use of dr.oldevomodel.substmodel.FrequencyModel in project beast-mcmc by beast-dev.
the class PMDTestProblem method testPMD.
public void testPMD() throws Exception {
Parameter popSize = new Parameter.Default(ConstantPopulationModelParser.POPULATION_SIZE, 496432.69917113904, 0, Double.POSITIVE_INFINITY);
ConstantPopulationModel constantModel = createRandomInitialTree(popSize);
CoalescentLikelihood coalescent = new CoalescentLikelihood(treeModel, null, new ArrayList<TaxonList>(), constantModel);
coalescent.setId("coalescent");
// clock model
Parameter rateParameter = new Parameter.Default(StrictClockBranchRates.RATE, 4.0E-7, 0, 100.0);
StrictClockBranchRates branchRateModel = new StrictClockBranchRates(rateParameter);
// Sub model
Parameter freqs = new Parameter.Default(new double[] { 0.25, 0.25, 0.25, 0.25 });
Parameter kappa = new Parameter.Default(HKYParser.KAPPA, 1.0, 1.0E-8, Double.POSITIVE_INFINITY);
FrequencyModel f = new FrequencyModel(Nucleotides.INSTANCE, freqs);
HKY hky = new HKY(kappa, f);
//siteModel
GammaSiteModel siteModel = new GammaSiteModel(hky);
Parameter mu = new Parameter.Default(GammaSiteModelParser.MUTATION_RATE, 1.0, 0, Double.POSITIVE_INFINITY);
siteModel.setMutationRateParameter(mu);
// SequenceErrorModel
Parameter ageRelatedRateParameter = new Parameter.Default(SequenceErrorModelParser.AGE_RELATED_RATE, 4.0E-7, 0, 100.0);
TipStatesModel aDNADamageModel = new SequenceErrorModel(null, null, SequenceErrorModel.ErrorType.TRANSITIONS_ONLY, null, ageRelatedRateParameter, null);
//treeLikelihood
SitePatterns patterns = new SitePatterns(alignment, null, 0, -1, 1, true);
TreeLikelihood treeLikelihood = new TreeLikelihood(patterns, treeModel, siteModel, branchRateModel, aDNADamageModel, false, false, true, false, false);
treeLikelihood.setId(TreeLikelihoodParser.TREE_LIKELIHOOD);
// Operators
OperatorSchedule schedule = new SimpleOperatorSchedule();
MCMCOperator operator = new ScaleOperator(kappa, 0.75);
operator.setWeight(1.0);
schedule.addOperator(operator);
operator = new ScaleOperator(rateParameter, 0.75);
operator.setWeight(3.0);
schedule.addOperator(operator);
Parameter allInternalHeights = treeModel.createNodeHeightsParameter(true, true, false);
operator = new UpDownOperator(new Scalable[] { new Scalable.Default(rateParameter) }, new Scalable[] { new Scalable.Default(allInternalHeights) }, 0.75, 3.0, CoercionMode.COERCION_ON);
schedule.addOperator(operator);
operator = new ScaleOperator(popSize, 0.75);
operator.setWeight(3.0);
schedule.addOperator(operator);
operator = new ScaleOperator(ageRelatedRateParameter, 0.75);
operator.setWeight(3.0);
schedule.addOperator(operator);
Parameter rootHeight = treeModel.getRootHeightParameter();
rootHeight.setId(TREE_HEIGHT);
operator = new ScaleOperator(rootHeight, 0.75);
operator.setWeight(3.0);
schedule.addOperator(operator);
Parameter internalHeights = treeModel.createNodeHeightsParameter(false, true, false);
operator = new UniformOperator(internalHeights, 30.0);
schedule.addOperator(operator);
operator = new SubtreeSlideOperator(treeModel, 15.0, 49643.2699171139, true, false, false, false, CoercionMode.COERCION_ON);
schedule.addOperator(operator);
operator = new ExchangeOperator(ExchangeOperator.NARROW, treeModel, 15.0);
// operator.doOperation();
schedule.addOperator(operator);
operator = new ExchangeOperator(ExchangeOperator.WIDE, treeModel, 3.0);
// operator.doOperation();
schedule.addOperator(operator);
operator = new WilsonBalding(treeModel, 3.0);
// operator.doOperation();
schedule.addOperator(operator);
// ??? correct?
operator = new DeltaExchangeOperator(freqs, new int[] { 1, 1, 1, 1 }, 0.01, 1.0, false, CoercionMode.COERCION_ON);
schedule.addOperator(operator);
//CompoundLikelihood
OneOnXPrior likelihood1 = new OneOnXPrior();
likelihood1.addData(popSize);
OneOnXPrior likelihood2 = new OneOnXPrior();
likelihood2.addData(kappa);
List<Likelihood> likelihoods = new ArrayList<Likelihood>();
likelihoods.add(likelihood1);
likelihoods.add(likelihood2);
likelihoods.add(coalescent);
Likelihood prior = new CompoundLikelihood(0, likelihoods);
prior.setId(CompoundLikelihoodParser.PRIOR);
likelihoods.clear();
likelihoods.add(treeLikelihood);
Likelihood likelihood = new CompoundLikelihood(-1, likelihoods);
likelihoods.clear();
likelihoods.add(prior);
likelihoods.add(likelihood);
Likelihood posterior = new CompoundLikelihood(0, likelihoods);
posterior.setId(CompoundLikelihoodParser.POSTERIOR);
// Log
ArrayLogFormatter formatter = new ArrayLogFormatter(false);
MCLogger[] loggers = new MCLogger[2];
loggers[0] = new MCLogger(formatter, 1000, false);
loggers[0].add(posterior);
loggers[0].add(treeLikelihood);
loggers[0].add(rootHeight);
loggers[0].add(rateParameter);
loggers[0].add(ageRelatedRateParameter);
loggers[0].add(popSize);
loggers[0].add(kappa);
loggers[0].add(coalescent);
loggers[1] = new MCLogger(new TabDelimitedFormatter(System.out), 10000, false);
loggers[1].add(posterior);
loggers[1].add(treeLikelihood);
loggers[1].add(rootHeight);
loggers[1].add(rateParameter);
// MCMC
MCMC mcmc = new MCMC("mcmc1");
MCMCOptions options = new MCMCOptions(1000000);
mcmc.setShowOperatorAnalysis(true);
mcmc.init(options, posterior, schedule, loggers);
mcmc.run();
// time
System.out.println(mcmc.getTimer().toString());
// Tracer
List<Trace> traces = formatter.getTraces();
ArrayTraceList traceList = new ArrayTraceList("PMDTest", traces, 0);
for (int i = 1; i < traces.size(); i++) {
traceList.analyseTrace(i);
}
// <expectation name="clock.rate" value="1.5E-7"/>
// <expectation name="errorModel.ageRate" value="0.7E-7"/>
// <expectation name="hky.kappa" value="10"/>
TraceCorrelation kappaStats = traceList.getCorrelationStatistics(traceList.getTraceIndex(HKYParser.KAPPA));
assertExpectation(HKYParser.KAPPA, kappaStats, 10);
TraceCorrelation rateStats = traceList.getCorrelationStatistics(traceList.getTraceIndex(StrictClockBranchRates.RATE));
assertExpectation(StrictClockBranchRates.RATE, rateStats, 1.5E-7);
TraceCorrelation ageRateStats = traceList.getCorrelationStatistics(traceList.getTraceIndex(SequenceErrorModelParser.AGE_RELATED_RATE));
assertExpectation(SequenceErrorModelParser.AGE_RELATED_RATE, ageRateStats, 0.7E-7);
}
use of dr.oldevomodel.substmodel.FrequencyModel in project beast-mcmc by beast-dev.
the class TimeIrreversibleTest method testComplexSubstitutionModel.
private double[] testComplexSubstitutionModel(Original test, double[] rates) {
System.out.println("\n*** Complex Substitution Model Test: " + test + " ***");
Parameter ratesP = new Parameter.Default(rates);
DataType dataType = test.getDataType();
FrequencyModel freqModel = new FrequencyModel(dataType, new Parameter.Default(test.getFrequencies()));
ComplexSubstitutionModel substModel = new ComplexSubstitutionModel("Complex Substitution Model Test", dataType, freqModel, ratesP);
double logL = substModel.getLogLikelihood();
System.out.println("Prior = " + logL);
double[] finiteTimeProbs = null;
if (!Double.isInfinite(logL)) {
finiteTimeProbs = new double[substModel.getDataType().getStateCount() * substModel.getDataType().getStateCount()];
substModel.getTransitionProbabilities(time, finiteTimeProbs);
System.out.println("Probs = ");
printRateMatrix(finiteTimeProbs, substModel.getDataType().getStateCount());
}
// assertEquals(1, 1, 1e-10);
return finiteTimeProbs;
}
use of dr.oldevomodel.substmodel.FrequencyModel in project beast-mcmc by beast-dev.
the class TimeIrreversibleTest method testSVSComplexSubstitutionModel.
private double[] testSVSComplexSubstitutionModel(Original test, double[] rates) {
System.out.println("\n*** SVS Complex Substitution Model Test: " + test + " ***");
double[] indicators = test.getIndicators();
Parameter ratesP = new Parameter.Default(rates);
Parameter indicatorsP = new Parameter.Default(indicators);
DataType dataType = test.getDataType();
FrequencyModel freqModel = new FrequencyModel(dataType, new Parameter.Default(test.getFrequencies()));
SVSComplexSubstitutionModel substModel = new SVSComplexSubstitutionModel("SVS Complex Substitution Model Test", dataType, freqModel, ratesP, indicatorsP);
double logL = substModel.getLogLikelihood();
System.out.println("Prior = " + logL);
double[] finiteTimeProbs = null;
if (!Double.isInfinite(logL)) {
finiteTimeProbs = new double[substModel.getDataType().getStateCount() * substModel.getDataType().getStateCount()];
substModel.getTransitionProbabilities(time, finiteTimeProbs);
System.out.println("Probs = ");
printRateMatrix(finiteTimeProbs, substModel.getDataType().getStateCount());
}
// assertEquals(1, 1, 1e-10);
return finiteTimeProbs;
}
use of dr.oldevomodel.substmodel.FrequencyModel in project beast-mcmc by beast-dev.
the class TwoStateCovarionModelTest method setUp.
public void setUp() throws Exception {
super.setUp();
frequencies = new Parameter.Default(new double[] { 0.25, 0.25, 0.25, 0.25 });
alpha = new Parameter.Default(0.0);
switchingRate = new Parameter.Default(1.0);
FrequencyModel freqModel = new FrequencyModel(TwoStateCovarion.INSTANCE, frequencies);
model = new TwoStateCovarionModel(TwoStateCovarion.INSTANCE, freqModel, alpha, switchingRate);
dataType = model.getDataType();
}
use of dr.oldevomodel.substmodel.FrequencyModel in project beast-mcmc by beast-dev.
the class testNtdBMA method testNtdBMA.
public void testNtdBMA() {
for (Instance test : all) {
Parameter logKappa = new Parameter.Default(1, test.getLogKappa());
Parameter logTN = new Parameter.Default(1, test.getLogTN());
Parameter logAC = new Parameter.Default(1, test.getLogAC());
Parameter logAT = new Parameter.Default(1, test.getLogAT());
Parameter logGC = new Parameter.Default(1, test.getLogGC());
Parameter logGT = new Parameter.Default(1, test.getLogGT());
Variable<Integer> modelChoose = test.getModelChoose();
double[] pi = test.getPi();
Parameter freqs = new Parameter.Default(pi);
FrequencyModel f = new FrequencyModel(Nucleotides.INSTANCE, freqs);
NtdBMA ntdBMA = new NtdBMA(logKappa, logTN, logAC, logAT, logGC, logGT, modelChoose, f);
double distance = test.getDistance();
double[] mat = new double[4 * 4];
ntdBMA.getTransitionProbabilities(distance, mat);
final double[] result = test.getExpectedResult();
for (int k = 0; k < mat.length; ++k) {
assertEquals(mat[k], result[k], 5e-10);
// System.out.print(" " + (mat[k] - result[k]));
}
}
}
Aggregations