def dumpCorrelationToMatlab(self, correlation, fileName, **analysisOptions):
results = self._performAnalysis(correlation, **analysisOptions)
results = self._organizeByGroup(results)
from mlabwrap import mlab
groups = []
for group, correlations in results.iteritems():
wrapped = []
for i, data in enumerate(correlations):
info = _infoToMatlab(data.info, mlab)
points = np.array(data.points)
wrapped.append(mlab.struct("info", info,
"points", points,
"name", str(data.name)))
groups.append(mlab.struct("group", group,
"correlations", _toStructArray(wrapped)))
mlab._set(correlation, _toStructArray(groups))
mlab.save(fileName, correlation)
def cobra_model_object_to_cobra_matlab_struct(cobra_model):
"""This function converts all of the object values to the
corresponding model fields to update the mlab model.
"""
try:
from mlabwrap import mlab as matlab
except:
raise Exception('mlabwrap and MATLAB are required to use these functions. '+\
'They only function on Mac OS X and GNU/Linux')
from cobra.mlab import python_list_to_matlab_cell, scipy_sparse_to_mlab_sparse
if hasattr(cobra_model, 'to_array_based_model'):
cobra_model = cobra_model.to_array_based_model()
else:
cobra_model.update()
matlab_struct = matlab.struct()
#Things that need a conversion: S, rxnGeneMat,
matlab_struct.mets = python_list_to_matlab_cell([x.id for x in cobra_model.metabolites],
transpose = True)
matlab_struct.metNames = python_list_to_matlab_cell([x.name for x in cobra_model.metabolites],
transpose = True)
matlab_struct.metFormulas = python_list_to_matlab_cell([x.formula
for x in cobra_model.metabolites],
transpose = True)
matlab_struct.genes = python_list_to_matlab_cell(cobra_model.genes, transpose = True)
matlab_struct.grRules = python_list_to_matlab_cell([x.gene_reaction_rule
for x in cobra_model.reactions],
transpose = True)
matlab_struct.rxns = python_list_to_matlab_cell([x.id for x in cobra_model.reactions],
transpose = True)
matlab_struct.rxnNames = python_list_to_matlab_cell([x.name
for x in cobra_model.reactions],
transpose = True)
matlab_struct.subSystems = python_list_to_matlab_cell([x.subsystem
for x in cobra_model.reactions],
transpose=True)
if hasattr(cobra_model, 'constraint_sense'):
matlab_struct.csense = reduce(lambda x,y: x+y, cobra_model.constraint_sense)
#matlab_struct.csense = python_list_to_matlab_cell(['E']*len(cobra_model.metabolites), transpose = True)
#TODO: inefficient conversion but who cares? matlab's on its way out
matlab_struct.S = scipy_sparse_to_mlab_sparse(cobra_model.S)
#Things that can be directly copied
matlab_struct.b = cobra_model.b
matlab_struct.c = cobra_model.objective_coefficients
matlab_struct.lb = cobra_model.lower_bounds
matlab_struct.ub = cobra_model.upper_bounds
matlab_struct.rev = [x.reversibility for x in cobra_model.reactions]
matlab_struct.description = cobra_model.description
return(matlab_struct)
def trackBacteria(images, debug=False):
from mlabwrap import mlab
masks = computeForegroundMasks(images)
#simpleImages = SimpleImageSeq(images)
#blobs = findBlobs(simpleImages, SimpleImageSeq(masks))
#if debug:
# debugBlobs(simpleImages, blobs)
#ellipses = findEllipses(blobs)
ellipses = findEllipses(masks)
params = mlab.struct("mem", 1, "dim", 2, "good", 2, "quiet", 0)
matrix = ellipses.selectColumns("x", "y", "angle", "area", "time")
maxDist = 5
while maxDist > 0.5:
try:
tracks = TracksMatrix(mlab.track(matrix, maxDist, params))
break
except:
maxDist /= 2.0
timeUnits = images.info.dt
lengthUnits = images.info.pixel
velocities = [VelocityMatrix(t*timeUnits) for t in range(1, len(images))]
oldId = tracks[0, "id"]
oldPosition = tracks[0, ("x", "y")]*lengthUnits
oldTime = 0
for row in range(1, tracks.shape[0]):
position = tracks[row, ("x", "y")]*lengthUnits
time = tracks[row, "time"]
angle = tracks[row, "angle"]
id_ = tracks[row, "id"]
if oldId != id_:
oldId = id_
else:
dt = (time - oldTime)*timeUnits
velocity = (position - oldPosition)/dt
combined = position.tolist() + velocity.tolist()
velocities[int(time) - 1].append(*(combined + [angle]))
oldPosition = position
oldTime = time
for matrix in velocities: matrix.compact()
return PathData(images.info, velocities)
def fit(self, X, y):
"""
Fits the IVM
Parameters
----------
X : array-like, shape (n_samples, n_features)
Set of samples, where n_samples is the number of samples and
n_features is the number of features.
Returns
-------
self : object
Returns self
"""
self.classes_, y = np.unique(y, return_inverse=True)
y = y + 1 #convert to matlab style indexing
if self.CV is None:
param_struct = mlab.struct('sigma', self.sigma, 'lambda', self._lambda, 'Nadd', self.Nadd, 'output', self.output,
'maxIter', self.maxIter, 'epsilon', self.epsilon, 'delta_k', self.delta_k, 'tempInt', self.tempInt,
'epsilon_back', self.epsilon_back, 'flyComputeK', self.flyComputeK, 'deselect', self.deselect)
self.model = mlab.ivm_learn(X.T, y, param_struct)
else:
param_struct = mlab.struct('sigmas', np.array(self.sigmas).reshape((1,-1)), 'lambdas', np.array(self.lambdas).reshape((1,-1)), 'CV', self.CV, 'Nadd', self.Nadd, 'output', self.output,
'maxIter', self.maxIter, 'epsilon', self.epsilon, 'delta_k', self.delta_k, 'tempInt', self.tempInt,
'epsilon_back', self.epsilon_back, 'flyComputeK', self.flyComputeK, 'deselect', self.deselect)
data_struct = mlab.struct('phi', X.T, 'c', y, 'phit', X.T, 'ct', y)
result = mlab.ivm(data_struct, param_struct)
self.model = result.model
return self
def dumpCorrelationToMatlab(self, correlation, fileName,
**analysisOptions):
results = self._performAnalysis(correlation, **analysisOptions)
results = self._organizeByGroup(results)
from mlabwrap import mlab
groups = []
for group, correlations in results.iteritems():
wrapped = []
for i, data in enumerate(correlations):
info = _infoToMatlab(data.info, mlab)
points = np.array(data.points)
wrapped.append(
mlab.struct("info", info, "points", points, "name",
str(data.name)))
groups.append(
mlab.struct("group", group, "correlations",
_toStructArray(wrapped)))
mlab._set(correlation, _toStructArray(groups))
mlab.save(fileName, correlation)
def dumpPathsToMatlab(self, fileName, **trackOptions):
self.track(**trackOptions)
from mlabwrap import mlab
allPaths = []
for info in self.infos:
try:
paths = self._loadPaths(info)
except:
continue
print info.uniqueName
frames = _makeCells([v.array for v in paths.velocities], mlab)
allPaths.append(mlab.struct("info", _infoToMatlab(paths.info, mlab),
"frames", _wrapCellValue(frames)))
mlab._set("paths", _toStructArray(allPaths))
mlab.save(fileName, "paths")
def dumpPathsToMatlab(self, fileName, **trackOptions):
self.track(**trackOptions)
from mlabwrap import mlab
allPaths = []
for info in self.infos:
try:
paths = self._loadPaths(info)
except:
continue
print info.uniqueName
frames = _makeCells([v.array for v in paths.velocities], mlab)
allPaths.append(
mlab.struct("info", _infoToMatlab(paths.info, mlab), "frames",
_wrapCellValue(frames)))
mlab._set("paths", _toStructArray(allPaths))
mlab.save(fileName, "paths")