def test_dtype_float32(self): """A BulkObjectsContainer with dtype=np.float32 should support fractional counts and deltas. """ container = BulkObjectsContainer(OBJECT_NAMES, dtype=np.float32) initialCounts = [10, 10.5, 20] container.countsIs(initialCounts) npt.assert_equal(container.counts(), initialCounts) incCounts = [10, 20.5, 30.5] newCounts = [20, 31, 50.5] container.countsInc(incCounts) npt.assert_equal(container.counts(), newCounts) decCounts = [1.5, 2, 3.5] newCounts = [18.5, 29, 47] container.countsDec(decCounts) npt.assert_equal(container.counts(), newCounts) countsView = container.countsView() newCounts = [28.5, 49.5, 77.5] countsView.countsInc(incCounts) npt.assert_equal(countsView.counts(), newCounts) newCounts = [27, 47.5, 74] countsView.countsDec(decCounts) npt.assert_equal(countsView.counts(), newCounts)
def fitSimData_2(kb, simOutDir):
subMass = kb.mass.subMass
proteinMass = subMass["proteinMass"].asUnit(units.g)
rnaMass = subMass["rnaMass"].asUnit(units.g)
# Construct bulk container
# We want to know something about the distribution of the copy numbers of
# macromolecules in the cell. While RNA and protein expression can be
# approximated using well-described statistical distributions, we need
# absolute copy numbers to form complexes. To get a distribution, we must
# instantiate many cells, form complexes, and finally compute the
# statistics we will use in the fitting operations.
bulkContainer = BulkObjectsContainer(kb.state.bulkMolecules.bulkData['id'])
rnaView = bulkContainer.countsView(kb.process.transcription.rnaData["id"])
proteinView = bulkContainer.countsView(kb.process.translation.monomerData["id"])
complexationMoleculesView = bulkContainer.countsView(kb.process.complexation.moleculeNames)
allMoleculesIDs = list(
set(kb.process.transcription.rnaData["id"]) | set(kb.process.translation.monomerData["id"]) | set(kb.process.complexation.moleculeNames)
)
allMoleculesView = bulkContainer.countsView(allMoleculesIDs)
allMoleculeCounts = np.empty((N_SEEDS, allMoleculesView.counts().size), np.int64)
complexationStoichMatrix = kb.process.complexation.stoichMatrix().astype(np.int64, order = "F")
complexationPrebuiltMatrices = mccBuildMatrices(
complexationStoichMatrix
)
rnaDistribution = kb.process.transcription.rnaData["expression"]
rnaTotalCounts = countsFromMassAndExpression(
rnaMass.asNumber(units.g),
kb.process.transcription.rnaData["mw"].asNumber(units.g / units.mol),
rnaDistribution,
kb.constants.nAvogadro.asNumber(1 / units.mol)
)
proteinDistribution = calcProteinDistribution(kb)
proteinTotalCounts = calcProteinTotalCounts(kb, proteinMass, proteinDistribution)
for seed in xrange(N_SEEDS):
randomState = np.random.RandomState(seed)
allMoleculesView.countsIs(0)
rnaView.countsIs(randomState.multinomial(
rnaTotalCounts,
rnaDistribution
))
proteinView.countsIs(randomState.multinomial(
proteinTotalCounts,
proteinDistribution
))
complexationMoleculeCounts = complexationMoleculesView.counts()
updatedCompMoleculeCounts = mccFormComplexesWithPrebuiltMatrices(
complexationMoleculeCounts,
seed,
complexationStoichMatrix,
*complexationPrebuiltMatrices
)
complexationMoleculesView.countsIs(updatedCompMoleculeCounts)
allMoleculeCounts[seed, :] = allMoleculesView.counts()
bulkAverageContainer = BulkObjectsContainer(kb.state.bulkMolecules.bulkData['id'], np.float64)
bulkDeviationContainer = BulkObjectsContainer(kb.state.bulkMolecules.bulkData['id'], np.float64)
bulkAverageContainer.countsIs(allMoleculeCounts.mean(0), allMoleculesIDs)
bulkDeviationContainer.countsIs(allMoleculeCounts.std(0), allMoleculesIDs)
# Free up memory
# TODO: make this more functional; one function for returning average & distribution
del allMoleculeCounts
del bulkContainer
# ----- Calculate ppGpp concentration ----- #
aminoAcidsInProtein = (bulkAverageContainer.counts(kb.process.translation.monomerData['id']) * kb.process.translation.monomerData['length'].asNumber()).sum()
aminoAcidsInComplex = 0.
for cplx in list(kb.process.complexation.complexNames):
cplx_data = kb.process.complexation.getMonomers(cplx)
cplx_subunit = cplx_data['subunitIds']
cplx_stoich = cplx_data['subunitStoich']
subunit_idxs = []
subunit_idxs_to_delete = []
for idx, subunit in enumerate(cplx_subunit):
try:
subunit_idxs.append(np.where(kb.process.translation.monomerData['id'] == subunit)[0][0])
except IndexError:
subunit_idxs_to_delete.append(idx)
cplx_stoich = np.delete(cplx_stoich, subunit_idxs_to_delete)
subunit_length = kb.process.translation.monomerData['length'][subunit_idxs].asNumber()
aminoAcidsInComplex += (bulkAverageContainer.count(cplx) * subunit_length * cplx_stoich).sum()
totalAminoAcidsInMacromolecules = (aminoAcidsInComplex + aminoAcidsInProtein)
totalAAInSolublePool = totalAminoAcidsInMacromolecules * 0.08 # Approximatly correct for one time calculature.
# TODO: Calculate soluble pools here too!
totalAminoAcidsInCell = totalAminoAcidsInMacromolecules + totalAAInSolublePool
ppGpp_per_cell = (totalAminoAcidsInCell * kb.constants.ppGpp_base_concentration).asUnit(units.count)
cellVolume = kb.mass.avgCellDryMassInit / kb.constants.cellDensity
ppGpp_concentration = (ppGpp_per_cell.asUnit(units.mol) / cellVolume).asUnit(units.mol / units.L)
# Finally set ppGpp concentration to maintain
kb.process.metabolism.metabolitePoolConcentrations[kb.process.metabolism.metabolitePoolIDs.index('PPGPP[c]')] = ppGpp_concentration
# ----- tRNA synthetase turnover rates ------
# Fit tRNA synthetase kcat values based on expected rates of translation
# compute values at initial time point
## Compute rate of AA incorperation
proteinComposition = kb.process.translation.monomerData["aaCounts"]
initialProteinMass = kb.mass.subMass['proteinMass']
initialProteinCounts = calcProteinCounts(kb, initialProteinMass)
initialProteinTranslationRate = (
(np.log(2) / kb.doubling_time + kb.process.translation.monomerData["degRate"]) * initialProteinCounts
).asUnit(1 / units.s)
initialAAPolymerizationRate = units.dot(
units.transpose(proteinComposition), initialProteinTranslationRate
).asUnit(units.aa / units.s)
## Compute expression of tRNA synthetases
## Assuming independence in variance
synthetase_counts_by_group = np.zeros(len(kb.process.translation.AA_SYNTHETASE_GROUPS), dtype = np.float64)
synthetase_variance_by_group = np.zeros(len(kb.process.translation.AA_SYNTHETASE_GROUPS), dtype = np.float)
for idx, synthetase_group in enumerate(kb.process.translation.AA_SYNTHETASE_GROUPS.itervalues()):
group_count = 0.
group_variance = 0.
for synthetase in synthetase_group:
counts = bulkAverageContainer.countsView([synthetase]).counts()
variance = bulkDeviationContainer.countsView([synthetase]).counts()
group_count += counts
group_variance += variance
synthetase_counts_by_group[idx] = group_count
synthetase_variance_by_group[idx] = group_variance
## Saved for plotting
kb.synthetase_counts = synthetase_counts_by_group
kb.synthetase_variance = synthetase_variance_by_group
kb.initial_aa_polymerization_rate = initialAAPolymerizationRate
kb.minimum_trna_synthetase_rates = initialAAPolymerizationRate / synthetase_counts_by_group
# TODO: Reimplement this with better fit taking into account the variance in aa
# utilization.
## Scaling synthetase counts by -2*variance so that rates will be high enough
## to accomodate stochastic behavior in the model without translation stalling.
# scaled_synthetase_counts = synthetase_counts_by_group - (2 * synthetase_variance_by_group)
scaled_synthetase_counts = synthetase_counts_by_group
assert all(scaled_synthetase_counts > 0)
predicted_trna_synthetase_rates = initialAAPolymerizationRate / scaled_synthetase_counts
kb.trna_synthetase_rates = 2 * predicted_trna_synthetase_rates