RateSmoothingTree.java

/*
 * RateSmoothingTree.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.evolution.tree;

import dr.math.ConjugateDirectionSearch;
import dr.math.MultivariateFunction;
import dr.math.MultivariateMinimum;

/**
 * An extension of simple tree that provides non-parametric rate smoothed
 * node heights in the manner described by Mike Sanderson. This takes a
 * tree with branch lengths as data and uses multidimensional optimization
 * to find the set of node heights that requires the smallest differences
 * in rates across nodes.
 *
 * @version $Id: RateSmoothingTree.java,v 1.8 2005/05/24 20:25:56 rambaut Exp $
 *
 * @author Andrew Rambaut
 * @author Alexei Drummond
 */
public class RateSmoothingTree extends SimpleTree {

	/**
	 * constructor 
	 */
	public RateSmoothingTree(Tree sourceTree) {
	
		super(sourceTree);
		this.sourceTree = sourceTree;
		this.mu = 1.0;
		this.optimizeMu = false;
	}

	/**
	 * constructor with a specific mutation rate
	 */
	public RateSmoothingTree(Tree sourceTree, double mu) {
	
		this.sourceTree = sourceTree;
		this.mu = mu;
		this.optimizeMu = false;
	}

	public double getMu() { return mu; }
	
	public void smoothRates() {
		nodeCount = getInternalNodeCount();
		int argumentCount = nodeCount;
		
		if (optimizeMu) {
			argumentCount++;
			muIndex = nodeCount;
		}
		
		MultivariateMinimum optimizer = new ConjugateDirectionSearch();
		
		nodeValues = new double[nodeCount];
		
		double[] xvec = new double[argumentCount];
		for (int i = 0; i < nodeCount; i++) { 
			xvec[i] = 1.0;
		}
		if (optimizeMu) {
			xvec[muIndex] = mu;
		}
		
		optimizer.optimize(nonParametricRateSmoothing, xvec, 1E-8, 1E-8);
	}

	public double getSumOfRates() {
	
		double[] score = new double[] { 0.0 };
		
		NodeRef root = getRoot();
		
		if (getChildCount(root) != 2) {
			throw new IllegalArgumentException("The tree must have a bifurcating root node");
		}
		
		sumDist = 0;
		sumTime = 0;
		
		double rate1 = sumScoreAtNode(getChild(root, 0), score);
		double rate2 = sumScoreAtNode(getChild(root, 1), score);
		
		mu = sumDist / sumTime;
		
		double diff = rate2 - rate1;
		score[0] += diff * diff;
		
		return score[0];
	}

	//
	// Private stuff
	//	
	
	private double sumScoreAtNode(NodeRef node, double[] score) {
		
		double rate0 = getRateAtNode(node);
		
		if (!isExternal(node)) {

			for (int i = 0; i < getChildCount(node); i++) {
				double rate1 = sumScoreAtNode(getChild(node, i), score);
				double diff = rate0 - rate1;
				score[0] += diff * diff;
			}
		}
		
		return rate0;
	}

	/**
	 * returns the rate on the branch to the node above.
	 */
	private double getRateAtNode(NodeRef node) {
		double time = getNodeHeight(getParent(node)) - getNodeHeight(node);
		double dist = sourceTree.getBranchLength(sourceTree.getNode(node.getNumber()));
		
		double rate;
		
		if (time == 0.0) {
			if (dist == 0.0) {
				rate = 1.0;
			} else {
				rate = Double.MIN_VALUE;
			}
		} else {
			rate = (dist / time);
		}
		sumDist += dist;
		sumTime += time;
	
		setNodeRate(node, rate);
		return rate;
	}


	private double setNodeHeightsFromValues(NodeRef node) {
		
		if (!isExternal(node)) {
			double maxHeight = setNodeHeightsFromValues(getChild(node, 0));

			for (int i = 1; i < getChildCount(node); i++) {
				double height = setNodeHeightsFromValues(getChild(node, i));
				if (height > maxHeight) maxHeight = height;
			}
			
			setNodeHeight(node, maxHeight + nodeValues[node.getNumber() - getExternalNodeCount()]);
		}
		
		return getNodeHeight(node);
	}

	private MultivariateFunction nonParametricRateSmoothing = new MultivariateFunction() {
	
		public double evaluate(double[] argument) {
			for (int i = 0; i < getInternalNodeCount(); i++) {
				nodeValues[i] = argument[i];
			}
			
			setNodeHeightsFromValues(getRoot());
			
			if (optimizeMu) {
				mu = argument[muIndex];
			}
			
			double score = getSumOfRates();
			return score;
		}
		
		public int getNumArguments() { 
			if (optimizeMu) {
				return getInternalNodeCount() + 1; 
			} else {
				return getInternalNodeCount(); 
			}
		}
		
		public double getLowerBound(int n) {
		
			if (optimizeMu && n == muIndex) {
				return Double.MIN_VALUE;
			} else {
				return 0.0;
			}
		}
		
		public double getUpperBound(int n) {

			if (optimizeMu && n == muIndex) {
				return Double.MAX_VALUE;
			} else {
				return Double.MAX_VALUE;
			}
		}
	};

	private int nodeCount;
	private double[] nodeValues;
	private Tree sourceTree;
	private double mu, sumDist, sumTime;
	private boolean optimizeMu;
	private int muIndex;
}