Framework for generalized association testing. Heavy lifting done in implementation of the AssociationContext(s) and AssociationContextAtom(s). Nothing really implemented.

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@5306 348d0f76-0448-11de-a6fe-93d51630548a

java/src/org/broadinstitute/sting/oneoffprojects/walkers/association/AssociationContext.java

@@ -0,0 +1,40 @@
+package org.broadinstitute.sting.oneoffprojects.walkers.association;
+
+import java.lang.reflect.InvocationTargetException;
+import java.util.*;
+
+/**
+ * @Author chartl
+ * @Date 2011-02-23
+ * A general context for windowed association
+ */
+public abstract class AssociationContext<X extends AssociationContextAtom> {
+    private Class<? extends AssociationContextAtom> clazz;
+    private List<X> window;
+
+    public AssociationContext( Class<X> zclaz ) {
+        window = new ArrayList<X>(getWindowSize());
+        clazz = zclaz;
+    }
+
+    public X map(MapExtender e) throws NoSuchMethodException, InstantiationException, IllegalAccessException, InvocationTargetException, ClassCastException {
+        return (X) clazz.getConstructor().newInstance(e);
+    }
+
+    public boolean filter(MapExtender m) { return true; }
+
+    public void reduce(X context) {
+        window.add(context);
+    }
+
+    public abstract int getWindowSize();
+    public abstract int slideByValue();
+
+    public boolean isFull() {
+        return window.size() >= getWindowSize();
+    }
+
+    public void slide() {
+        window = window.subList(slideByValue(),window.size());
+    }
+}

java/src/org/broadinstitute/sting/oneoffprojects/walkers/association/AssociationContextAtom.java

@@ -0,0 +1,13 @@
+package org.broadinstitute.sting.oneoffprojects.walkers.association;
+
+/**
+ * @Author chartl
+ * @Date 2011-02-23
+ * general class for calculating base data for generalized association tests
+ */
+public abstract class AssociationContextAtom {
+
+    public AssociationContextAtom(MapExtender e) {
+
+    }
+}

java/src/org/broadinstitute/sting/oneoffprojects/walkers/association/AssociationTestRunner.java

@@ -0,0 +1,44 @@
+package org.broadinstitute.sting.oneoffprojects.walkers.association;
+
+import java.util.ArrayList;
+import java.util.List;
+import org.broadinstitute.sting.oneoffprojects.walkers.association.interfaces.*;
+
+/**
+ * Created by IntelliJ IDEA.
+ * User: chartl
+ * Date: Feb 24, 2011
+ * Time: 11:48:26 AM
+ * To change this template use File | Settings | File Templates.
+ */
+public class AssociationTestRunner {
+
+    public static List<String> runTests(AssociationContext context) {
+        List<String> results = new ArrayList<String>();
+        if ( context.getClass().isInstance(TStatistic.class)) {
+            results.add(runStudentT(context));
+        }
+
+        if ( context.getClass().isInstance(ZStatistic.class)) {
+            results.add(runZ(context));
+        }
+
+        if ( context.getClass().isInstance(FisherExact.class) ) {
+            results.add(runFisherExact(context));
+        }
+
+        return results;
+    }
+
+    public static String runStudentT(AssociationContext context) {
+        return "Test not yet implemented";
+    }
+
+    public static String runZ(AssociationContext context) {
+        return "Test not yet implemented";
+    }
+
+    public static String runFisherExact(AssociationContext context) {
+        return "Test not yet implemented";
+    }
+}

java/src/org/broadinstitute/sting/oneoffprojects/walkers/association/MapExtender.java

@@ -0,0 +1,52 @@
+package org.broadinstitute.sting.oneoffprojects.walkers.association;
+
+import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
+import org.broadinstitute.sting.gatk.contexts.StratifiedAlignmentContext;
+import org.broadinstitute.sting.gatk.datasources.sample.Sample;
+import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
+
+import java.util.Map;
+
+/**
+ * @Author chartl
+ * @Date 2011-02-23
+ * Holds multiple map contexts for use in the regional association walker
+ */
+public class MapExtender {
+
+    private MapHolder previous = null;
+    private MapHolder current = null;
+
+    public MapExtender() {
+        // no need to do anything
+    }
+
+    public void set(MapHolder holder) {
+        previous = current;
+        current = holder;
+    }
+
+    public Map<Sample,StratifiedAlignmentContext> getPreviousContext() {
+        return previous.getContext();
+    }
+
+    public ReferenceContext getPreviousRef() {
+        return previous.getRef();
+    }
+
+    public RefMetaDataTracker getPreviousTracker() {
+        return previous.getTracker();
+    }
+
+    public Map<Sample,StratifiedAlignmentContext> getContext() {
+        return current.getContext();
+    }
+
+    public ReferenceContext getReferenceContext() {
+        return current.getRef();
+    }
+
+    public RefMetaDataTracker getTracker() {
+        return current.getTracker();
+    }
+}

java/src/org/broadinstitute/sting/oneoffprojects/walkers/association/MapHolder.java

@@ -0,0 +1,33 @@
+package org.broadinstitute.sting.oneoffprojects.walkers.association;
+
+import org.broadinstitute.sting.gatk.contexts.AlignmentContext;
+import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
+import org.broadinstitute.sting.gatk.contexts.StratifiedAlignmentContext;
+import org.broadinstitute.sting.gatk.datasources.sample.Sample;
+import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
+
+import java.util.Map;
+
+public class MapHolder {
+    private RefMetaDataTracker tracker;
+    private ReferenceContext ref;
+    private Map<Sample, StratifiedAlignmentContext> alignments;
+
+    public MapHolder(RefMetaDataTracker t, ReferenceContext r, AlignmentContext a) {
+        tracker = t;
+        ref = r;
+        alignments = StratifiedAlignmentContext.splitContextBySample(a.getBasePileup());
+    }
+
+    public Map<Sample, StratifiedAlignmentContext> getContext() {
+        return alignments;
+    }
+
+    public ReferenceContext getRef() {
+        return ref;
+    }
+
+    public RefMetaDataTracker getTracker() {
+        return tracker;
+    }
+}

java/src/org/broadinstitute/sting/oneoffprojects/walkers/association/RegionalAssociationHandler.java

@@ -0,0 +1,62 @@
+package org.broadinstitute.sting.oneoffprojects.walkers.association;
+
+import org.broadinstitute.sting.utils.GenomeLoc;
+
+import java.lang.reflect.InvocationTargetException;
+import java.util.*;
+
+/**
+ * @Author chartl
+ * @Date 2011-02-23
+ * A general context for windowed association
+ */
+public class RegionalAssociationHandler {
+    private MapExtender maps;
+    // todo -- the correct way to do this is via the PluginManager (a la VariantEval) but this is a QND implementation
+    private Set<AssociationContext> associations;
+
+    public RegionalAssociationHandler(Set<AssociationContext> contexts) {
+        maps = null;
+        associations = contexts;
+    }
+
+    public void updateExtender(MapHolder mapHolder) {
+        maps.set(mapHolder);
+    }
+
+    /**
+     * Once the instances are collected; run map-reduce
+     * @map: MapExtender --> A child of AssociationContextAtom
+     *  @implementation: Indirect construction via newinstance
+     * @reduce: (List<AssociationContext>,AssociationContext) --> List<AssociationContextAtom>
+     *  @implementation: just append
+     */
+    public void runMapReduce() throws NoSuchMethodException, IllegalAccessException, InstantiationException, InvocationTargetException {
+        for ( AssociationContext w : associations ) {
+            if ( w.filter(maps) ) {
+                w.reduce(w.map(maps));
+            }
+        }
+    }
+
+    /**
+     * For those AssociationContexts with full windows:
+     *  1) Run their associated test(s)
+     *  2) Slide the windows
+     */
+    public List<String> runTests() {
+        List<String> testResults = new ArrayList<String>(associations.size());
+        for ( AssociationContext w : associations ) {
+            if ( w.isFull() ) {
+                testResults.addAll(AssociationTestRunner.runTests(w));
+                w.slide();
+            }
+        }
+        return testResults;
+    }
+
+    public GenomeLoc getLocation() {
+        return maps.getReferenceContext().getLocus();
+    }
+
+}

java/src/org/broadinstitute/sting/oneoffprojects/walkers/association/RegionalAssociationWalker.java

@@ -0,0 +1,78 @@
+package org.broadinstitute.sting.oneoffprojects.walkers.association;
+
+import org.broadinstitute.sting.commandline.Argument;
+import org.broadinstitute.sting.commandline.Output;
+import org.broadinstitute.sting.gatk.contexts.AlignmentContext;
+import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
+import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
+import org.broadinstitute.sting.gatk.walkers.LocusWalker;
+import org.broadinstitute.sting.gatk.walkers.TreeReducible;
+import org.broadinstitute.sting.utils.exceptions.StingException;
+
+import java.io.PrintStream;
+import java.util.HashSet;
+import java.util.List;
+import java.util.Set;
+
+/**
+ * @Author chartl
+ * @Date 2011-02-23
+ * Generalized framework for regional (windowed) associations
+ */
+public class RegionalAssociationWalker extends LocusWalker<MapHolder, RegionalAssociationHandler> implements TreeReducible<RegionalAssociationHandler> {
+    @Argument(doc="foo",shortName="AT",fullName="associationType",required=false)
+    public String[] associationsToUse = null;
+
+    @Output
+    PrintStream out;
+
+    public RegionalAssociationHandler reduceInit() {
+        Set<AssociationContext> validAssociations = getAssociations();
+        RegionalAssociationHandler wac = new RegionalAssociationHandler(validAssociations);
+
+        return wac;
+    }
+
+    public MapHolder map(RefMetaDataTracker tracker, ReferenceContext ref, AlignmentContext context) {
+        return new MapHolder(tracker,ref,context);
+    }
+
+    public RegionalAssociationHandler reduce(MapHolder map, RegionalAssociationHandler rac) {
+        rac.updateExtender(map);
+        try {
+            rac.runMapReduce();
+        } catch (Exception e) {
+            throw new StingException("Error in map reduce",e);
+        }
+        List<String> testsHere = rac.runTests();
+        // todo -- really awful shitty formatting
+        out.printf("%s%n",rac.getLocation().toString());
+        for ( String s : testsHere ) {
+            out.printf("%s%n",s);
+        }
+        return rac;
+    }
+
+    private AssociationContext stringToAssociationContext(String s) {
+        return null;
+    }
+
+    private Set<AssociationContext> getAssociations() {
+        // todo -- this should use the package handler like variant eval
+        Set<AssociationContext> validAssociations = new HashSet<AssociationContext>();
+        for ( String s : associationsToUse ) {
+            validAssociations.add(stringToAssociationContext(s));
+        }
+        return validAssociations;
+    }
+
+    public RegionalAssociationHandler treeReduce(RegionalAssociationHandler left, RegionalAssociationHandler right) {
+        // for now be dumb; in future fix the fact that left-most intervals of a 16kb shard won't see the context from
+        // the right-most locus of the previous shard
+        return right;
+    }
+
+    public void onTraversalDone(RegionalAssociationHandler rac) {
+        // do nothing
+    }
+}

java/src/org/broadinstitute/sting/oneoffprojects/walkers/association/interfaces/FisherExact.java

@@ -0,0 +1,14 @@
+package org.broadinstitute.sting.oneoffprojects.walkers.association.interfaces;
+
+/**
+ * Created by IntelliJ IDEA.
+ * User: chartl
+ * Date: Feb 24, 2011
+ * Time: 12:58:56 PM
+ * To change this template use File | Settings | File Templates.
+ */
+public interface FisherExact {
+    public abstract int[][] getCounts();
+    public abstract String[] getRowNames();
+    public abstract String[] getColNames();
+}

java/src/org/broadinstitute/sting/oneoffprojects/walkers/association/interfaces/TStatistic.java

@@ -0,0 +1,12 @@
+package org.broadinstitute.sting.oneoffprojects.walkers.association.interfaces;
+
+/**
+ * Created by IntelliJ IDEA.
+ * User: chartl
+ * Date: Feb 24, 2011
+ * Time: 12:58:16 PM
+ * To change this template use File | Settings | File Templates.
+ */
+public interface TStatistic {
+    public abstract double getTStatistic();
+}

java/src/org/broadinstitute/sting/oneoffprojects/walkers/association/interfaces/ZStatistic.java

@@ -0,0 +1,12 @@
+package org.broadinstitute.sting.oneoffprojects.walkers.association.interfaces;
+
+/**
+ * Created by IntelliJ IDEA.
+ * User: chartl
+ * Date: Feb 24, 2011
+ * Time: 12:58:38 PM
+ * To change this template use File | Settings | File Templates.
+ */
+public interface ZStatistic {
+    public abstract double getZStatistic();
+}