/*
* DiscretizedBranchRates.java
*
* Copyright (C) 2002-2009 Alexei Drummond and Andrew Rambaut
*
* 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.evomodel.branchratemodel;
import dr.evolution.tree.NodeRef;
import dr.evolution.tree.Tree;
import dr.evolution.util.Taxon;
import dr.evomodel.tree.TreeModel;
import dr.evomodel.tree.TreeParameterModel;
import dr.evomodelxml.branchratemodel.DiscretizedBranchRatesParser;
import dr.inference.distribution.ParametricDistributionModel;
import dr.inference.model.Model;
import dr.inference.model.ModelListener;
import dr.inference.model.Parameter;
import dr.inference.model.Variable;
/**
* @author Alexei Drummond
* @author Andrew Rambaut
* @author Michael Defoin Platel
* @version $Id: DiscretizedBranchRates.java,v 1.11 2006/01/09 17:44:30 rambaut Exp $
*/
public class DiscretizedBranchRates extends AbstractBranchRateModel {
private final ParametricDistributionModel distributionModel;
// The rate categories of each branch
final TreeParameterModel rateCategories;
private final int categoryCount;
private final double step;
private final double[] rates;
private boolean normalize = false;
private double normalizeBranchRateTo = Double.NaN;
private double scaleFactor = 1.0;
private TreeModel treeModel;
private final double logDensityNormalizationConstant;
//overSampling control the number of effective categories
public DiscretizedBranchRates(
TreeModel tree,
Parameter rateCategoryParameter,
ParametricDistributionModel model,
int overSampling) {
this(tree, rateCategoryParameter, model, overSampling, false, Double.NaN);
}
public DiscretizedBranchRates(
TreeModel tree,
Parameter rateCategoryParameter,
ParametricDistributionModel model,
int overSampling,
boolean normalize,
double normalizeBranchRateTo) {
super(DiscretizedBranchRatesParser.DISCRETIZED_BRANCH_RATES);
this.rateCategories = new TreeParameterModel(tree, rateCategoryParameter, false);
categoryCount = (tree.getNodeCount() - 1) * overSampling;
step = 1.0 / (double) categoryCount;
rates = new double[categoryCount];
this.normalize = normalize;
this.treeModel = tree;
this.distributionModel = model;
this.normalizeBranchRateTo = normalizeBranchRateTo;
//Force the boundaries of rateCategoryParameter to match the category count
Parameter.DefaultBounds bound = new Parameter.DefaultBounds(categoryCount - 1, 0, rateCategoryParameter.getDimension());
rateCategoryParameter.addBounds(bound);
for (int i = 0; i < rateCategoryParameter.getDimension(); i++) {
int index = (int) Math.floor((i + 0.5) * overSampling);
rateCategoryParameter.setParameterValue(i, index);
}
addModel(model);
// AR - commented out: changes to the tree are handled by model changed events fired by rateCategories
// addModel(tree);
addModel(rateCategories);
// AR - commented out: changes to rateCategoryParameter are handled by model changed events fired by rateCategories
// addVariable(rateCategoryParameter);
if (normalize) {
tree.addModelListener(new ModelListener() {
public void modelChangedEvent(Model model, Object object, int index) {
computeFactor();
}
public void modelRestored(Model model) {
computeFactor();
}
});
}
setupRates();
// Each parameter take any value in [1, \ldots, categoryCount]
// NB But this depends on the transition kernel employed. Using swap-only results in a different constant
logDensityNormalizationConstant = -rateCategoryParameter.getDimension() * Math.log(categoryCount);
}
// compute scale factor
private void computeFactor() {
//scale mean rate to 1.0 or separate parameter
double treeRate = 0.0;
double treeTime = 0.0;
//normalizeBranchRateTo = 1.0;
for (int i = 0; i < treeModel.getNodeCount(); i++) {
NodeRef node = treeModel.getNode(i);
if (!treeModel.isRoot(node)) {
int rateCategory = (int) Math.round(rateCategories.getNodeValue(treeModel, node));
treeRate += rates[rateCategory] * treeModel.getBranchLength(node);
treeTime += treeModel.getBranchLength(node);
//System.out.println("rates and time\t" + rates[rateCategory] + "\t" + treeModel.getBranchLength(node));
}
}
//treeRate /= treeTime;
scaleFactor = normalizeBranchRateTo / (treeRate / treeTime);
//System.out.println("scaleFactor\t\t\t\t\t" + scaleFactor);
}
public void handleModelChangedEvent(Model model, Object object, int index) {
if (model == distributionModel) {
setupRates();
fireModelChanged();
} else if (model == rateCategories) {
// AR - commented out: if just the rate categories have changed the rates will be the same
// setupRates();
fireModelChanged(null, index);
}
}
protected final void handleVariableChangedEvent(Variable variable, int index, Parameter.ChangeType type) {
// AR - commented out: changes to rateCategoryParameter are handled by model changed events
// setupRates();
}
protected void storeState() {
}
protected void restoreState() {
setupRates();
}
protected void acceptState() {
}
public double getBranchRate(final Tree tree, final NodeRef node) {
assert !tree.isRoot(node) : "root node doesn't have a rate!";
int rateCategory = (int) Math.round(rateCategories.getNodeValue(tree, node));
// whdc: Do not sample for the rate of branches leading up
// to adventitious leaves. I would return zero, but that
// cause the tree likelihood to return NaN.
Taxon taxon = tree.getNodeTaxon( node);
if( taxon != null && taxon.getAdventitious()) {
return 0.0000001;
}
//System.out.println(rates[rateCategory] + "\t" + rateCategory);
return rates[rateCategory] * scaleFactor;
}
/**
* Calculates the actual rates corresponding to the category indices.
*/
protected void setupRates() {
double z = step / 2.0;
for (int i = 0; i < categoryCount; i++) {
rates[i] = distributionModel.quantile(z);
//System.out.print(rates[i]+"\t");
z += step;
}
/*if(distributionModel.getClass().getName().equals("dr.inference.distribution.LogNormalDistributionModel")) {
LogNormalDistributionModel lndm = (LogNormalDistributionModel) distributionModel;
System.out.println("chur " + lndm.getS() +"\t" + lndm.getM());
}
else { System.out.println(distributionModel.getClass().getName());}*/
if (normalize) computeFactor();
}
public double getLogLikelihood() {
return logDensityNormalizationConstant;
}
}