def braineacData(self, probeList, naVal="NA"):
"""
Get braineac data for the specified list of probes.
"""
# create a brain object
self.b = mbo.brainObj()
# the list of brineac nodes
nodeList = [v for v in self.regValDict.keys()]
nodeList.sort()
# iterate through the braineac regions
for n, k in enumerate(nodeList):
# open the data file for the appropriate regions
inFile = "expr_" + self.regValDict[k] + '.txt'
f = open(path.join(self.braineacDir, inFile), "r")
lines = f.readlines()
# have a look through each line of the data file to look for the appropriate probe data
for line in lines:
bits = line.split()
if bits[0] in probeList:
pMean = np.mean(
[float(v) for v in bits[1:] if not v == naVal])
self.b.G.add_node(n, attr_dict={bits[0]: pMean})
def comparison(self, assMat, delim=" ", spatialFile="parcel_500.txt", parcellation="parcels/parcel_500.nii",
outFile="braineacNodes.txt", midLine=44.5):
"""
now let's match up the imaging data to the braineac data
to create a brain object and import an association matrix.
"""
self.a = mbo.brainObj()
self.a.importAdjFile(assMat, delimiter=delim)
# get the spatial information and parcellation scheme
self.a.importSpatialInfo(spatialFile)
self.a.importISO(parcellation)
self.a.parcels(self.a.G.nodes())
# set up output file
outFile = open(outFile, "w")
for node in self.a.G.nodes():
self.a.G.node['braineac'] = None
self.nodeDict = {n:{"L":[], "R":[]} for n in range(1,8)}
for n in self.nodeDict.keys():
tmp = self.a.parcelList[np.where(self.out==n)]
nodeList = [int(v) for v in np.unique(tmp.data.flatten()) if not v==0]
# check if 50% of parcel is in the braineac region
for node in nodeList:
rat = float(len(self.a.parcelList[np.where(self.a.parcelList==node) and np.where(self.a.parcelList==n)])) / float(len(self.a.parcelList[np.where(self.a.parcelList==node)]))
if rat < 0.5:
print str(node)+" less than 50% in "+str(n)
elif self.a.G.node[node-1]["xyz"][0] < midLine:
self.nodeDict[n]["L"].append(node-1)
self.a.G.node[node-1]['braineac'] = self.regValDict[n]+"L"
else:
self.nodeDict[n]["R"].append(node-1)
self.a.G.node[node-1]['braineac'] = self.regValDict[n]+"R"
# write the output in to a file
outFile.writelines(self.regValDict[n]+"L" + ' ' + ' '.join([str(v) for v in self.nodeDict[n]["L"]])+'\n')
outFile.writelines(self.regValDict[n]+"R" + ' ' + ' '.join([str(v) for v in self.nodeDict[n]["R"]])+'\n')
outFile.close()
def braineacData(self, probeList, naVal="NA"):
"""
Get braineac data for the specified list of probes.
"""
# create a brain object
self.b = mbo.brainObj()
# the list of brineac nodes
nodeList = [v for v in self.regValDict.keys()]
nodeList.sort()
# iterate through the braineac regions
for n,k in enumerate(nodeList):
# open the data file for the appropriate regions
inFile = "expr_"+self.regValDict[k]+'.txt'
f = open(path.join(self.braineacDir,inFile), "r")
lines = f.readlines()
# have a look through each line of the data file to look for the appropriate probe data
for line in lines:
bits = line.split()
if bits[0] in probeList:
pMean = np.mean([float(v) for v in bits[1:] if not v==naVal])
self.b.G.add_node(n, attr_dict={bits[0]:pMean})
# 355, 250, 296, 491, 423, 140, 205, 434, 26, 244,
# 21, 142, 493, 76, 232, 234, 422]
# excluded nodes for Tim's original analysis
#excludedNodes = [28, 303, 355, 339, 131, 250, 491, 205, 423, 140, 434, 142, 235,
# 244, 493, 21, 26, 232, 76, 234, 422] # nodes with insufficient coverage
excludedNodes = [21, 26, 28, 35, 54, 73, 75, 76, 129, 133, 135, 140, 142, 160, 169, 177, 189, 232, 234, 235, 244, 245, 294, 304, 322, 339, 363, 386, 396, 419, 422, 423, 434, 476, 481, 493]
parcelFile = "parcel_500.txt"
thresholdtype = "local"
dVal = "2"
adjMatFile = "wave_cor_mat_level_"+dVal+"d_500_z.txt" # = "mean_2d_500.txt"
delim=" "
a = mbt.brainObj()
appVal = False
# unweighted measures
for e in edgePCCons:
ofb = '_'.join(["brain", thresholdtype, str(e), "d"+dVal+"_"])
propDict = {"edgePC":str(e)}
a.importAdjFile(adjMatFile, delimiter=delim, exclnodes=excludedNodes)
a.localThresholding(edgePC=e)
a.removeUnconnectedNodes()
# degs = a.G.degree(weight='weight')
# extras.writeResults(degs, "degreeWt", ofb, append=appVal)
a.binarise()
def __init__(self,
assocMat,
nodesToExclude=[],
delim=" ",
subjList=None,
spatialFile="parcel_500.txt",
symmetrise=False,
convertMNI=False,
mirror=True):
if subjList:
self.subjList = subjList
else:
self.subjList = [v for v in glob("17823*") if path.isdir(v)]
self.fName = "SampleAnnot.csv"
self.maFile = "MicroarrayExpression.csv"
self.probeFile = "Probes.csv"
self.mirror = mirror
# if symmetrise is true then regions are identified by the structure name,
# if symmetrise is false, then regions are identified by the structural acronym
# these refer to headers in the SampleAnnot.csv file
if symmetrise:
self.sLab = "structure_acronym"
else:
self.sLab = "structure_name"
# set up brain for expression data
self.a = mbo.brainObj()
node_counter = 0
self.headers = {}
self.sIDDict = {
} # dictionary storing the list of nodes for each structural ID by subject - for use later in averaging across all subjects
for subj in self.subjList:
self.sIDDict[subj] = {}
# import probe data
f = open(path.join(subj, self.fName), "rb")
reader = csv.DictReader(f, delimiter=",", quotechar='"')
self.headers[subj] = ['probe']
for l in reader:
# n = int(l["structure_id"])
sID = l[self.sLab]
n = node_counter # GIVE NODES UNIQUE INCREMENTAL ID (across all subjects)
if not self.a.G.has_node(n):
self.a.G.add_node(n)
self.a.G.node[n] = l
self.a.G.node[n][
'sID'] = sID # store the structure_acronym/structure_name for the node
node_counter += 1
# self.headers[subj].append(l["structure_id"])
self.headers[subj].append(
sID
) #STORE structure_acronym or structure_name depending on symmetrise
if not sID in self.sIDDict[subj].keys():
self.sIDDict[subj][sID] = [n]
else:
self.sIDDict[subj][sID].append(n)
f.close()
if convertMNI:
# convert location data for Allen brain from MNI space
for n in self.a.G.nodes():
x = 45 - (float(self.a.G.node[n]['mni_x']) / 2)
y = 63 + (float(self.a.G.node[n]['mni_y']) / 2)
z = 36 + (float(self.a.G.node[n]['mni_z']) / 2)
self.a.G.node[n]['xyz'] = (x, y, z)
else:
for n in self.a.G.nodes():
x = float(self.a.G.node[n]['mni_x'])
y = float(self.a.G.node[n]['mni_y'])
z = float(self.a.G.node[n]['mni_z'])
self.a.G.node[n]['xyz'] = (x, y, z)
if self.mirror and len(self.a.G.nodes()) < 600:
self.a.copyHemisphere()
# f.write('%s, %s, %s, %s \n' % (str(n), str(self.a.G.node[n]['xyz'][0]),str(self.a.G.node[n]['xyz'][1]),str(self.a.G.node[n]['xyz'][2])))
#f.close()
# set up brain with graph properties
self.c = mbo.brainObj()
self.c.importAdjFile(assocMat,
delimiter=delim,
exclnodes=nodesToExclude)
self.c.importSpatialInfo(spatialFile)
def __init__(self,
allenSubj,
assocMat,
delim=",",
spatialFile="atlas471_xyz_flip_xy.txt",
nodesToExclude=[],
symmetrise=False,
mirror=False,
convertMNI=False):
"""
This object contains two 'brain' network objects, one for the imaging data and
one for the Allen data. Embedded functions make a comparison between the imaging
and Allen data by pairing nodes between the two network objects.
"""
self.subj = allenSubj
self.fName = "SampleAnnot.csv"
self.maFile = "MicroarrayExpression.csv"
self.probeFile = "Probes.csv"
# if symmetrise is true then regions are identified by the structure name,
# if symmetrise is false, then regions are identified by the structural acronym
# these refer to headers in the SampleAnnot.csv file
if symmetrise:
self.sLab = "structure_acronym"
else:
self.sLab = "structure_name"
self.mirror = mirror
if symmetrise and self.mirror:
print "Please select either a symmetrised or mirrored graph, ignoring mirror=True"
self.mirror = False
# set up brain for expression data
self.a = mbo.brainObj()
node_counter = 0
# import probe data
f = open(path.join(self.subj, self.fName), "rb")
reader = csv.DictReader(f, delimiter=",", quotechar='"')
self.headers = ['probe']
self.sIDDict = {}
### deprecated code for removal ####
# for l in reader:
# n = int(l[self.sLab])
# self.a.G.add_node(n)
# self.a.G.node[n] = l
# self.headers.append(l["structure_id"])
# f.close()
for l in reader:
# n = int(l["structure_id"])
sID = l[self.sLab]
n = node_counter # GIVE NODES UNIQUE INCREMENTAL ID (across all subjects)
if not self.a.G.has_node(n):
self.a.G.add_node(n)
self.a.G.node[n] = l
self.a.G.node[n][
'sID'] = sID # store the structure_acronym/structure_name for the node
node_counter += 1
# self.headers[subj].append(l["structure_id"])
self.headers.append(
sID
) #STORE structure_acronym or structure_name depending on symmetrise
if not sID in self.sIDDict.keys():
self.sIDDict[sID] = [n]
else:
self.sIDDict[sID].append(n)
f.close()
if convertMNI:
# convert location data for Allen brain from MNI space
for n in self.a.G.nodes():
x = 45 - (float(self.a.G.node[n]['mni_x']) / 2)
y = 63 + (float(self.a.G.node[n]['mni_y']) / 2)
z = 36 + (float(self.a.G.node[n]['mni_z']) / 2)
self.a.G.node[n]['xyz'] = (x, y, z)
else:
for n in self.a.G.nodes():
x = float(self.a.G.node[n]['mni_x'])
y = float(self.a.G.node[n]['mni_y'])
z = float(self.a.G.node[n]['mni_z'])
self.a.G.node[n]['xyz'] = (x, y, z)
# copy hemisphere if required
if self.mirror and len(self.a.G.nodes()) < 600:
self.a.copyHemisphere()
# set up brain with graph properties
self.c = mbo.brainObj()
self.c.importAdjFile(assocMat,
delimiter=delim,
exclnodes=nodesToExclude)
self.c.importSpatialInfo(spatialFile)
def __init__(self, assocMat, nodesToExclude=[], delim=" ",
subjList=None, spatialFile="parcel_500.txt", symmetrise=False,
convertMNI=False, mirror=True):
if subjList:
self.subjList = subjList
else:
self.subjList = [v for v in glob("17823*") if path.isdir(v)]
self.fName = "SampleAnnot.csv"
self.maFile = "MicroarrayExpression.csv"
self.probeFile = "Probes.csv"
self.mirror=mirror
# if symmetrise is true then regions are identified by the structure name,
# if symmetrise is false, then regions are identified by the structural acronym
# these refer to headers in the SampleAnnot.csv file
if symmetrise:
self.sLab = "structure_acronym"
else:
self.sLab = "structure_name"
# set up brain for expression data
self.a = mbo.brainObj()
node_counter = 0
self.headers={}
self.sIDDict = {} # dictionary storing the list of nodes for each structural ID by subject - for use later in averaging across all subjects
for subj in self.subjList:
self.sIDDict[subj] = {}
# import probe data
f = open(path.join(subj, self.fName), "rb")
reader = csv.DictReader(f, delimiter=",", quotechar='"')
self.headers[subj] = ['probe']
for l in reader:
# n = int(l["structure_id"])
sID = l[self.sLab]
n = node_counter # GIVE NODES UNIQUE INCREMENTAL ID (across all subjects)
if not self.a.G.has_node(n):
self.a.G.add_node(n)
self.a.G.node[n] = l
self.a.G.node[n]['sID'] = sID # store the structure_acronym/structure_name for the node
node_counter += 1
# self.headers[subj].append(l["structure_id"])
self.headers[subj].append(sID) #STORE structure_acronym or structure_name depending on symmetrise
if not sID in self.sIDDict[subj].keys():
self.sIDDict[subj][sID] = [n]
else:
self.sIDDict[subj][sID].append(n)
f.close()
if convertMNI:
# convert location data for Allen brain from MNI space
for n in self.a.G.nodes():
x = 45 - (float(self.a.G.node[n]['mni_x'])/2)
y = 63 + (float(self.a.G.node[n]['mni_y'])/2)
z = 36 + (float(self.a.G.node[n]['mni_z'])/2)
self.a.G.node[n]['xyz'] = (x,y,z)
else:
for n in self.a.G.nodes():
x = float(self.a.G.node[n]['mni_x'])
y = float(self.a.G.node[n]['mni_y'])
z = float(self.a.G.node[n]['mni_z'])
self.a.G.node[n]['xyz'] = (x,y,z)
if self.mirror and len(self.a.G.nodes()) < 600:
self.a.copyHemisphere()
# f.write('%s, %s, %s, %s \n' % (str(n), str(self.a.G.node[n]['xyz'][0]),str(self.a.G.node[n]['xyz'][1]),str(self.a.G.node[n]['xyz'][2])))
#f.close()
# set up brain with graph properties
self.c = mbo.brainObj()
self.c.importAdjFile(assocMat, delimiter=delim, exclnodes=nodesToExclude)
self.c.importSpatialInfo(spatialFile)
def __init__(self, allenSubj, assocMat, delim=",",
spatialFile="atlas471_xyz_flip_xy.txt", nodesToExclude=[],
symmetrise=False, mirror=False, convertMNI=False):
"""
This object contains two 'brain' network objects, one for the imaging data and
one for the Allen data. Embedded functions make a comparison between the imaging
and Allen data by pairing nodes between the two network objects.
"""
self.subj = allenSubj
self.fName = "SampleAnnot.csv"
self.maFile = "MicroarrayExpression.csv"
self.probeFile = "Probes.csv"
# if symmetrise is true then regions are identified by the structure name,
# if symmetrise is false, then regions are identified by the structural acronym
# these refer to headers in the SampleAnnot.csv file
if symmetrise:
self.sLab = "structure_acronym"
else:
self.sLab = "structure_name"
self.mirror = mirror
if symmetrise and self.mirror:
print "Please select either a symmetrised or mirrored graph, ignoring mirror=True"
self.mirror = False
# set up brain for expression data
self.a = mbo.brainObj()
node_counter = 0
# import probe data
f = open(path.join(self.subj, self.fName), "rb")
reader = csv.DictReader(f, delimiter=",", quotechar='"')
self.headers = ['probe']
self.sIDDict = {}
### deprecated code for removal ####
# for l in reader:
# n = int(l[self.sLab])
# self.a.G.add_node(n)
# self.a.G.node[n] = l
# self.headers.append(l["structure_id"])
# f.close()
for l in reader:
# n = int(l["structure_id"])
sID = l[self.sLab]
n = node_counter # GIVE NODES UNIQUE INCREMENTAL ID (across all subjects)
if not self.a.G.has_node(n):
self.a.G.add_node(n)
self.a.G.node[n] = l
self.a.G.node[n]['sID'] = sID # store the structure_acronym/structure_name for the node
node_counter += 1
# self.headers[subj].append(l["structure_id"])
self.headers.append(sID) #STORE structure_acronym or structure_name depending on symmetrise
if not sID in self.sIDDict.keys():
self.sIDDict[sID] = [n]
else:
self.sIDDict[sID].append(n)
f.close()
if convertMNI:
# convert location data for Allen brain from MNI space
for n in self.a.G.nodes():
x = 45 - (float(self.a.G.node[n]['mni_x'])/2)
y = 63 + (float(self.a.G.node[n]['mni_y'])/2)
z = 36 + (float(self.a.G.node[n]['mni_z'])/2)
self.a.G.node[n]['xyz'] = (x,y,z)
else:
for n in self.a.G.nodes():
x = float(self.a.G.node[n]['mni_x'])
y = float(self.a.G.node[n]['mni_y'])
z = float(self.a.G.node[n]['mni_z'])
self.a.G.node[n]['xyz'] = (x,y,z)
# copy hemisphere if required
if self.mirror and len(self.a.G.nodes()) < 600:
self.a.copyHemisphere()
# set up brain with graph properties
self.c = mbo.brainObj()
self.c.importAdjFile(assocMat, delimiter=delim,
exclnodes=nodesToExclude)
self.c.importSpatialInfo(spatialFile)
"""
from maybrain import brainObjs as mbt
from datetime import datetime
startTime = datetime.now() # record the starting time
# define some variables
parcelFile = "parcel_500.txt" # the file containing the spatial position of nodes in the brain
degenName = "randomAttack" # name of the degenerative model
dVal = "2" # this refers to the wavelet scale used to construct the association matrix
thresholdtype = "local" # the type of thresholding to use
am = "mean_500.txt" # the association matrix
nodesToExclude = [28, 303, 355, 339, 131, 250, 491, 205, 423, 140, 434, 142, 235, 244, 493, 21, 26, 232, 76, 234, 422] # a list of nodes that are not covered in most/all subjects on the imaging
a = mbt.brainObj() # create a brain object
a.importAdjFile(am, exclnodes=nodesToExclude, delimiter=",") # import the association matrix
a.localThresholding(edgePC=3) # apply local thresholding with 3 percent of all possible connections retained
a.removeUnconnectedNodes() # remove any unconnected nodes
a.importSpatialInfo(parcelFile) # import the spatial information
## total weights
weights = [mbt.np.absolute(a.G.edge[v[0]][v[1]]['weight']) for v in a.G.edges() ] # list all the weights in the graph
T_start = mbt.np.sum(weights) # starting sum of all graph weights
print T_start # record the starting time
wtLossPC = 5. # five percent loss of connections at each iteration
wtLoss = T_start * (wtLossPC/100) # calculate 5% of the connections
for n in range(3,4):
print "doing degenerative process"