class LinearModel(SecondOrderCache):
'''
Core math matrix object for linear optimization modeling in reflux
Uses the Second Order Cache object as its foundation
Three main parts:
1)Sparce matrix of coeffecents
2)Row and column limits
3)Objective vector
Minimize:
Z = c*x
Subject to:
S*x <=> b
l < x < u
Extended Linear Control Model
With additional components:
4)Annotations
5)Targets
6)ControlMap
6)NaturalObjective
7)SyntheticObjective
'''
def __init__(self):
self.verbose = False
#Data Containers
self.modelName = ''
self.data = SecondOrderCache()
self.rowLimits = {}
self.columnLimits = {}
self.mipColumns = IndexedCache()
self.objective = {}
#Report of model annotation values (dict)
self.annotation = None
#Control
self.targets = None
self.controlMap = None
self.controlClusters = None
self.naturalObjective = None
self.syntheticObjective = None
#Variables
self.scale = 1 #! to be removed
self.defaultLowerLimit = None
self.defaultUpperLimit = None
def __str__(self):
return self.data.__str__()
def _addString(self,other):
print "string"
def _addScalar(self,other):
print "scalar"
def _addDict(self,other):
print "dict"
def _addLinearModel(self,other):
print "model"
def __add__(self,other):
if type(other) == type(""):
self._addString(other)
if type(other) == type(0) or type(other) ==type(0.0):
self._addScalar(other)
if type(other) == type({}):
self._addDict(other)
if type(other) == type(self):
self._addLinearModel(other)
return self
def __eq__(self,value):
if type(self) != type(value):
return False
a = self.data.equals(value.data)
b = self.rowLimits == value.rowLimits
c = self.columnLimits == value.columnLimits
d = self.objective == value.objective
e = self.mipColumns == value.mipColumns
result = a and b and c and d and e
return result
def _scaleTuple(self,value):
'''
scales a par of values
used in scaling of whole matrix
@type value: float
'''
(v1,v2) = value
if v1 != None:
v1 = float(v1*self.scale)
if v2 != None:
v2 = float(v2*self.scale)
return (v1, v2)
def _getDefaultColumnLimits(self):
'''
returns the column limits with defaults for values not given
'''
columnNames = set(self.getColumnNames())
limitNames = set(self.columnLimits.keys())
defaultingNames = columnNames.difference(limitNames)
result = {}
for name in defaultingNames:
if not (self.defaultLowerLimit == None and self.defaultUpperLimit == None):
result[name] = (self.defaultLowerLimit,self.defaultUpperLimit)
return result
def _getDefaultRowLimits(self):
'''
returns the row limits with defaults for values not given
'''
columnNames = set(self.getRowNames())
limitNames = set(self.rowLimits.keys())
defaultingNames = columnNames.difference(limitNames)
result = {}
for name in defaultingNames:
if not (self.defaultLowerLimit == None and self.defaultUpperLimit == None):
result[name] = (self.defaultLowerLimit,self.defaultUpperLimit)
return result
def _annotateString(self,value,annotationMap,regex,nsep=" "):
result = ''
tags = re.findall(regex,value)
for s in tags:
if s in annotationMap.keys():
r = annotationMap[s]
else:
r = s
result += nsep + r
result = result[len(nsep):]
return result
def _annotateMap(self,data,annotationMap,regex):
result = {}
for (key,value) in data.items():
ikey = self._annotateString(key, annotationMap, regex)
result[ikey] = value
return result
def _annotateList(self,data,annotationMap,regex):
result = []
for value in data:
ivalue = self._annotateString(value, annotationMap, regex)
result.append(ivalue)
return result
def annotateGeneList(self,data,annotationName = "bnumber", regex="[a-zA-Z0-9\(\)]+"):
result = data
if self.annotation == None:
return result
if annotationName in self.annotation.keys():
annotationMap = self.annotation[annotationName]
gMap = annotationMap.getColumn("gene")
if annotationMap != None:
result = self._annotateList(data,gMap,regex)
return result
def annotateGenes(self,objective,annotationName = "bnumber", regex="[a-zA-Z0-9\(\)]+"):
result = objective
if self.annotation == None:
return result
if annotationName in self.annotation.keys():
annotationMap = self.annotation[annotationName]
gMap = annotationMap.getColumn("gene")
if annotationMap != None:
result = self._annotateMap(objective,gMap,regex)
return result
def getGeneTargetMap(self):
if self.controlMap == None:
return None
result = {}
for (r,gs) in self.controlMap.items():
for g in gs:
if g not in result.keys():
result[g] = set()
result[g].add(r)
return result
def setProperty(self,name,value):
self.properties[name] = value
def getProperty(self,name):
return self.properties[name]
def setScalex(self,scale):
'''
Sets the scale for the matrix
Not entirely checked for completeness and usage
@param scale: the scaling factor for the limits
@type scale: float
'''
self.scale = scale
def addRowName(self,name):
'''
Adds a row name to matrix
@type name: string
'''
self.data.rowCache.addValue(name)
return None
def addColumnName(self,name):
'''
Adds column name to matrix
@type name: string
'''
self.data.columnCache.addValue(name)
return None
def setMipColumnName(self,name):
'''
Sets column with value name as a integer column
@type name: string
'''
self.mipColumns.addValue(name)
def setMipColumnNames(self,names,tag="%s"):
'''
@type names: string[]
'''
for name in names:
iname = tag % name
self.setMipColumnName(iname)
return None
def getMipColumnNames(self):
'''
Returns array of strings of column names which are set to integers
@rtype: string[]
'''
return self.mipColumns.getValues()
def getRowIndex(self,name):
'''
Returns index of row of value name
@type name: string
@rtype: int
'''
return self.data.rowCache.getindex(name)
def getColumnIndex(self,name):
'''
Returns index of column of value name
@type name: string
@rtype: int
'''
return self.data.columnCache.getindex(name)
def getRowNames(self):
'''
Returns an list of the row names
@rtype: string[]
'''
return self.data.rowCache.getValues()
def getColumnNames(self):
'''
Returns a list of the column names
@rtype: string[]
'''
return self.data.columnCache.getValues()
def addRowLimit(self,rowName,limit):
'''
Sets limits for selected row
@type rowName: string
@type limit (float,float)
'''
self.addRowName(rowName)
self.rowLimits[rowName] = self._scaleTuple(limit)
def addColumnLimit(self,columnName,limit):
'''
Sets limits for selected column
@columnName: string
@type limit (float,float)
'''
self.addColumnName(columnName)
self.columnLimits[columnName] = self._scaleTuple(limit)
def getRowLimit(self,name):
'''
returns limit for row
@rtype: (float,float)
'''
if name in self.rowLimits:
return self.rowLimits[name]
else:
return (None,None)
def getColumnLimit(self,name):
'''
returns limit for column
@rtype: (float,float)
'''
if name in self.columnLimits:
return self.columnLimits[name]
else:
return (None,None)
def addRowLimits(self,limits):
'''
Sets row limits from map
@type limits: {string:(float,float)}
'''
for key in limits.keys():
(lower,upper) = limits[key]
self.addRowLimit(key,(lower,upper))
def addColumnLimits(self,limits):
'''
Sets colum limits from map
@type limits: {string:(float,float)}
'''
for key in limits.keys():
(lower,upper) = limits[key]
self.addColumnLimit(key,(lower,upper))
def getRowLimits(self):
'''
returns map of row limits
@rtype: {name:(float,float)}
'''
result = self.rowLimits.copy()
defaultLimits = self._getDefaultRowLimits()
result.update(defaultLimits)
return result
def getColumnLimits(self):
'''
returns map of column limits
@rtype: {name:(float,float)}
'''
result = self.columnLimits.copy()
defaultLimits = self._getDefaultColumnLimits()
result.update(defaultLimits)
return result
def addObjective(self,columnName, value):
'''
sets objective coeffecent for a column
@type columnName: string
@type value: float
'''
self.objective[columnName] = value
def setObjective(self,objectiveMap):
'''
Sets objective coeffecents from map
@type objectiveMap: {sting:float}
'''
self.objective = {}
for key in objectiveMap.keys():
value = objectiveMap[key]
self.objective[key]=value
def getObjective(self):
'''
returns objective map
@rtype {string,float}
'''
return self.objective
def getRowValueMap(self,rowName):
'''
returns a dict of row values (column name: value)
@type rowName: string
@rtype {string,float}
'''
r = self.data.getRow(rowName)
#return r
result = {}
for key in self.data.keys():
if key[0] == rowName:
colName = key[1]
value = self.data.getValue(key[0],key[1])
result[colName] = value
if r != result:
pass
return result
def getColumnValueMap(self,colName):
'''
returns a dict of row values (column name: value)
@type rowName: string
@rtype {string,float}
'''
r = self.data.getColumn(colName)
#return r
result = {}
for key in self.data.keys():
if key[1] == colName:
rowName = key[0]
value = self.data.getValue(key[0],key[1])
#self.getValue(key[0],key[1])
result[rowName] = value
if r != result:
pass
return result
def getRowByValue(self,name,function):
result = {}
values = self.getRowValueMap(name)
for (k,v) in values.items():
if function(v):
result[k] = v
return result
def getColumnByValue(self,name,function):
result = {}
values = self.getColumnValueMap(name)
for (k,v) in values.items():
if function(v):
result[k] = v
return result
def getRowValuesFromPred(self,name,values):
result = {}
rvalues = self.getRowValueMap(name)
for (k,v) in rvalues.items():
vi = None
if k in values:
vi = values[k]
result[k] = (v,vi)
return result
def addData(self,rowName,columnName,value):
'''
Add data value to model matrix
@type rowName: string
@type columnName: string
@type value: string
'''
if value == None or value == 0.0:
return None
floatValue = float(value)
self.data.addValue(rowName,columnName,floatValue)
#self.addValue(rowName,columnName,floatValue)
return (0,0)
def addDataCache(self,data):
'''
Adds a data cach object to the coeffecent matrix
@type data: SecondOrderCache {(string,string):float}
'''
self.data.extend(data)
#self.data.rowCache.extend(data.rowCache)
#self.data.columnCache.extend(data.columnCache)
def getData(self,rowName,columnName):
'''
Get the coeffecent of the matrix
@type rowName: string
@type columnName: string
'''
return self.data.getValue(rowName,columnName)
#return self.getValue(rowName,columnName)
def removeData(self,rowName,columnName):
'''
removes a datapoint by row and column name
@type rowName: string
@type columnName: string
'''
self.data.removeValue(rowName,columnName)
def removeRow(self,rowName):
columnNames = self.data.rowMap[rowName]
self.data.removeRow(rowName)
if rowName in self.rowLimits:
del self.rowLimits[rowName]
for columnName in columnNames:
rCount = len(self.data.columnMap[columnName])
if rCount == 0:
self.removeColumn(columnName)
#if self.verbose: print "removing column [%s]" % (columnName)
if columnName in self.columnLimits.keys():
del self.columnLimits[columnName]
return None
def removeColumn(self, columnName):
rowNames = self.data.columnMap[columnName]
self.data.removeColumn(columnName)
if columnName in self.columnLimits:
del self.columnLimits[columnName]
if columnName in self.mipColumns.dataArray:
self.mipColumns.removeValue(columnName)
if self.targets != None:
if columnName in self.targets:
self.targets.removeValue(columnName)
if self.controlMap != None:
if columnName in self.controlMap.keys():
del self.controlMap[columnName]
#print "checking rownames [%s]" % (rowNames)
for rowName in rowNames:
cNames = self.data.rowMap[rowName]
if len(cNames) == 0:
self.removeRow(rowName)
#if self.verbose: print "removing row [%s]" % (rowName)
if rowName in self.rowLimits.keys():
del self.rowLimits[rowName]
return None
def addRow(self,rowName,data):
'''
add a row in the form of a dictonary
@type rowName: string
@type data: {string,float}
'''
for key in data.keys():
value = data[key]
self.addData(rowName,key,value)
def addColumn(self,columnName,modelVector):
'''
@type columnName: string
@type data: {string:float}
'''
for key in modelVector.keys():
value = modelVector[key]
self.addData(key,columnName,value)
def getSparseMatrix(self):
'''
@rtype (int,int,float)[]
'''
return self.data.getSparseMatrix()
def getSparseMatrixMap(self):
'''
return the index matrix
'''
return self.data.getIndexMatrix()
def addConstraints(self,model):
'''
Add a linear constraints to the current model
@type model: LinearModel
'''
self.data.extend(model.data)
rowLimits = model.getRowLimits()
columnLimits = model.getColumnLimits()
self.addRowLimits(rowLimits)
self.addColumnLimits(columnLimits)
self.mipColumns.extend(model.mipColumns)
return None
def addModel(self,model):
'''
Add a linear model to the current model
@type model: LinearModel
'''
self.data.extend(model.data)
rowLimits = model.getRowLimits()
columnLimits = model.getColumnLimits()
self.addRowLimits(rowLimits)
self.addColumnLimits(columnLimits)
self.mipColumns.extend(model.mipColumns)
return None
def extend(self,model):
'''
Adda linear model and update the objective function
@type model: LinearModel
'''
self.addModel(model)
self.objective = model.objective
def multiply(self,modelMatrix,value):
'''
Multiply the values of the model by a scalar
@type value: float
'''
result = LinearModel()
result.rowCache.extend(modelMatrix.rowCache)
result.columnCache.extend(modelMatrix.columnCache)
result.data.extend(modelMatrix.data.multiply(value))
rowLimits = modelMatrix.getRowLimits()
columnLimits = modelMatrix.getColumnLimits()
result.addRowLimits(rowLimits)
result.addColumnLimits(columnLimits)
def transpose(self):
'''
Get the transpose of the current model
@rtype: LinearModel
'''
result = LinearModel()
result.scale=self.scale
result.rowCache=self.columnCache
result.columnCache=self.rowCache
result.data = self.data.getTranspose()
return result
def getIndexMatrix(self):
'''
Returns the index matrix of the model
@rtype: (int,int,float)[]
'''
return self.data.getIndexMatrix()
def _reportRow(self,dataValues,limit,predValues=None):
result = ""
for (k,v) in dataValues.items():
p = None
if predValues != None:
if k in predValues:
p = predValues[k]
r = "%s(%s)[%s] + " %(v,p,k)
result += r
result = result[:-2]
result += " = (%s,%s)" % (limit[0],limit[1])
return result
def _vectorValue(self,v1,v2):
result = 0
for k in set(v1.keys()).intersection(v2.keys()):
result += v1[k] * v2[k]
return result
def floatLimit(self,limit):
if limit[0] != None:
r1 = limit[0]
else:
r1 = float("-inf")
if limit[1] != None:
r2 = limit[1]
else:
r2 = float("inf")
result = (r1,r2)
return result
def modelReport(self,dir=1,prediction=None):
report = Report()
delta = 1e-6
for rowName in self.getRowNames():
rLimit = self.floatLimit(self.getRowLimit(rowName))
rValues =self.getRowValueMap(rowName)
rString = self._reportRow(rValues,rLimit,prediction)
report.addElement(rowName,"Type","row")
report.addElement(rowName,"Equation",rString)
rLimitS = "(%s,%s)" % (rLimit[0],rLimit[1])
report.addElement(rowName,"Limit",rLimitS)
if prediction != None:
rValue = self._vectorValue(rValues, prediction)
rValue = round(rValue,6)
rValid = rLimit[0]-delta < rValue < rLimit[1] + delta
report.addElement(rowName,"Value",rValue)
report.addElement(rowName,"Valid",rValid)
for colName in self.getColumnNames():
cLimit = self.floatLimit(self.getColumnLimit(colName))
cValues =self.getColumnValueMap(colName)
cString = self._reportRow(cValues,cLimit,prediction)
report.addElement(colName,"Type","column")
report.addElement(colName,"Equation",cString)
if prediction != None:
if colName in prediction.keys():
cValue = prediction[colName]
cValid = cLimit[0]-delta < cValue < cLimit[1] + delta
report.addElement(colName,"Value",cValue)
report.addElement(colName,"Valid",cValid)
return report
def _convertToGeneNames(self,reactionName):
if reactionName not in self.controlMap.keys():
return None
geneNames = self.controlMap[reactionName]
if type(geneNames) == type(""):
return set([geneNames])
else:
return geneNames
def _convertToGeneTag(self,reactionName,geneClusters = None):
geneNames = self._convertToGeneNames(reactionName)
if geneNames == None:
return reactionName
iGeneNames = set(geneNames)
for geneName in geneNames:
if geneClusters != None:
if geneName in geneClusters.keys():
for iGeneName in geneClusters[geneName]:
iGeneNames.add(iGeneName)
geneTag = ''
for iGeneName in iGeneNames:
geneTag = geneTag + " " + iGeneName
geneTag = geneTag[1:]
return geneTag
def getEnzymeControlMap(self):
result = {}
for (key,values) in self.controlMap.items():
for v in values:
if v not in result.keys():
result[v] = set()
result[v].add(key)
return result
def getControlsForNames(self,variableNames):
result = set()
for name in variableNames:
geneNames = self._convertToGeneNames(name)
if geneNames != None:
for gName in geneNames:
result.add(gName)
return result
def printGeneObjective(self,iObjective,geneClusters=None):
iGeneObjective = {}
iOtherObjective = {}
igControl = {}
for rxnName in iObjective.keys():
rControl = iObjective[rxnName]
if rxnName not in self.controlMap.keys():
iOtherObjective[rxnName] = rControl
else:
geneTag = self._convertToGeneTag(rxnName, geneClusters)
if geneTag not in iGeneObjective:
igControl[geneTag] = []
igControl[geneTag].append(rControl)
for k in igControl.keys():
iGeneObjective[k] = mean(igControl[k])
return (iGeneObjective,iOtherObjective)
class Report:
"""
Creates a 2nd matrix of data.
Largely designed for reading and writing from files.
"""
def __init__(self, blank='N/A'):
self.data = SecondOrderCache()
self.blank = blank
def __setitem__(self,key,value):
self.addColumnHash(key,value)
def __getitem__(self,key):
result = {}
for k in self.data.keys():
if k[1] == key:
value = self.data[k]
result[k[0]] = value
return result
def setEmpty(self,blank):
"""
@param blank: the value to be used for blank fields
@type blank: string
self.blank = blank
"""
self.blank = blank
def add(self,rowName,columnName,value):
"""
@param rowName: name of row
@type rowName: string
@param columnName: name of column
@type columnName: string
@param value: the value to be added
@type value: string
"""
self.data.addValue(rowName,columnName,str(value))
return None
def addElement(self,rowName,columnName,value):
return self.add(rowName,columnName,value)
def get(self,rowName,columnName):
"""
@param rowName: name of row
@type rowName: string
@param columnName: name of column
@type columnName: string
@param value: the value to be added
@type value: string
"""
result = self.data.getValue(rowName,columnName)
return result
def getElement(self,rowName,columnName):
return self.get(rowName,columnName)
def addPairList(self,data):
for ((rowName,columnName),value) in data.items():
self.addElement(rowName,columnName,value)
return None
def addColumnHash(self,columnName,columnMap):
"""
@param columnName: name of column to be added
@type columnName: string
@param columnMap: values to be added
@type columnMap: dict
"""
if columnMap == None:
return None
for rowName in columnMap.keys():
value = columnMap[rowName]
if type(value) != type(""):
value = str(value)
self.data.addValue(rowName,columnName,columnMap[rowName])
return None
def extend(self,report):
"""
@param report: report to be appendent to this one
@type report: Report
"""
self.data.extend(report.data)
def getRowNames(self):
"""
@return: the names of all the rows in the report
@rtype: string[]
"""
return self.data.getRowKeys()
def returnRowNames(self):
return self.getRowNames()
def getColumnNames(self):
"""
@return: the names of all the columns in the report
@rtype: string[]
"""
return self.data.getColumnKeys()
def returnColumnNames(self):
return self.getColumnNames()
def returnRowArray(self,rowName):
"""
@param rowName: the name of the row of interest
@type rowName: string
@return: the values of the row in order from the report
@rtype: string[]
"""
columnNames = self.returnColumnNames()
result = []
for name in columnNames:
value = self.data.getValue(rowName,name)
if value == None:
result.append(self.blank)
else:
result.append(value)
return result
def getRow(self,name):
'''
@param name: name of column to return
@return: the values of column of report
@rtype: {string:string}
'''
result = {}
if name not in self.getRowNames():
return None
for columnName in self.getColumnNames():
value = self.data.getValue(name,columnName)
if value == None:
continue
else:
result[columnName] = value
return result
def getColumn(self,name):
'''
@param name: name of column to return
@return: the values of column of report
@rtype: {string:string}
'''
result = {}
if name not in self.getColumnNames():
return None
for rowName in self.getRowNames():
value = self.data.getValue(rowName,name)
if value == None:
continue
else:
result[rowName] = value
return result