/* * DiscreteTraitsComponentGenerator.java * * Copyright (c) 2002-2015 Alexei Drummond, Andrew Rambaut and Marc Suchard * * This file is part of BEAST. * See the NOTICE file distributed with this work for additional * information regarding copyright ownership and licensing. * * BEAST is free software; you can redistribute it and/or modify * it under the terms of the GNU Lesser General Public License as * published by the Free Software Foundation; either version 2 * of the License, or (at your option) any later version. * * BEAST is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU Lesser General Public License for more details. * * You should have received a copy of the GNU Lesser General Public * License along with BEAST; if not, write to the * Free Software Foundation, Inc., 51 Franklin St, Fifth Floor, * Boston, MA 02110-1301 USA */ package dr.app.beauti.components.discrete; import dr.evomodelxml.treelikelihood.MarkovJumpsTreeLikelihoodParser; import dr.app.beauti.components.ancestralstates.AncestralStatesComponentOptions; import dr.app.beauti.generator.BaseComponentGenerator; import dr.app.beauti.generator.ClockModelGenerator; import dr.app.beauti.generator.ComponentGenerator; import dr.app.beauti.options.*; import dr.app.beauti.util.XMLWriter; import dr.evolution.datatype.GeneralDataType; import dr.inference.distribution.BinomialLikelihood; import dr.inference.model.DesignMatrix; import dr.inference.operators.MultivariateNormalOperator; import dr.inferencexml.distribution.BinomialLikelihoodParser; import dr.inferencexml.distribution.GeneralizedLinearModelParser; import dr.oldevomodel.sitemodel.SiteModel; import dr.oldevomodel.substmodel.AbstractSubstitutionModel; import dr.evomodel.tree.TreeModel; import dr.oldevomodelxml.sitemodel.GammaSiteModelParser; import dr.oldevomodelxml.substmodel.ComplexSubstitutionModelParser; import dr.oldevomodelxml.substmodel.FrequencyModelParser; import dr.oldevomodelxml.substmodel.GLMSubstitutionModelParser; import dr.oldevomodelxml.substmodel.GeneralSubstitutionModelParser; import dr.oldevomodelxml.treelikelihood.AncestralStateTreeLikelihoodParser; import dr.oldevomodelxml.treelikelihood.TreeLikelihoodParser; import dr.evoxml.AttributePatternsParser; import dr.evoxml.GeneralDataTypeParser; import dr.evoxml.TaxaParser; import dr.inference.model.ParameterParser; import dr.inferencexml.loggers.ColumnsParser; import dr.inferencexml.loggers.LoggerParser; import dr.inferencexml.model.ProductStatisticParser; import dr.inferencexml.model.SumStatisticParser; import dr.util.Attribute; import dr.xml.XMLParser; import java.util.Set; import static dr.evomodelxml.substmodel.ComplexSubstitutionModelParser.ROOT_FREQUENCIES; /** * @author Andrew Rambaut * @version $Id$ */ public class DiscreteTraitsComponentGenerator extends BaseComponentGenerator { public DiscreteTraitsComponentGenerator(final BeautiOptions options) { super(options); } @Override public void checkOptions() throws GeneratorException { super.checkOptions(); for (PartitionSubstitutionModel model : options.getPartitionSubstitutionModels(GeneralDataType.INSTANCE)) { if (model.getDiscreteSubstType() == DiscreteSubstModelType.GLM_SUBST) { if (model.getTraitData().getIncludedPredictors().size() < 1) { throw new GeneratorException("The GLM model for trait, " + model.getTraitData().getName() + ", has no predictors included."); } for (Predictor predictor : model.getTraitData().getIncludedPredictors()) { if (predictor.isLogged() && predictor.hasZeroValues(true)) { throw new GeneratorException("The GLM predictor, " + predictor.getName() + ", for trait, " + model.getTraitData().getName() + ", has zero values and therefore should not be logged."); } if (predictor.isStandardized() && predictor.isBinary()) { throw new GeneratorException("The GLM predictor, " + predictor.getName() + ", for trait, " + model.getTraitData().getName() + ", is binary and therefore should not be standardized."); } } } } } public boolean usesInsertionPoint(final InsertionPoint point) { if (options.getDataPartitions(GeneralDataType.INSTANCE).size() == 0) { // Empty, so do nothing return false; } boolean hasGLM = false; for (PartitionSubstitutionModel model : options.getPartitionSubstitutionModels(GeneralDataType.INSTANCE)) { if (model.getDiscreteSubstType() == DiscreteSubstModelType.GLM_SUBST) { hasGLM = true; } } switch (point) { case AFTER_PATTERNS: case AFTER_TREE_LIKELIHOOD: case IN_MCMC_LIKELIHOOD: case IN_SCREEN_LOG: case IN_FILE_LOG_PARAMETERS: case IN_FILE_LOG_LIKELIHOODS: case AFTER_FILE_LOG: return true; case IN_OPERATORS: return hasGLM; case IN_MCMC_PRIOR: return hasGLM || hasBSSVS(); default: return false; } } protected void generate(final InsertionPoint point, final Object item, String prefix, final XMLWriter writer) { DiscreteTraitsComponentOptions comp = (DiscreteTraitsComponentOptions)options.getComponentOptions(DiscreteTraitsComponentOptions.class); switch (point) { case AFTER_PATTERNS: writeDiscreteTraitPatterns(writer, comp); break; case AFTER_TREE_LIKELIHOOD: writeDiscreteTraitsModels(writer); writeTreeLikelihoods(writer, comp); break; case IN_OPERATORS: for (PartitionSubstitutionModel model : options.getPartitionSubstitutionModels(GeneralDataType.INSTANCE)) { if (model.getDiscreteSubstType() == DiscreteSubstModelType.GLM_SUBST) { writeGLMCoefficientOperator(model, writer); } } break; case IN_MCMC_PRIOR: for (PartitionSubstitutionModel model : options.getPartitionSubstitutionModels(GeneralDataType.INSTANCE)) { if (model.getDiscreteSubstType() == DiscreteSubstModelType.GLM_SUBST) { writeGLMBinomialLikelihood(model, writer); } } writeDiscreteTraitsSubstitutionModelReferences(writer); break; case IN_MCMC_LIKELIHOOD: writeTreeLikelihoodReferences(writer); break; case IN_SCREEN_LOG: writeScreenLogEntries(writer); break; case IN_FILE_LOG_PARAMETERS: writeFileLogEntries(writer); break; case IN_FILE_LOG_LIKELIHOODS: writeTreeLikelihoodReferences(writer); break; case AFTER_FILE_LOG: writeDiscreteTraitFileLoggers(writer); break; default: throw new IllegalArgumentException("This insertion point is not implemented for " + this.getClass().getName()); } } protected String getCommentLabel() { return "Discrete Traits Model"; } private void writeDiscreteTraitPatterns(XMLWriter writer, DiscreteTraitsComponentOptions component) { boolean first = true; for (PartitionSubstitutionModel model : options.getPartitionSubstitutionModels(GeneralDataType.INSTANCE)) { if (!first) { writer.writeBlankLine(); } else { first = false; } writeGeneralDataType(model, writer); } // now create an attribute pattern for each trait that uses it for (AbstractPartitionData partition : options.getDataPartitions(GeneralDataType.INSTANCE)) { if (partition.getTraits() != null) { writer.writeBlankLine(); writeAttributePatterns(partition, writer); } } } private void writeGeneralDataType(PartitionSubstitutionModel model, XMLWriter writer) { writer.writeComment("general data type for discrete trait model, '" + model.getName() + "'"); Set states = options.getStatesForDiscreteModel(model); String prefix = model.getName() + "."; // writer.writeOpenTag(GeneralDataTypeParser.GENERAL_DATA_TYPE, new Attribute[]{ new Attribute.Default(XMLParser.ID, prefix + "dataType")}); int numOfStates = states.size(); writer.writeComment("Number Of States = " + numOfStates); for (String eachGD : states) { writer.writeTag(GeneralDataTypeParser.STATE, new Attribute[]{ new Attribute.Default(GeneralDataTypeParser.CODE, eachGD)}, true); } writer.writeCloseTag(GeneralDataTypeParser.GENERAL_DATA_TYPE); } /** * write * * @param partition PartitionData * @param writer XMLWriter */ private void writeAttributePatterns(AbstractPartitionData partition, XMLWriter writer) { String traitName = partition.getTraits().get(0).getName(); writer.writeComment("Data pattern for discrete trait, '" + traitName + "'"); String prefix = partition.getName() + "."; // writer.writeOpenTag(AttributePatternsParser.ATTRIBUTE_PATTERNS, new Attribute[]{ new Attribute.Default(XMLParser.ID, prefix + "pattern"), new Attribute.Default(AttributePatternsParser.ATTRIBUTE, traitName)}); String prefix2 = ""; if (!options.hasIdenticalTaxa()) { prefix2 = partition.getPartitionTreeModel().getPrefix(); } writer.writeIDref(TaxaParser.TAXA, prefix2 + TaxaParser.TAXA); writer.writeIDref(GeneralDataTypeParser.GENERAL_DATA_TYPE, partition.getPartitionSubstitutionModel().getName() + ".dataType"); writer.writeCloseTag(AttributePatternsParser.ATTRIBUTE_PATTERNS); } private void writeDiscreteTraitsModels(XMLWriter writer) { for (PartitionSubstitutionModel model : options.getPartitionSubstitutionModels(GeneralDataType.INSTANCE)) { writeDiscreteTraitsSubstitutionModel(model, writer); writeDiscreteTraitsSiteModel(model, writer); } } private void writeDiscreteTraitsSubstitutionModel(PartitionSubstitutionModel model, XMLWriter writer) { int stateCount = options.getStatesForDiscreteModel(model).size(); String prefix = model.getName() + "."; if (model.getDiscreteSubstType() == DiscreteSubstModelType.SYM_SUBST) { writer.writeComment("symmetric CTMC model for discrete state reconstructions"); writer.writeOpenTag(GeneralSubstitutionModelParser.GENERAL_SUBSTITUTION_MODEL, new Attribute[]{ new Attribute.Default(XMLParser.ID, prefix + AbstractSubstitutionModel.MODEL)}); writer.writeIDref(GeneralDataTypeParser.GENERAL_DATA_TYPE, prefix + "dataType"); writer.writeOpenTag(GeneralSubstitutionModelParser.FREQUENCIES); writeDiscreteFrequencyModel(prefix, prefix, stateCount, true, writer); writer.writeCloseTag(GeneralSubstitutionModelParser.FREQUENCIES); //---------------- rates and indicators ----------------- writeRatesAndIndicators(model, stateCount * (stateCount - 1) / 2, null, writer); writer.writeCloseTag(GeneralSubstitutionModelParser.GENERAL_SUBSTITUTION_MODEL); } else if (model.getDiscreteSubstType() == DiscreteSubstModelType.ASYM_SUBST) { writer.writeComment("asymmetric CTMC model for discrete state reconstructions"); writer.writeOpenTag(GeneralSubstitutionModelParser.GENERAL_SUBSTITUTION_MODEL, new Attribute[]{ new Attribute.Default(XMLParser.ID, prefix + AbstractSubstitutionModel.MODEL), new Attribute.Default(ComplexSubstitutionModelParser.RANDOMIZE, false)}); writer.writeIDref(GeneralDataTypeParser.GENERAL_DATA_TYPE, prefix + "dataType"); writer.writeOpenTag(GeneralSubstitutionModelParser.FREQUENCIES); writeDiscreteFrequencyModel(prefix, prefix, stateCount, true, writer); writer.writeCloseTag(GeneralSubstitutionModelParser.FREQUENCIES); //---------------- rates and indicators ----------------- writeRatesAndIndicators(model, stateCount * (stateCount - 1), null, writer); writer.writeCloseTag(GeneralSubstitutionModelParser.GENERAL_SUBSTITUTION_MODEL); } else if (model.getDiscreteSubstType() == DiscreteSubstModelType.GLM_SUBST) { writer.writeComment("GLM substitution model"); writer.writeOpenTag(GLMSubstitutionModelParser.GLM_SUBSTITUTION_MODEL, new Attribute[] { new Attribute.Default(XMLParser.ID, prefix + AbstractSubstitutionModel.MODEL) }); writer.writeIDref(GeneralDataTypeParser.GENERAL_DATA_TYPE, prefix + "dataType"); writer.writeOpenTag(ROOT_FREQUENCIES); writeDiscreteFrequencyModel(prefix, prefix, stateCount, true, writer); writer.writeCloseTag(ROOT_FREQUENCIES); writer.writeOpenTag(GeneralizedLinearModelParser.GLM_LIKELIHOOD, new Attribute[] { new Attribute.Default("family", "logLinear"), new Attribute.Default("checkIdentifiability", "true") }); writer.writeOpenTag(GeneralizedLinearModelParser.INDEPENDENT_VARIABLES); writeParameter(options.getParameter(prefix + "coefficients"), 1, writer); writeParameter(GeneralizedLinearModelParser.INDICATOR, prefix + "coefIndicators", 1, writer); writer.writeOpenTag(DesignMatrix.DESIGN_MATRIX, new Attribute[] { new Attribute.Default("id", prefix + DesignMatrix.DESIGN_MATRIX) } ); for (Predictor predictor : model.getTraitData().getIncludedPredictors()) { if (predictor.getType() == Predictor.Type.ORIGIN_VECTOR || predictor.getType() == Predictor.Type.BOTH_VECTOR) { writer.writeTag(ParameterParser.PARAMETER, new Attribute[]{ new Attribute.Default(XMLParser.ID, prefix + predictor + "_origin"), new Attribute.Default(ParameterParser.VALUE, predictor.getValueString(Predictor.Type.ORIGIN_VECTOR)) }, true); } if (predictor.getType() == Predictor.Type.DESTINATION_VECTOR || predictor.getType() == Predictor.Type.BOTH_VECTOR) { writer.writeTag(ParameterParser.PARAMETER, new Attribute[]{ new Attribute.Default(XMLParser.ID, prefix + predictor + "_destination"), new Attribute.Default(ParameterParser.VALUE, predictor.getValueString(Predictor.Type.DESTINATION_VECTOR)) }, true); } if (predictor.getType() == Predictor.Type.MATRIX) { writer.writeTag(ParameterParser.PARAMETER, new Attribute[]{ new Attribute.Default(XMLParser.ID, prefix + predictor), new Attribute.Default(ParameterParser.VALUE, predictor.getValueString(Predictor.Type.MATRIX)) }, true); } } writer.writeCloseTag(DesignMatrix.DESIGN_MATRIX); writer.writeCloseTag(GeneralizedLinearModelParser.INDEPENDENT_VARIABLES); writer.writeCloseTag(GeneralizedLinearModelParser.GLM_LIKELIHOOD); writer.writeCloseTag(GLMSubstitutionModelParser.GLM_SUBSTITUTION_MODEL); writer.writeComment("GLM: statistic that returns the product of the coefficients and the respective indicators for the predictors"); writer.writeOpenTag(ProductStatisticParser.PRODUCT_STATISTIC, new Attribute[] { new Attribute.Default(XMLParser.ID, prefix + "coefficientsTimesIndicators") }); writer.writeTag(ParameterParser.PARAMETER, new Attribute[]{ new Attribute.Default(XMLParser.IDREF, prefix + "coefficients") }, true); writer.writeTag(ParameterParser.PARAMETER, new Attribute[]{ new Attribute.Default(XMLParser.IDREF, prefix + "coefIndicators") }, true); writer.writeCloseTag(ProductStatisticParser.PRODUCT_STATISTIC); writer.writeComment("GLM: the number of predictors included in the model (non-zero indicators)"); writer.writeOpenTag(SumStatisticParser.SUM_STATISTIC, new Attribute[]{ new Attribute.Default(XMLParser.ID, prefix + "includedPredictors")}); writer.writeIDref(ParameterParser.PARAMETER, prefix + "coefIndicators"); writer.writeCloseTag(SumStatisticParser.SUM_STATISTIC); } else { throw new IllegalArgumentException("Unknown discreteSubstType"); } if (model.isActivateBSSVS()) { // If "BSSVS" is not activated, rateIndicator should not be there. writeStatisticModel(model, writer); } } private void writeDiscreteTraitsSubstitutionModelReferences(XMLWriter writer) { for (PartitionSubstitutionModel model : options.getPartitionSubstitutionModels(GeneralDataType.INSTANCE)) { if (model.getDiscreteSubstType() == DiscreteSubstModelType.GLM_SUBST) { // not strictly necessary but makes the XML consistent writer.writeIDref(GLMSubstitutionModelParser.GLM_SUBSTITUTION_MODEL, model.getName() + "." + AbstractSubstitutionModel.MODEL); } else { writer.writeIDref(GeneralSubstitutionModelParser.GENERAL_SUBSTITUTION_MODEL, model.getName() + "." + AbstractSubstitutionModel.MODEL); } } } private boolean hasBSSVS() { for (PartitionSubstitutionModel model : options.getPartitionSubstitutionModels(GeneralDataType.INSTANCE)) { if (model.isActivateBSSVS()) return true; } return false; } private void writeDiscreteFrequencyModel(String prefix, String dataTypePrefix, int stateCount, Boolean normalize, XMLWriter writer) { if (normalize == null) { writer.writeOpenTag(FrequencyModelParser.FREQUENCY_MODEL, new Attribute[]{ new Attribute.Default(XMLParser.ID, prefix + FrequencyModelParser.FREQUENCY_MODEL)}); } else { writer.writeOpenTag(FrequencyModelParser.FREQUENCY_MODEL, new Attribute[]{ new Attribute.Default(XMLParser.ID, prefix + FrequencyModelParser.FREQUENCY_MODEL), new Attribute.Default(FrequencyModelParser.NORMALIZE, normalize)}); } writer.writeIDref(GeneralDataTypeParser.GENERAL_DATA_TYPE, dataTypePrefix + "dataType"); writer.writeOpenTag(FrequencyModelParser.FREQUENCIES); writeParameter(prefix + "frequencies", stateCount, Double.NaN, Double.NaN, Double.NaN, writer); writer.writeCloseTag(FrequencyModelParser.FREQUENCIES); writer.writeCloseTag(FrequencyModelParser.FREQUENCY_MODEL); } private void writeDiscreteTraitsSiteModel(PartitionSubstitutionModel model, XMLWriter writer) { String prefix = model.getName() + "."; writer.writeOpenTag(SiteModel.SITE_MODEL, new Attribute[]{ new Attribute.Default(XMLParser.ID, prefix + SiteModel.SITE_MODEL)}); writer.writeOpenTag(GammaSiteModelParser.SUBSTITUTION_MODEL); if (model.getDiscreteSubstType() == DiscreteSubstModelType.GLM_SUBST) { // not strictly necessary but makes the XML consistent writer.writeIDref(GLMSubstitutionModelParser.GLM_SUBSTITUTION_MODEL, prefix + AbstractSubstitutionModel.MODEL); } else { writer.writeIDref(GeneralSubstitutionModelParser.GENERAL_SUBSTITUTION_MODEL, prefix + AbstractSubstitutionModel.MODEL); } writer.writeCloseTag(GammaSiteModelParser.SUBSTITUTION_MODEL); writer.writeCloseTag(SiteModel.SITE_MODEL); } private void writeRatesAndIndicators(PartitionSubstitutionModel model, int dimension, Integer relativeTo, XMLWriter writer) { writer.writeComment("rates and indicators"); String prefix = model.getName() + "."; if (relativeTo == null) { writer.writeOpenTag(GeneralSubstitutionModelParser.RATES); } else { writer.writeOpenTag(GeneralSubstitutionModelParser.RATES, new Attribute[]{ new Attribute.Default(GeneralSubstitutionModelParser.RELATIVE_TO, relativeTo)}); } writeParameter(options.getParameter(prefix + "rates"), dimension, writer); writer.writeCloseTag(GeneralSubstitutionModelParser.RATES); if (model.isActivateBSSVS()) { //If "BSSVS" is not activated, rateIndicator should not be there. writer.writeOpenTag(GeneralSubstitutionModelParser.INDICATOR); writeParameter(options.getParameter(prefix + "indicators"), dimension, writer); writer.writeCloseTag(GeneralSubstitutionModelParser.INDICATOR); } } private void writeStatisticModel(PartitionSubstitutionModel model, XMLWriter writer) { String prefix = model.getName() + "."; writer.writeOpenTag(SumStatisticParser.SUM_STATISTIC, new Attribute[]{ new Attribute.Default(XMLParser.ID, prefix + "nonZeroRates"), new Attribute.Default(SumStatisticParser.ELEMENTWISE, true)}); writer.writeIDref(ParameterParser.PARAMETER, prefix + "indicators"); writer.writeCloseTag(SumStatisticParser.SUM_STATISTIC); writer.writeOpenTag(ProductStatisticParser.PRODUCT_STATISTIC, new Attribute[]{ new Attribute.Default(XMLParser.ID, prefix + "actualRates"), new Attribute.Default(SumStatisticParser.ELEMENTWISE, false)}); writer.writeIDref(ParameterParser.PARAMETER, prefix + "indicators"); writer.writeIDref(ParameterParser.PARAMETER, prefix + "rates"); writer.writeCloseTag(ProductStatisticParser.PRODUCT_STATISTIC); } private void writeTreeLikelihoods(XMLWriter writer, DiscreteTraitsComponentOptions component) { // generate tree likelihoods for discrete trait partitions if (options.hasDiscreteTraitPartition()) { writer.writeComment("Likelihood for tree given discrete trait data"); } for (AbstractPartitionData partition : options.dataPartitions) { if (partition.getTraits() != null) { TraitData trait = partition.getTraits().get(0); if (trait.getTraitType() == TraitData.TraitType.DISCRETE) { writeTreeLikelihood(partition, writer); } } } } /** * Write Tree Likelihood * * @param partition PartitionData * @param writer XMLWriter */ public void writeTreeLikelihood(AbstractPartitionData partition, XMLWriter writer) { String prefix = partition.getName() + "."; PartitionSubstitutionModel substModel = partition.getPartitionSubstitutionModel(); PartitionTreeModel treeModel = partition.getPartitionTreeModel(); PartitionClockModel clockModel = partition.getPartitionClockModel(); AncestralStatesComponentOptions ancestralStatesOptions = (AncestralStatesComponentOptions)options.getComponentOptions(AncestralStatesComponentOptions.class); String treeLikelihoodTag = TreeLikelihoodParser.ANCESTRAL_TREE_LIKELIHOOD; if (ancestralStatesOptions.isCountingStates(partition)) { treeLikelihoodTag = MarkovJumpsTreeLikelihoodParser.MARKOV_JUMP_TREE_LIKELIHOOD; } boolean saveCompleteHistory = ancestralStatesOptions.isCompleteHistoryLogging(partition); writer.writeOpenTag(treeLikelihoodTag, new Attribute[]{ new Attribute.Default(XMLParser.ID, prefix + TreeLikelihoodParser.TREE_LIKELIHOOD), new Attribute.Default(AncestralStateTreeLikelihoodParser.RECONSTRUCTION_TAG_NAME, prefix + AncestralStateTreeLikelihoodParser.RECONSTRUCTION_TAG), new Attribute.Default(MarkovJumpsTreeLikelihoodParser.USE_UNIFORMIZATION, true), new Attribute.Default(MarkovJumpsTreeLikelihoodParser.SAVE_HISTORY, saveCompleteHistory ? "true" : "false"), new Attribute.Default(MarkovJumpsTreeLikelihoodParser.LOG_HISTORY, saveCompleteHistory ? "true" : "false"), }); writer.writeIDref(AttributePatternsParser.ATTRIBUTE_PATTERNS, prefix + "pattern"); writer.writeIDref(TreeModel.TREE_MODEL, treeModel.getPrefix() + TreeModel.TREE_MODEL); writer.writeIDref(SiteModel.SITE_MODEL, substModel.getName() + "." + SiteModel.SITE_MODEL); if (partition.getPartitionSubstitutionModel().getDiscreteSubstType() == DiscreteSubstModelType.GLM_SUBST) { // not strictly necessary but makes the XML consistent writer.writeIDref(GLMSubstitutionModelParser.GLM_SUBSTITUTION_MODEL, prefix + AbstractSubstitutionModel.MODEL); } else { writer.writeIDref(GeneralSubstitutionModelParser.GENERAL_SUBSTITUTION_MODEL, prefix + AbstractSubstitutionModel.MODEL); } writer.writeIDref(GeneralSubstitutionModelParser.GENERAL_SUBSTITUTION_MODEL, substModel.getName() + "." + AbstractSubstitutionModel.MODEL); ClockModelGenerator.writeBranchRatesModelRef(clockModel, writer); if (substModel.getDiscreteSubstType() == DiscreteSubstModelType.ASYM_SUBST) { int stateCount = options.getStatesForDiscreteModel(substModel).size(); writer.writeComment("The root state frequencies"); writeDiscreteFrequencyModel(partition.getPrefix() + "root.", substModel.getName() + ".", stateCount, true, writer); } getCallingGenerator().generateInsertionPoint(ComponentGenerator.InsertionPoint.IN_TREE_LIKELIHOOD, partition, writer); writer.writeCloseTag(treeLikelihoodTag); } private void writeTreeLikelihoodReferences(XMLWriter writer) { for (AbstractPartitionData partition : options.dataPartitions) { if (partition.getTraits() != null) { AncestralStatesComponentOptions ancestralStatesOptions = (AncestralStatesComponentOptions)options.getComponentOptions(AncestralStatesComponentOptions.class); String treeLikelihoodTag = TreeLikelihoodParser.ANCESTRAL_TREE_LIKELIHOOD; if (ancestralStatesOptions.isCountingStates(partition)) { treeLikelihoodTag = MarkovJumpsTreeLikelihoodParser.MARKOV_JUMP_TREE_LIKELIHOOD; } TraitData trait = partition.getTraits().get(0); String prefix = partition.getName() + "."; if (trait.getTraitType() == TraitData.TraitType.DISCRETE) { writer.writeIDref(treeLikelihoodTag, prefix + TreeLikelihoodParser.TREE_LIKELIHOOD); } } } } private void writeGLMCoefficientOperator(PartitionSubstitutionModel model, XMLWriter writer) { writer.writeOpenTag(MultivariateNormalOperator.MVN_OPERATOR, new Attribute[]{ new Attribute.Default(MultivariateNormalOperator.SCALE_FACTOR, 1.0), new Attribute.Default(MultivariateNormalOperator.WEIGHT,5.0), new Attribute.Default(MultivariateNormalOperator.FORM_XTX, "true") } ); String prefix = model.getName() + "."; writeParameterRef(prefix + "coefficients", writer); writer.writeOpenTag(MultivariateNormalOperator.VARIANCE_MATRIX); writeParameterRef(prefix + DesignMatrix.DESIGN_MATRIX, writer); writer.writeCloseTag(MultivariateNormalOperator.VARIANCE_MATRIX); writer.writeCloseTag(MultivariateNormalOperator.MVN_OPERATOR); } private void writeGLMBinomialLikelihood(PartitionSubstitutionModel model, XMLWriter writer) { double proportion = 1.0 - Math.exp(Math.log(0.5) / model.getTraitData().getIncludedPredictorCount()); String prefix = model.getName() + "."; writer.writeComment("Using the binomialLikelihood we specify a 50% prior mass on no predictors being included."); writer.writeOpenTag(BinomialLikelihood.BINOMIAL_LIKELIHOOD); writer.writeOpenTag(BinomialLikelihoodParser.PROPORTION); writer.writeTag("parameter", new Attribute.Default("value", proportion), true); writer.writeCloseTag(BinomialLikelihoodParser.PROPORTION); // the dimension of this parameter will be set automatically to be the same as the counts. writer.writeOpenTag(BinomialLikelihoodParser.TRIALS); writer.writeTag("parameter", new Attribute[]{ new Attribute.Default("value", 1.0) }, true); writer.writeCloseTag(BinomialLikelihoodParser.TRIALS); writer.writeOpenTag(BinomialLikelihoodParser.COUNTS); writeParameterRef(prefix + "coefIndicators", writer); writer.writeCloseTag(BinomialLikelihoodParser.COUNTS); writer.writeCloseTag(BinomialLikelihood.BINOMIAL_LIKELIHOOD); } public void writeScreenLogEntries(XMLWriter writer) { for (PartitionSubstitutionModel model : options.getPartitionSubstitutionModels(GeneralDataType.INSTANCE)) { String prefix = model.getName() + "."; if (model.isActivateBSSVS()) { //If "BSSVS" is not activated, rateIndicator should not be there. writer.writeOpenTag(ColumnsParser.COLUMN, new Attribute[]{ new Attribute.Default(ColumnsParser.LABEL, prefix + "nonZeroRates"), new Attribute.Default(ColumnsParser.DECIMAL_PLACES, "0"), new Attribute.Default(ColumnsParser.WIDTH, "6") } ); writer.writeIDref(SumStatisticParser.SUM_STATISTIC, prefix + "nonZeroRates"); writer.writeCloseTag(ColumnsParser.COLUMN); } if (model.getDiscreteSubstType() == DiscreteSubstModelType.GLM_SUBST) { writer.writeOpenTag(ColumnsParser.COLUMN, new Attribute[]{ new Attribute.Default(ColumnsParser.LABEL, prefix + "incPredictors"), new Attribute.Default(ColumnsParser.DECIMAL_PLACES, "0"), new Attribute.Default(ColumnsParser.WIDTH, "6") } ); writer.writeIDref(SumStatisticParser.SUM_STATISTIC, prefix + "includedPredictors"); writer.writeCloseTag(ColumnsParser.COLUMN); } } } private void writeFileLogEntries(XMLWriter writer) { for (PartitionSubstitutionModel model : options.getPartitionSubstitutionModels(GeneralDataType.INSTANCE)) { String prefix = model.getName() + "."; if (model.getDiscreteSubstType() == DiscreteSubstModelType.GLM_SUBST) { writer.writeIDref(SumStatisticParser.SUM_STATISTIC, prefix + "includedPredictors"); writer.writeIDref(ProductStatisticParser.PRODUCT_STATISTIC, prefix + "coefficientsTimesIndicators"); } else { writer.writeIDref(ParameterParser.PARAMETER, prefix + "rates"); if (model.isActivateBSSVS()) { //If "BSSVS" is not activated, rateIndicator should not be there. writer.writeIDref(ParameterParser.PARAMETER, prefix + "indicators"); writer.writeIDref(SumStatisticParser.SUM_STATISTIC, prefix + "nonZeroRates"); } } } } private void writeDiscreteTraitFileLoggers(XMLWriter writer) { for (PartitionSubstitutionModel model : options.getPartitionSubstitutionModels(GeneralDataType.INSTANCE)) { writeDiscreteTraitFileLogger(writer, model); } } private void writeDiscreteTraitFileLogger(XMLWriter writer, PartitionSubstitutionModel model) { String prefix = options.fileNameStem + "." + model.getName(); if (model.getDiscreteSubstType() == DiscreteSubstModelType.GLM_SUBST) { writer.writeOpenTag(LoggerParser.LOG, new Attribute[]{ new Attribute.Default(XMLParser.ID, prefix + "glmLog"), new Attribute.Default(LoggerParser.LOG_EVERY, options.logEvery + ""), new Attribute.Default(LoggerParser.FILE_NAME, prefix + ".glm.log")}); } else { writer.writeOpenTag(LoggerParser.LOG, new Attribute[]{ new Attribute.Default(XMLParser.ID, prefix + "rateMatrixLog"), new Attribute.Default(LoggerParser.LOG_EVERY, options.logEvery + ""), new Attribute.Default(LoggerParser.FILE_NAME, prefix + ".rates.log")}); } writeLogEntries(model, writer); writer.writeCloseTag(LoggerParser.LOG); } private void writeLogEntries(PartitionSubstitutionModel model, XMLWriter writer) { String prefix = model.getName() + "."; if (model.getDiscreteSubstType() == DiscreteSubstModelType.GLM_SUBST) { writer.writeIDref(ParameterParser.PARAMETER, prefix + "coefficients"); writer.writeIDref(ParameterParser.PARAMETER, prefix + "coefIndicators"); writer.writeIDref(ProductStatisticParser.PRODUCT_STATISTIC, prefix + "coefficientsTimesIndicators"); } else { writer.writeIDref(ParameterParser.PARAMETER, prefix + "rates"); if (model.isActivateBSSVS()) { //If "BSSVS" is not activated, rateIndicator should not be there. writer.writeIDref(ParameterParser.PARAMETER, prefix + "indicators"); writer.writeIDref(SumStatisticParser.SUM_STATISTIC, prefix + "nonZeroRates"); } } } }