def read_integrated_data(filename, listOfEntities, dataSet=None):
"""
DEPRECTATED! Just for reminiscing in the past... :)
Reads the results of a Fortran-BioParkin run.
The list of SBML entities is used to associate columns in the entityData file
with their corresponding entityData entities.
format:
C{[time [WHITESPACE] par 1 [WHITESPACE] par n}
e.g.:
C{0.0000000000000000E+00 1.4999999999999999E-02 0.0000000000000000E+00}
"""
try:
csv.register_dialect("SimulationData", delimiter=' ', quotechar='"', skipinitialspace=True)
reader = csv.reader(open(filename), dialect="SimulationData")
time = []
values = {}
for (i, row) in enumerate(reader): # we don't use the i... yet
for (j, value) in enumerate(row):
if j == 0: # jump over the 0th row
time.append(value) # time value is always in the 0th row
continue
columnID = j - 1 # possibility to compensate for column offset
entity = listOfEntities[columnID]
if values.has_key(entity):
values[entity].append(value) # append to existing list
else:
values[entity] = [value] # create list for the first time
# create final data structure
dataMap = {}
if len(values) > 0: # if we have dataMap
for entity in values.keys():
entityData = EntityData()
if dataSet:
entityData.setAssociatedDataSet(dataSet)
entityData.dataDescriptors = time
entityData.datapoints = values[entity]
entityData.type = entitydata.TYPE_SIMULATED
entityData.sbmlEntity = entity.wrappedEntity
entityData.originalFilename = filename
dataMap[entity] = entityData
return dataMap
except Exception, e:
logging.error("Error while trying to read simulation dataMap file: %s" % e)
def read_csv_data(filename, dataSet=None, parkinController=None):
'''
Parses a CSV (comma separated values) file with experimental data.
The CSV has to have a header line. Fields may be encapsulated by " (but they
don't have to be). The value format is e.g. "3.14".
Headers have the following convention:
- First column called C{"Timepoint [%UNIT%]"}, e.g. C{"Timepoint [s]"}
- Second column called C{"Unit"} for the time unit.
- One column per Species ID, given the unit in square brackets. E.g: C{"LH [mg/ml]"}
- Optionally: One column per Species Weight called "Weight" *immediately* after the ID column.
- All other columns are handled as kind of "meta data". So, there can be e.g. a "patient id" column.
The data of this column will be put in its own EntityData object with isMetaData = True.
'''
try:
csv.register_dialect("ExperimentalDataCSV",
delimiter='\t',
quotechar='"',
skipinitialspace=True)
reader = csv.reader(open(filename), dialect="ExperimentalDataCSV")
descriptors = []
descriptorUnit = ""
columnIDs = []
columnIDToUnit = {}
columnIDHasWeightColumn = {}
columnIDsToValues = {}
columnIndexToColumnID = {}
idToName = {}
originalHeaders = {}
columnIndexesWithWeightData = []
columnIndexesWithMetaData = [
] # for columns with "non-critical" information
columnIDsWithMetaData = []
previousId = None
# getting data from file and preparing it for a DataSet
for (i, row) in enumerate(reader):
if i == 0: # in header row
for (j, header) in enumerate(row):
if j == 0: # timepoint header
originalDescriptorHeader = header
if not "[" in header:
descriptorUnit = "N/A"
else:
descriptorUnit = header.split("[")[1].rsplit(
"]")[0].strip()
else: # all non-timepoint headers
if "[" in header: # the current cell is a Species definition
splitID = header.split("[")
id = splitID[0].strip()
unit = splitID[1].strip()[:-1]
columnIDs.append(id)
columnIDToUnit[id] = unit
columnIDsToValues[id] = []
columnIndexToColumnID[j] = id
originalHeaders[id] = header
previousId = id
elif previousId and STANDARD_DEVIATION in header.lower(
): # if in weight-defining cell
# radical change: handle weight columns as "normal" data, save them in separate EntityData
columnIDHasWeightColumn[
previousId] = True # will be used later on
columnIndexesWithWeightData.append(j)
id = "weight_%s" % previousId
columnIDs.append(id)
columnIDsToValues[id] = []
columnIndexToColumnID[j] = id
idToName[id] = "SD"
originalHeaders[id] = header
previousId = None
else: # in non-defined arbitrary cell, e.g. "Patient ID"
columnIndexesWithMetaData.append(j)
columnIDsWithMetaData.append(header)
columnIDs.append(header)
columnIDsToValues[header] = []
columnIndexToColumnID[j] = header
originalHeaders[header] = header
previousId = None
logging.debug(
"DataHandling.read_csv_data(): Reading 'meta data' of column %s"
% header)
else: # for all non-header rows
for (j, value) in enumerate(row):
if j == 0: # data descriptor column (e.g. timepoints)
descriptors.append(value)
elif j in columnIndexesWithMetaData: # handling "non-species" columns
id = columnIndexToColumnID[j]
columnIDsToValues[id].append(value)
previousId = None
elif columnIDHasWeightColumn.get(
previousId,
False): # this is a weight-defining cell
columnIDsToValues[previousId].append(value)
previousId = None
else:
id = columnIndexToColumnID[j]
columnIDsToValues[id].append(value)
# loop through the collected data and feed it into a OrderedDict with EntityData objects
# as values. This is the internal data structure of a DataSet object.
dataSet.dataDescriptors = descriptors # e.g. global timepoint list
dataSet.dataDescriptorUnit = descriptorUnit
dataSet.descriptorHeader = originalDescriptorHeader
lastEntityData = None
entityMap = OrderedDict() # keep Species order
for j, columnID in enumerate(columnIDs):
logging.debug("About to save data for column %s" % columnID)
entityData = EntityData()
entityData.dataDescriptorUnit = descriptorUnit
entityData.datapointUnit = columnIDToUnit.get(columnID, None)
entityData.setType(entitydata.TYPE_EXPERIMENTAL)
entityData.setAssociatedDataSet(
dataSet
) # always use this method, so SIGNALs are correctly connected!
entityData.originalId = columnID # can be used as an ID fallback (e.g. for plotting) later
entityData.originalHeader = originalHeaders[columnID]
if idToName.has_key(columnID):
entityData.setName(idToName[columnID])
if columnID in columnIDsWithMetaData:
entityData.isMetaData = True
if j in columnIndexesWithWeightData:
entityData.isWeightData = True
entityData.setWeightData(lastEntityData)
values = columnIDsToValues[columnID]
if not values:
logging.error(
"datahandling.load_csv_data(): No values for column %s" %
columnID)
continue
for i, value in enumerate(values):
try:
value = float(value)
except:
value = None
timepoint = descriptors[i]
entityData.dataDescriptors.append(timepoint)
entityData.datapoints.append(value)
# try to associate a Species in the currently loaded model (if any) with the current Species ID
if not columnID in columnIDsWithMetaData and parkinController and parkinController.ActiveModelController and parkinController.ActiveModelController.sbmlModel:
entityData.sbmlEntity = findSbmlEntity(
columnID, parkinController.ActiveModelController.sbmlModel)
if entityData.sbmlEntity:
entityMap[entityData.sbmlEntity] = entityData
else:
entityMap[columnID] = entityData
lastEntityData = entityData
return entityMap
except Exception, e:
logging.error(
"DataHandling.read_csv_data(): Error while reading file %s:\n%s" %
(filename, e))
def read_integrated_data(filename, listOfEntities, dataSet=None):
'''
Reads the results of a Fortran-BioParkin run.
The list of SBML entities is used to associate columns in the entityData file
with their corresponding entityData entities.
format:
C{[time [WHITESPACE] par 1 [WHITESPACE] par n}
e.g.:
C{0.0000000000000000E+00 1.4999999999999999E-02 0.0000000000000000E+00}
'''
try:
csv.register_dialect("SimulationData",
delimiter=' ',
quotechar='"',
skipinitialspace=True)
reader = csv.reader(open(filename), dialect="SimulationData")
time = []
values = {}
for (i, row) in enumerate(reader): # we don't use the i... yet
for (j, value) in enumerate(row):
if j == 0: # jump over the 0th row
time.append(value) # time value is always in the 0th row
continue
columnID = j - 1 # possibility to compensate for column offset
entity = listOfEntities[columnID]
if values.has_key(entity):
values[entity].append(value) # append to existing list
else:
values[entity] = [value] # create list for the first time
# create final data structure
dataMap = {}
if len(values) > 0: # if we have dataMap
for entity in values.keys():
entityData = EntityData()
if dataSet:
entityData.setAssociatedDataSet(dataSet)
entityData.dataDescriptors = time
entityData.datapoints = values[entity]
entityData.type = entitydata.TYPE_SIMULATED
entityData.sbmlEntity = entity.wrappedEntity
entityData.originalFilename = filename
dataMap[entity] = entityData
return dataMap
except Exception, e:
logging.error(
"Error while trying to read simulation dataMap file: %s" % e)
#def read_integrated_data_old(filename):
# '''
# OLD VERSION, DON'T USE!
# Only for looking up old code!
#
# Reads the results of a Fortran-BioParkin run.
#
# input format, e.g. see:
# /home/bzfwadem/workspace/BioParkin/compute/temp/P_gnrh_rec4_Solution.dat
#
# format:
# time \t par 1 \t par n
#
# '''
# data = {}
#
# #with csv.reader(open(filename), delimiter=' ', quotechar='"') as reader:
# try:
# csv.register_dialect("SimulationData", delimiter=' ', quotechar='"', skipinitialspace=True)
# reader = csv.reader(open(filename), dialect="SimulationData")
# time = []
# values = {}
# for (i, row) in enumerate(reader): # we don't use the i... yet
#
# for (j, value) in enumerate(row):
# if j == 0: # jump over the 0th row
# time.append(value) # time value is always in the 0th row
# continue
#
# columnID = j # possibility to compensate for column offset
# # we will use the column index as the key
# # TODO: We need something better than the column index. Column indices should be matched to previously loaded experimental data (or parts of the SBML model?)
# if values.has_key(columnID):
# values[columnID].append(value) # append to existing list
# else:
# values[columnID] = [value] # create list for the first time
#
# # create final data structure
# if len(values) > 0: # if we have data
# for dataID in values.keys():
# #if not data.has_key(dataID):
# entry = {services.dataservice.TIME: time,
# services.dataservice.VALUE: values[dataID]}
# data[str(dataID)] = entry #str: because the experimental data's keys are strings
# #else:
# # entry = data[dataID]
# # entry[Services.DataService.TIME] = time
# # entry[Services.DataService.VALUE].append(value)
#
# except Exception, e:
# logging.error("Error while trying to read simulation data file: %s" % e)
# # finally:
# # reader.close() # does not exist and does not seem to be necessary
#
#
# #
# # dictReader = csv.DictReader(open(filename), ["time", range(nrCols)-1])
# # unformattedData = dictReader.
#
# # TODO
# # for row in reader:
# # #time = float(row[0])
# # for (i, value) in row.enumerate_stripped():
# # if i == 0:
# # time = float(row[0])
# # else:
#
# return data
def read_csv_data(filename, dataSet=None, parkinController=None):
'''
Parses a CSV (comma separated values) file with experimental data.
The CSV has to have a header line. Fields may be encapsulated by " (but they
don't have to be). The value format is e.g. "3.14".
Headers have the following convention:
- First column called C{"Timepoint [%UNIT%]"}, e.g. C{"Timepoint [s]"}
- Second column called C{"Unit"} for the time unit.
- One column per Species ID, given the unit in square brackets. E.g: C{"LH [mg/ml]"}
- Optionally: One column per Species Weight called "Weight" *immediately* after the ID column.
- All other columns are handled as kind of "meta data". So, there can be e.g. a "patient id" column.
The data of this column will be put in its own EntityData object with isMetaData = True.
'''
try:
csv.register_dialect("ExperimentalDataCSV", delimiter='\t', quotechar='"', skipinitialspace=True)
reader = csv.reader(open(filename), dialect="ExperimentalDataCSV")
descriptors = []
descriptorUnit = ""
columnIDs = []
columnIDToUnit = {}
columnIDHasWeightColumn = {}
columnIDsToValues = {}
columnIndexToColumnID = {}
idToName = {}
originalHeaders = {}
columnIndexesWithWeightData = []
columnIndexesWithMetaData = [] # for columns with "non-critical" information
columnIDsWithMetaData = []
previousId = None
# getting data from file and preparing it for a DataSet
for (i, row) in enumerate(reader):
if i == 0: # in header row
for (j, header) in enumerate(row):
if j == 0: # timepoint header
originalDescriptorHeader = header
if not "[" in header:
descriptorUnit = "N/A"
else:
descriptorUnit = header.split("[")[1].rsplit("]")[0].strip()
else: # all non-timepoint headers
if "[" in header: # the current cell is a Species definition
splitID = header.split("[")
id = splitID[0].strip()
unit = splitID[1].strip()[:-1]
columnIDs.append(id)
columnIDToUnit[id] = unit
columnIDsToValues[id] = []
columnIndexToColumnID[j] = id
originalHeaders[id] = header
previousId = id
elif previousId and STANDARD_DEVIATION in header.lower(): # if in weight-defining cell
# radical change: handle weight columns as "normal" data, save them in separate EntityData
columnIDHasWeightColumn[previousId] = True # will be used later on
columnIndexesWithWeightData.append(j)
id = "weight_%s" % previousId
columnIDs.append(id)
columnIDsToValues[id] = []
columnIndexToColumnID[j] = id
idToName[id] = "SD"
originalHeaders[id] = header
previousId = None
else: # in non-defined arbitrary cell, e.g. "Patient ID"
columnIndexesWithMetaData.append(j)
columnIDsWithMetaData.append(header)
columnIDs.append(header)
columnIDsToValues[header] = []
columnIndexToColumnID[j] = header
originalHeaders[header] = header
previousId = None
logging.debug("DataHandling.read_csv_data(): Reading 'meta data' of column %s" % header)
else: # for all non-header rows
for (j, value) in enumerate(row):
if j == 0: # data descriptor column (e.g. timepoints)
descriptors.append(value)
elif j in columnIndexesWithMetaData: # handling "non-species" columns
id = columnIndexToColumnID[j]
columnIDsToValues[id].append(value)
previousId = None
elif columnIDHasWeightColumn.get(previousId, False): # this is a weight-defining cell
columnIDsToValues[previousId].append(value)
previousId = None
else:
id = columnIndexToColumnID[j]
columnIDsToValues[id].append(value)
# loop through the collected data and feed it into a OrderedDict with EntityData objects
# as values. This is the internal data structure of a DataSet object.
dataSet.dataDescriptors = descriptors # e.g. global timepoint list
dataSet.dataDescriptorUnit = descriptorUnit
dataSet.descriptorHeader = originalDescriptorHeader
lastEntityData = None
entityMap = OrderedDict() # keep Species order
for j, columnID in enumerate(columnIDs):
logging.debug("About to save data for column %s" % columnID)
entityData = EntityData()
entityData.dataDescriptorUnit = descriptorUnit
entityData.datapointUnit = columnIDToUnit.get(columnID, None)
entityData.setType(entitydata.TYPE_EXPERIMENTAL)
entityData.setAssociatedDataSet(dataSet) # always use this method, so SIGNALs are correctly connected!
entityData.originalId = columnID # can be used as an ID fallback (e.g. for plotting) later
entityData.originalHeader = originalHeaders[columnID]
if idToName.has_key(columnID):
entityData.setName(idToName[columnID])
if columnID in columnIDsWithMetaData:
entityData.isMetaData = True
if j in columnIndexesWithWeightData:
entityData.isWeightData = True
entityData.setWeightData(lastEntityData)
values = columnIDsToValues[columnID]
if not values:
logging.error("datahandling.load_csv_data(): No values for column %s" % columnID)
continue
for i, value in enumerate(values):
try:
value = float(value)
except :
value = None
timepoint = descriptors[i]
entityData.dataDescriptors.append(timepoint)
entityData.datapoints.append(value)
# try to associate a Species in the currently loaded model (if any) with the current Species ID
if not columnID in columnIDsWithMetaData and parkinController and parkinController.ActiveModelController and parkinController.ActiveModelController.sbmlModel:
entityData.sbmlEntity = findSbmlEntity(columnID, parkinController.ActiveModelController.sbmlModel)
if entityData.sbmlEntity:
entityMap[entityData.sbmlEntity] = entityData
else:
entityMap[columnID] = entityData
lastEntityData = entityData
return entityMap
except Exception, e:
logging.error("DataHandling.read_csv_data(): Error while reading file %s:\n%s" % (filename, e))
def read_integrated_data(filename, listOfEntities, dataSet=None):
'''
Reads the results of a Fortran-BioParkin run.
The list of SBML entities is used to associate columns in the entityData file
with their corresponding entityData entities.
format:
C{[time [WHITESPACE] par 1 [WHITESPACE] par n}
e.g.:
C{0.0000000000000000E+00 1.4999999999999999E-02 0.0000000000000000E+00}
'''
try:
csv.register_dialect("SimulationData", delimiter=' ', quotechar='"', skipinitialspace=True)
reader = csv.reader(open(filename), dialect="SimulationData")
time = []
values = {}
for (i, row) in enumerate(reader): # we don't use the i... yet
for (j, value) in enumerate(row):
if j == 0: # jump over the 0th row
time.append(value) # time value is always in the 0th row
continue
columnID = j - 1 # possibility to compensate for column offset
entity = listOfEntities[columnID]
if values.has_key(entity):
values[entity].append(value) # append to existing list
else:
values[entity] = [value] # create list for the first time
# create final data structure
dataMap = {}
if len(values) > 0: # if we have dataMap
for entity in values.keys():
entityData = EntityData()
if dataSet:
entityData.setAssociatedDataSet(dataSet)
entityData.dataDescriptors = time
entityData.datapoints = values[entity]
entityData.type = entitydata.TYPE_SIMULATED
entityData.sbmlEntity = entity.wrappedEntity
entityData.originalFilename = filename
dataMap[entity] = entityData
return dataMap
except Exception, e:
logging.error("Error while trying to read simulation dataMap file: %s" % e)
#def read_integrated_data_old(filename):
# '''
# OLD VERSION, DON'T USE!
# Only for looking up old code!
#
# Reads the results of a Fortran-BioParkin run.
#
# input format, e.g. see:
# /home/bzfwadem/workspace/BioParkin/compute/temp/P_gnrh_rec4_Solution.dat
#
# format:
# time \t par 1 \t par n
#
# '''
# data = {}
#
# #with csv.reader(open(filename), delimiter=' ', quotechar='"') as reader:
# try:
# csv.register_dialect("SimulationData", delimiter=' ', quotechar='"', skipinitialspace=True)
# reader = csv.reader(open(filename), dialect="SimulationData")
# time = []
# values = {}
# for (i, row) in enumerate(reader): # we don't use the i... yet
#
# for (j, value) in enumerate(row):
# if j == 0: # jump over the 0th row
# time.append(value) # time value is always in the 0th row
# continue
#
# columnID = j # possibility to compensate for column offset
# # we will use the column index as the key
# # TODO: We need something better than the column index. Column indices should be matched to previously loaded experimental data (or parts of the SBML model?)
# if values.has_key(columnID):
# values[columnID].append(value) # append to existing list
# else:
# values[columnID] = [value] # create list for the first time
#
# # create final data structure
# if len(values) > 0: # if we have data
# for dataID in values.keys():
# #if not data.has_key(dataID):
# entry = {services.dataservice.TIME: time,
# services.dataservice.VALUE: values[dataID]}
# data[str(dataID)] = entry #str: because the experimental data's keys are strings
# #else:
# # entry = data[dataID]
# # entry[Services.DataService.TIME] = time
# # entry[Services.DataService.VALUE].append(value)
#
# except Exception, e:
# logging.error("Error while trying to read simulation data file: %s" % e)
# # finally:
# # reader.close() # does not exist and does not seem to be necessary
#
#
# #
# # dictReader = csv.DictReader(open(filename), ["time", range(nrCols)-1])
# # unformattedData = dictReader.
#
# # TODO
# # for row in reader:
# # #time = float(row[0])
# # for (i, value) in row.enumerate_stripped():
# # if i == 0:
# # time = float(row[0])
# # else:
#
# return data
