 contains rot; 3,0:3 contains trans
result = N0.concatenate((r, N0.transpose([t.tolist()])), 1)
## make it square
result = N0.concatenate((result, N0.array([[0, 0, 0, 1]], N0.Float32)),
0)
return result.astype(N0.Float32)
def rmsInterface(self, ref, cutoff=4.5, fit=1):
"""
Rmsd between this and reference interface. The interface is
defined as any residue that has an atom which is within the
distance given by |cutoff| from its partner.
@param ref: reference complex
@type ref: Complex
@param cutoff: atom distance cutoff for interface residue definition
(default: 4.5)
@type cutoff: float
@param fit: least-squares fit before calculating the rms (default: 1)
@type fit: 1|0
@return: interface rmad
@rtype: float
"""
## casting
this = self
if not ref.rec_model.equals( self.rec_model )[1] \
or not ref.lig_model.equals( self.lig_model )[1]:
m_rec, m_rec_ref, m_lig, m_lig_ref = self.equalAtoms(ref)
this = self.compress(m_rec, m_lig)
ref = ref.compress(m_rec_ref, m_lig_ref)
## determine interface
contacts = ref.resContacts(cutoff)
if_rec = ref.rec_model.res2atomMask(N0.sum(contacts, 1))
if_lig = ref.lig_model.res2atomMask(N0.sum(contacts, 0))
mask_interface = N0.concatenate((if_rec, if_lig))
mask_heavy = N0.concatenate(
(ref.rec().maskHeavy(), ref.lig_model.maskHeavy()))
mask_interface = mask_interface * mask_heavy
## rms
ref_model = ref.model()
this_model = this.model()
return ref_model.rms(this_model, mask_interface, fit=fit)
def convertChainIdsCter(self, model, chains):
"""
Convert normal chain ids to chain ids considering chain breaks.
"""
if len(chains) == 0:
return chains
## fetch last atom of given chains
index = N0.concatenate((model.chainIndex(), [len(model)]))
i = N0.take(index, N0.array(chains) + 1) - 1
## convert back to chain indices but this time including chain breaks
return model.atom2chainIndices(i, breaks=1)
def prepare( self ):
"""
Write a xyzrn coordinate file to disc.
Overrides Executor method.
"""
## get radiia and name array
p2x = Pdb2xyzrn(self.model, verbose=self.verbose, debug=self.debug )
r, n = p2x.run()
xyz = self.model.xyz
xyzr = N0.concatenate( ( xyz, N0.transpose([r]) ) ,axis=1 )
f = open( self.f_xyzrn, 'w' )
i = 0
for line in xyzr:
f.write( str(line)[2:-1] + ' 1 ' + n[i] + '\n')
i += 1
f.close()
def prepare( self ):
"""
Write a xyzrn coordinate file to disc.
Overrides Executor method.
"""
## get radiia and name array
p2x = Pdb2xyzrn(self.model, verbose=self.verbose, debug=self.debug )
r, n = p2x.run()
xyz = self.model.xyz
xyzr = N0.concatenate( ( xyz, N0.transpose([r]) ) ,axis=1 )
f = open( self.f_xyzrn, 'w' )
i = 0
for line in xyzr:
f.write( str(line)[2:-1] + ' 1 ' + n[i] + '\n')
i += 1
f.close()
def test_FuzzyCluster( self):
"""FuzzyCluster test"""
import gnuplot as G
x1 = R.random_sample((500,2))
x2 = R.random_sample((500,2)) + 1
x3 = R.random_sample((500,2)) + 2
self.x = N0.concatenate((x1, x2, x3))
self.fuzzy = FuzzyCluster(self.x, n_cluster=5, weight=1.5)
self.centers = self.fuzzy.go(1.e-30, n_iterations=50, nstep=10,
verbose=self.local)
if self.local:
print "cluster centers are displayed in green"
G.scatter( self.x, self.centers )
self.assertEqual( N0.shape(self.centers), (5, 2) )
def reduceXyz(self, xyz, axis=0):
"""
Reduce the number of atoms in the given coordinate set. The set must
have the same length and order as the reference model. It may have
an additional (time) dimension as first axis.
@param xyz: coordinates (N_atoms x 3) or (N_frames x N_atoms x 3)
@type xyz: array
@param axis: axis with atoms (default: 0)
@type axis: int
@return: coordinate array (N_less_atoms x 3) or
(N_frames x N_less_atoms x 3)
@rtype: array
"""
masses = self.m.atoms.get('mass')
r_xyz = None
for atom_indices in self.groups:
x = N0.take(xyz, atom_indices, axis)
m = N0.take(masses, atom_indices)
center = N0.sum(x * N0.transpose([
m,
]), axis=axis) / N0.sum(m)
if axis == 0:
center = center[N0.NewAxis, :]
if axis == 1:
center = center[:, N0.NewAxis, :]
if r_xyz is None:
r_xyz = center
else:
r_xyz = N0.concatenate((r_xyz, center), axis)
return r_xyz
def __alignMatrixDimension(self, cm, thisSeq, castSeq, axis=0):
"""
Correct one dimension of contactMatrix by inserting and deleting
columns, so that it can be later compared to contact matrices based
on slightly different sequences.
@param cm: contact matrix, 2D matrix of residue contacts
recceptor x ligand sequence
@type cm: array
@param thisSeq: AA sequence of this dimension of the contactMatrix
@type thisSeq: string
@param castSeq: AA sequence of this dimension in the other contact
@type castSeq: string
@param axis: which dimension to adapt (0=receptor, 1=ligand)
@type axis: 1|0
@return: contact matrix with residue contacts compatible to refSeq.
@rtype: 2D array
"""
# compare the two sequences
seqdiff = SequenceMatcher(None, thisSeq, castSeq)
seqDiff = seqdiff.get_opcodes()
## print seqDiff
# decide which dimension to work on
if not axis:
cm = N0.transpose(cm)
seqCount = 0 # keep track of sequence length changes
i = 0
for list in seqDiff:
# remove the column corresponding to the deletion in the
# docked sequence
if str(seqDiff[i][0]) == 'delete':
# separate matrix into before and after deletion
matrixSeg1 = cm[:, :seqDiff[i][1] + seqCount]
matrixSeg2 = cm[:, seqDiff[i][2] + seqCount:]
# concatenate part
cm = N0.concatenate((matrixSeg1, matrixSeg2), 1)
seqCount = seqCount + seqDiff[i][1] - seqDiff[i][2]
# inserts zeros in the column where there is a insertion in the
# docked sequence
if str(seqDiff[i][0]) == 'insert':
# create a matrix to be inserted
insertZeros = seqDiff[i][4] - seqDiff[i][3]
insertColumns = N0.array([[0] * insertZeros] * N0.size(cm, 0))
# separate matrix into before and after insertion
matrixSeg1 = cm[:, :seqDiff[i][1] + seqCount]
matrixSeg2 = cm[:, seqDiff[i][2] + seqCount:]
# concatenate parts with the zero matrix
cm = N0.concatenate((matrixSeg1, insertColumns, matrixSeg2), 1)
seqCount = seqCount + seqDiff[i][4] - seqDiff[i][3]
i = i + 1
if not axis:
return N0.transpose(cm)
return cm
