def atomize(dependencyGraph, weights, translator, reactionProperties,
equivalenceDictionary, bioGridFlag, sbmlAnalyzer, database, parser):
'''
The atomizer's main methods. Receives a dependency graph
'''
redrawflag = True
loops = 0
while redrawflag and loops < 10:
loops +=1
bindingCounter = Counter()
bindingFailureDict = {}
for idx, element in enumerate(weights):
# 0 molecule
if element[0] == '0':
continue
# user defined molecules to be the zero molecule
if dependencyGraph[element[0]] == [['0']]:
zeroSpecies = st.Species()
zeroMolecule = st.Molecule('0')
zeroSpecies.addMolecule(zeroMolecule)
translator[element[0]] = zeroSpecies
continue
# undivisible molecules
elif dependencyGraph[element[0]] == []:
if element[0] not in translator:
translator[element[0]] = createEmptySpecies(element[0])
else:
if len(dependencyGraph[element[0]][0]) == 1:
# catalysis
createCatalysisRBM(dependencyGraph, element, translator, reactionProperties,
equivalenceDictionary, sbmlAnalyzer, database)
else:
try:
createBindingRBM(element, translator, dependencyGraph,
bioGridFlag, database.pathwaycommons, parser, database)
except BindingException as e:
for c in e.combinations:
bindingCounter[c] += 1
bindingFailureDict[element[0]] = e.combinations
logMess('DEBUG:ATO003', "We don't know how {0} binds together in complex {1}. Not atomizing".format(
e.value, element[0]))
# there awas an issue during binding, don't atomize
translator[element[0]] = createEmptySpecies(element[0])
# evaluate species that weren't bound properly and see if we can get information from all over the model to find the right binding partner
bindingTroubleLog = defaultdict(list)
modifiedPairs = set()
redrawflag = False
for molecule in bindingFailureDict:
bindingWinner = defaultdict(list)
for candidateTuple in bindingFailureDict[molecule]:
bindingWinner[bindingCounter[candidateTuple]].append(candidateTuple)
bestBindingCandidates = bindingWinner[max(bindingWinner.keys())]
if len(bestBindingCandidates) > 1:
bindingTroubleLog[tuple(sorted(bestBindingCandidates))].append(molecule)
else:
bindingPair = bestBindingCandidates[0]
if bindingPair not in modifiedPairs:
modifiedPairs.add(bindingPair)
else:
continue
c1 = st.Component(bindingPair[1].lower())
c2 = st.Component(bindingPair[0].lower())
molecule1 = translator[translator[bindingPair[0]].molecules[0].name].molecules[0]
molecule2 = translator[translator[bindingPair[1]].molecules[0].name].molecules[0]
molecule1.addComponent(c1)
molecule2.addComponent(c2)
redrawflag = True
logMess('INFO:ATO031','Determining that {0} binds together based on frequency of the bond in the reaction network.'.format(bindingPair))
for trouble in bindingTroubleLog:
logMess('ERROR:ATO202','{0}:{1}:We need information to resolve the bond structure of these complexes . \
Please choose among the possible binding candidates that had the most observed frequency in the reaction network or provide a new one'.format(bindingTroubleLog[trouble],trouble))
def createEmptySpecies(name):
species = st.Species()
molecule = st.Molecule(name)
species.addMolecule(molecule)
return species
def createBindingRBM(element, translator, dependencyGraph, bioGridFlag, pathwaycommonsFlag, parser, database):
species = st.Species()
# go over the sct and reuse existing stuff
for molecule in dependencyGraph[element[0]][0]:
if molecule in translator:
tmpSpecies = translator[molecule]
if molecule != getTrueTag(dependencyGraph, molecule):
original = translator[getTrueTag(dependencyGraph, molecule)]
updateSpecies(tmpSpecies, original.molecules[0])
if tmpSpecies.molecules[0].name in database.constructedSpecies:
tmpSpecies.molecules[0].trueName = molecule
else:
tmpSpecies.molecules[0].trueName = tmpSpecies.molecules[0].name
species.addMolecule(deepcopy(tmpSpecies.molecules[0]))
else:
mol = st.Molecule(molecule)
mol.trueName = molecule
#dependencyGraph[molecule] = deepcopy(mol)
species.addMolecule(mol)
dependencyGraphCounter = Counter(dependencyGraph[element[0]][0])
# XXX: this wont work for species with more than one molecule with the
# same name
changeFlag = False
partialBonds = defaultdict(list)
for partialUserEntry in database.partialUserLabelDictionary:
partialCounter = Counter(partialUserEntry)
if all([partialCounter[x] <= dependencyGraphCounter[x] for x in partialCounter]):
changeFlag = True
for molecule in database.partialUserLabelDictionary[partialUserEntry].molecules:
for molecule2 in species.molecules:
if molecule.name == molecule2.name:
for component in molecule.components:
for bond in component.bonds:
if molecule2.name not in [x.name for x in partialBonds[bond]]:
partialBonds[bond].append(molecule2)
'''
for component in molecule.components:
component2 = [x for x in molecule2.components if x.name == component.name]
# component already exists in species template
if component2:
if component.bonds:
component2[0].bonds = component.bonds
else:
molecule2.addComponent(deepcopy(component))
'''
bondSeeding = [partialBonds[x] for x in partialBonds if x > 0]
bondExclusion = [partialBonds[x] for x in partialBonds if x < 0]
# how do things bind together?
moleculePairsList = getComplexationComponents2(
element[0], species, bioGridFlag, pathwaycommonsFlag, parser, bondSeeding, bondExclusion, database)
#moleculeCount = Counter([y for x in moleculePairsList for y in x])
# print moleculeCount
#moleculePairsList = [sorted(x) for x in moleculePairsList]
#moleculePairsList.sort(key=lambda x: [-moleculeCount[x[0]],(str(x[0]), x[0],str(x[1]),x[1])])
# TODO: update basic molecules with new components
# translator[molecule[0].name].molecules[0].components.append(deepcopy(newComponent1))
# translator[molecule[1].name].molecules[0].components.append(deepcopy(newComponent2))
moleculeCounter = defaultdict(list)
for molecule in moleculePairsList:
flag = False
# create an index on m0 and m1 depending on their name and repeats in
# the species
if molecule[0] not in moleculeCounter[molecule[0].name]:
moleculeCounter[molecule[0].name].append(molecule[0])
if molecule[1] not in moleculeCounter[molecule[1].name]:
moleculeCounter[molecule[1].name].append(molecule[1])
m0index = moleculeCounter[molecule[0].name].index(molecule[0])
m1index = moleculeCounter[molecule[1].name].index(molecule[1])
bondIdx = getBondNumber('{0}{1}'.format(
molecule[0].name, m0index), '{0}{1}'.format(molecule[1].name, m1index))
# add bonds where binding components already exist and they are not
# occupied
for component in molecule[0].components:
if component.name == molecule[1].name.lower() and \
len(component.bonds) == 0:
component.bonds.append(bondIdx)
flag = True
break
if not flag:
# create components if they dont exist already.
# Add a bond afterwards
newComponent1 = st.Component(molecule[1].name.lower())
molecule[0].components.append(newComponent1)
if newComponent1.name not in [x.name for x in translator[molecule[0].name].molecules[0].
components]:
translator[molecule[0].name].molecules[
0].components.append(deepcopy(newComponent1))
molecule[0].components[-1].bonds.append(bondIdx)
flag = False
# same thing for the other member of the bond
for component in molecule[1].components:
if component.name == molecule[0].name.lower() and len(component.bonds) == 0:
component.bonds.append(bondIdx)
flag = True
break
if not flag:
newComponent2 = st.Component(molecule[0].name.lower())
molecule[1].components.append(newComponent2)
if molecule[0].name != molecule[1].name:
if newComponent2.name not in [x.name for x in translator[molecule[1].name].molecules[0].
components]:
translator[molecule[1].name].molecules[
0].components.append(deepcopy(newComponent2))
molecule[1].components[-1].bonds.append(bondIdx)
# update the translator
translator[element[0]] = species