zzh
/
gatk-3.8
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

RODs for reads phase 2: modified RODRecordList to implement List<ReferenceOrderedDatum> so I could stub it out for testing, added a FlashBackIterator which is needed to prevent the ResourcePool from opening infinity+1 iterators, and some other interfaces to make unit testing much smoother. 

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2892 348d0f76-0448-11de-a6fe-93d51630548a
This commit is contained in:
aaron 2010-02-25 22:48:55 +00:00
parent 591102a841
commit 246fa28386
31 changed files with 551 additions and 135 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
java/src/org/broadinstitute/sting/gatk/datasources/providers/ManagingReferenceOrderedView.java
View File

@ -1,8 +1,9 @@
package org.broadinstitute.sting.gatk.datasources.providers;
import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
import org.broadinstitute.sting.gatk.refdata.SeekableRODIterator;
import org.broadinstitute.sting.gatk.datasources.simpleDataSources.ReferenceOrderedDataSource;
import org.broadinstitute.sting.gatk.refdata.utils.FlashBackIterator;
import org.broadinstitute.sting.gatk.refdata.utils.RODRecordList;
import org.broadinstitute.sting.utils.GenomeLoc;
import java.util.List;
@ -37,7 +38,7 @@ public class ManagingReferenceOrderedView implements ReferenceOrderedView {
*/
public ManagingReferenceOrderedView( ShardDataProvider provider ) {
for( ReferenceOrderedDataSource dataSource: provider.getReferenceOrderedData() )
states.add( new ReferenceOrderedDataState( dataSource, (SeekableRODIterator)dataSource.seek(provider.getShard()) ) );
states.add( new ReferenceOrderedDataState( dataSource, ((FlashBackIterator)dataSource.seek(provider.getShard()) )) );
provider.register(this);
}
@ -73,9 +74,9 @@ public class ManagingReferenceOrderedView implements ReferenceOrderedView {
*/
class ReferenceOrderedDataState {
public final ReferenceOrderedDataSource dataSource;
public final SeekableRODIterator iterator;
public final FlashBackIterator iterator;
public ReferenceOrderedDataState( ReferenceOrderedDataSource dataSource, SeekableRODIterator iterator ) {
public ReferenceOrderedDataState( ReferenceOrderedDataSource dataSource, FlashBackIterator iterator ) {
this.dataSource = dataSource;
this.iterator = iterator;
}

8
java/src/org/broadinstitute/sting/gatk/datasources/providers/RodLocusView.java
View File

@ -3,6 +3,8 @@ package org.broadinstitute.sting.gatk.datasources.providers;
import org.broadinstitute.sting.gatk.refdata.*;
import org.broadinstitute.sting.gatk.datasources.simpleDataSources.ReferenceOrderedDataSource;
import org.broadinstitute.sting.gatk.contexts.AlignmentContext;
import org.broadinstitute.sting.gatk.refdata.utils.FlashBackIterator;
import org.broadinstitute.sting.gatk.refdata.utils.RODRecordList;
import org.broadinstitute.sting.utils.GenomeLoc;
import org.broadinstitute.sting.utils.MergingIterator;
import org.broadinstitute.sting.utils.GenomeLocParser;
@ -59,14 +61,14 @@ public class RodLocusView extends LocusView implements ReferenceOrderedView {
GenomeLoc loc = provider.getLocus();
List< Iterator<List<ReferenceOrderedDatum>> > iterators = new LinkedList< Iterator<List<ReferenceOrderedDatum>> >();
List< Iterator<RODRecordList> > iterators = new LinkedList< Iterator<RODRecordList> >();
for( ReferenceOrderedDataSource dataSource: provider.getReferenceOrderedData() ) {
if ( DEBUG ) System.out.printf("Shard is %s%n", provider.getLocus());
// grab the ROD iterator from the data source, and compute the first location in this shard, forwarding
// the iterator to immediately before it, so that it can be added to the merging iterator primed for
// next() to return the first real ROD in this shard
SeekableRODIterator it = (SeekableRODIterator)dataSource.seek(provider.getShard());
FlashBackIterator it = (FlashBackIterator)dataSource.seek(provider.getShard());
it.seekForward(GenomeLocParser.createGenomeLoc(loc.getContigIndex(), loc.getStart()-1));
states.add(new ReferenceOrderedDataState(dataSource,it));
@ -199,7 +201,7 @@ public class RodLocusView extends LocusView implements ReferenceOrderedView {
*/
public void close() {
for( ReferenceOrderedDataState state: states )
state.dataSource.close( state.iterator );
state.dataSource.close( state.iterator );
rodQueue = null;
tracker = null;

42
java/src/org/broadinstitute/sting/gatk/datasources/simpleDataSources/ReferenceOrderedDataSource.java
View File

@ -2,8 +2,8 @@ package org.broadinstitute.sting.gatk.datasources.simpleDataSources;
import org.broadinstitute.sting.gatk.refdata.ReferenceOrderedDatum;
import org.broadinstitute.sting.gatk.refdata.ReferenceOrderedData;
import org.broadinstitute.sting.gatk.refdata.SeekableRODIterator;
import org.broadinstitute.sting.gatk.datasources.shards.Shard;
import org.broadinstitute.sting.gatk.refdata.utils.FlashBackIterator;
import org.broadinstitute.sting.utils.GenomeLoc;
import org.broadinstitute.sting.utils.StingException;
@ -68,15 +68,15 @@ public class ReferenceOrderedDataSource implements SimpleDataSource {
*/
public Iterator seek( Shard shard ) {
DataStreamSegment dataStreamSegment = shard.getGenomeLocs().size() != 0 ? new MappedStreamSegment(shard.getGenomeLocs().get(0)) : new EntireStream();
SeekableRODIterator iterator = iteratorPool.iterator(dataStreamSegment);
return iterator;
FlashBackIterator RODIterator = iteratorPool.iterator(dataStreamSegment);
return RODIterator;
}
/**
* Close the specified iterator, returning it to the pool.
* @param iterator Iterator to close.
*/
public void close( SeekableRODIterator iterator ) {
public void close( FlashBackIterator iterator ) {
this.iteratorPool.release(iterator);
}
@ -85,9 +85,8 @@ public class ReferenceOrderedDataSource implements SimpleDataSource {
/**
* A pool of reference-ordered data iterators.
*/
class ReferenceOrderedDataPool extends ResourcePool<SeekableRODIterator,SeekableRODIterator> {
class ReferenceOrderedDataPool extends ResourcePool<FlashBackIterator, FlashBackIterator> {
private final ReferenceOrderedData<? extends ReferenceOrderedDatum> rod;
public ReferenceOrderedDataPool( ReferenceOrderedData<? extends ReferenceOrderedDatum> rod ) {
this.rod = rod;
}
@ -97,8 +96,8 @@ class ReferenceOrderedDataPool extends ResourcePool<SeekableRODIterator,Seekable
* to be completely independent of any other iterator.
* @return The newly created resource.
*/
public SeekableRODIterator createNewResource() {
return rod.iterator();
public FlashBackIterator createNewResource() {
return new FlashBackIterator(rod.iterator());
}
/**
@ -108,18 +107,21 @@ class ReferenceOrderedDataPool extends ResourcePool<SeekableRODIterator,Seekable
* @param resources @{inheritedDoc}
* @return @{inheritedDoc}
*/
public SeekableRODIterator selectBestExistingResource( DataStreamSegment segment, List<SeekableRODIterator> resources ) {
public FlashBackIterator selectBestExistingResource( DataStreamSegment segment, List<FlashBackIterator> resources ) {
if(segment instanceof MappedStreamSegment) {
GenomeLoc position = ((MappedStreamSegment)segment).getFirstLocation();
//#########################################
//## System.out.printf("Searching for iterator at locus %s; %d resources available%n", position, resources.size());
for( SeekableRODIterator iterator: resources ) {
//##System.out.printf("Examining iterator at position %s [last query location: %s]%n", iterator.position(),iterator.lastQueryLocation());
if( (iterator.position() == null && iterator.hasNext()) ||
(iterator.position() != null && iterator.position().isBefore(position)) )
return iterator;
for( FlashBackIterator RODIterator : resources ) {
if( (RODIterator.position() == null && RODIterator.hasNext()) ||
(RODIterator.position() != null && RODIterator.position().isBefore(position)) )
return RODIterator;
if ((RODIterator.position() != null && RODIterator.canFlashBackTo(position))) {
RODIterator.flashBackTo(position);
return RODIterator;
}
}
//##System.out.printf("Failed to find iterator at locus %s%n", position);
return null;
}
else if(segment instanceof EntireStream) {
@ -135,15 +137,15 @@ class ReferenceOrderedDataPool extends ResourcePool<SeekableRODIterator,Seekable
/**
* In this case, the iterator is the resource. Pass it through.
*/
public SeekableRODIterator createIteratorFromResource( DataStreamSegment segment, SeekableRODIterator resource ) {
public FlashBackIterator createIteratorFromResource( DataStreamSegment segment, FlashBackIterator resource ) {
return resource;
}
/**
* Don't worry about closing the resource; let the file handles expire naturally for the moment.
*/
public void closeResource( SeekableRODIterator resource ) {
public void closeResource( FlashBackIterator resource ) {
}
}

8
java/src/org/broadinstitute/sting/gatk/refdata/RODRecordList.java → java/src/org/broadinstitute/sting/gatk/refdata/RODRecordListImpl.java
View File

@ -1,8 +1,8 @@
package org.broadinstitute.sting.gatk.refdata;
import org.broadinstitute.sting.gatk.refdata.utils.RODRecordList;
import org.broadinstitute.sting.utils.GenomeLoc;
import org.broadinstitute.sting.utils.StingException;
import org.broadinstitute.sting.utils.GenomeLocParser;
import java.util.*;
@ -13,12 +13,12 @@ import java.util.*;
* Time: 6:10:48 PM
* To change this template use File | Settings | File Templates.
*/
public class RODRecordList extends AbstractList<ReferenceOrderedDatum> implements Comparable<RODRecordList>, Cloneable {
public class RODRecordListImpl extends AbstractList<ReferenceOrderedDatum> implements Comparable<RODRecordList>, Cloneable, RODRecordList {
private List<ReferenceOrderedDatum> records;
private GenomeLoc location = null;
private String name = null;
public RODRecordList(String name) {
public RODRecordListImpl(String name) {
records = new ArrayList<ReferenceOrderedDatum>();
this.name = name;
}
@ -37,7 +37,7 @@ public class RODRecordList extends AbstractList<ReferenceOrderedDatum> implement
* @param data
* @param loc
*/
public RODRecordList(String name, Collection<ReferenceOrderedDatum> data, GenomeLoc loc) {
public RODRecordListImpl(String name, Collection<ReferenceOrderedDatum> data, GenomeLoc loc) {
this.records = new ArrayList<ReferenceOrderedDatum>(data==null?0:data.size());
this.name = name;
this.location = loc;

15
java/src/org/broadinstitute/sting/gatk/refdata/RefMetaDataTracker.java
View File

@ -2,6 +2,7 @@ package org.broadinstitute.sting.gatk.refdata;
import org.apache.log4j.Logger;
import org.broadinstitute.sting.gatk.contexts.variantcontext.VariantContext;
import org.broadinstitute.sting.gatk.refdata.utils.RODRecordList;
import org.broadinstitute.sting.utils.GenomeLoc;
import org.broadinstitute.sting.utils.StingException;
@ -43,8 +44,8 @@ public class RefMetaDataTracker {
if ( map.containsKey(luName) ) {
RODRecordList value = map.get(luName) ;
if ( value != null ) {
List<ReferenceOrderedDatum> l = value.getRecords();
if ( l != null & l.size() > 0 ) return value.getRecords().get(0);
List<ReferenceOrderedDatum> l = value;
if ( l != null & l.size() > 0 ) return value.get(0);
}
}
return defaultValue;
@ -76,9 +77,9 @@ public class RefMetaDataTracker {
for ( Map.Entry<String, RODRecordList> datum : map.entrySet() ) {
final String rodName = datum.getKey();
if ( rodName.startsWith(luName) ) {
if ( trackData == null ) trackData = new RODRecordList(name);
if ( trackData == null ) trackData = new RODRecordListImpl(name);
//System.out.printf("Adding bindings from %s to %s at %s%n", rodName, name, datum.getValue().getLocation());
trackData.add(datum.getValue(), true);
((RODRecordListImpl)trackData).add(datum.getValue(), true);
}
}
}
@ -88,7 +89,7 @@ public class RefMetaDataTracker {
else if ( defaultValue == null )
return null;
else
return new RODRecordList(defaultValue.getName(),
return new RODRecordListImpl(defaultValue.getName(),
Collections.singletonList(defaultValue),
defaultValue.getLocation());
}
@ -144,7 +145,7 @@ public class RefMetaDataTracker {
List<ReferenceOrderedDatum> l = new ArrayList<ReferenceOrderedDatum>();
for ( RODRecordList rl : map.values() ) {
if ( rl == null ) continue; // how do we get null value stored for a track? shouldn't the track be missing from the map alltogether?
l.addAll(rl.getRecords());
l.addAll(rl);
}
return l;
@ -285,7 +286,7 @@ public class RefMetaDataTracker {
}
private void addVariantContexts(Collection<VariantContext> contexts, RODRecordList rodList, EnumSet<VariantContext.Type> allowedTypes, GenomeLoc curLocation, boolean requireStartHere, boolean takeFirstOnly ) {
for ( ReferenceOrderedDatum rec : rodList.getRecords() ) {
for ( ReferenceOrderedDatum rec : rodList ) {
if ( VariantContextAdaptors.canBeConvertedToVariantContext(rec) ) {
// ok, we might actually be able to turn this record in a variant context
VariantContext vc = VariantContextAdaptors.toVariantContext(rodList.getName(), rec);

6
java/src/org/broadinstitute/sting/gatk/refdata/ReferenceOrderedData.java
View File

@ -5,6 +5,8 @@ import org.broadinstitute.sting.gatk.refdata.tracks.RMDTrack;
import org.broadinstitute.sting.gatk.refdata.tracks.RMDTrackCreationException;
import org.broadinstitute.sting.gatk.refdata.tracks.RODRMDTrack;
import org.broadinstitute.sting.gatk.refdata.tracks.builders.RMDTrackBuilder;
import org.broadinstitute.sting.gatk.refdata.utils.LocationAwareSeekableRODIterator;
import org.broadinstitute.sting.gatk.refdata.utils.RODRecordList;
import org.broadinstitute.sting.oneoffprojects.refdata.HapmapVCFROD;
import org.broadinstitute.sting.utils.StingException;
import org.broadinstitute.sting.utils.Utils;
@ -21,7 +23,7 @@ import java.util.*;
* Time: 10:47:14 AM
* To change this template use File | Settings | File Templates.
*/
public class ReferenceOrderedData<ROD extends ReferenceOrderedDatum> implements Iterable<List<ReferenceOrderedDatum>> { // }, RMDTrackBuilder {
public class ReferenceOrderedData<ROD extends ReferenceOrderedDatum> implements Iterable<RODRecordList> { // }, RMDTrackBuilder {
private String name;
private File file = null;
// private String fieldDelimiter;
@ -235,7 +237,7 @@ public class ReferenceOrderedData<ROD extends ReferenceOrderedDatum> implements
return this.name.equals(name) && type.isAssignableFrom(this.type);
}
public SeekableRODIterator iterator() {
public LocationAwareSeekableRODIterator iterator() {
Iterator<ReferenceOrderedDatum> it;
try {
Method m = type.getDeclaredMethod("createIterator", String.class, java.io.File.class);

28
java/src/org/broadinstitute/sting/gatk/refdata/SeekableRODIterator.java
View File

@ -1,7 +1,8 @@
package org.broadinstitute.sting.gatk.refdata;
import org.broadinstitute.sting.gatk.iterators.PushbackIterator;
import org.broadinstitute.sting.gatk.iterators.PeekingIterator;
import org.broadinstitute.sting.gatk.refdata.utils.LocationAwareSeekableRODIterator;
import org.broadinstitute.sting.gatk.refdata.utils.RODRecordList;
import org.broadinstitute.sting.utils.GenomeLoc;
import org.broadinstitute.sting.utils.GenomeLocParser;
import org.broadinstitute.sting.utils.StingException;
@ -33,7 +34,7 @@ import java.util.LinkedList;
* Time: 6:20:46 PM
* To change this template use File | Settings | File Templates.
*/
public class SeekableRODIterator implements Iterator<List<ReferenceOrderedDatum>> {
public class SeekableRODIterator implements LocationAwareSeekableRODIterator {
private PushbackIterator<ReferenceOrderedDatum> it;
List<ReferenceOrderedDatum> records = null; // here we will keep a pile of records overlaping with current position; when we iterate
// and step out of record's scope, we purge it from the list
@ -161,12 +162,12 @@ public class SeekableRODIterator implements Iterator<List<ReferenceOrderedDatum>
}
int that_contig = r.getLocation().getContigIndex();
if ( curr_contig > that_contig )
throw new StingException("SeekableRODIterator: contig " +r.getLocation().getContig() +
throw new StingException("LocationAwareSeekableRODIterator: contig " +r.getLocation().getContig() +
" occurs out of order in track " + r.getName() );
if ( curr_contig < that_contig ) break; // next record is on a higher contig, we do not need it yet...
if ( r.getLocation().getStart() < curr_position )
throw new StingException("SeekableRODIterator: track "+r.getName() +
throw new StingException("LocationAwareSeekableRODIterator: track "+r.getName() +
" is out of coordinate order on contig "+r.getLocation().getContig());
if ( r.getLocation().getStart() > curr_position ) break; // next record starts after the current position; we do not need it yet
@ -182,7 +183,7 @@ public class SeekableRODIterator implements Iterator<List<ReferenceOrderedDatum>
// 'records' and current position are fully updated. Last, we need to set the location of the whole track
// (collection of ROD records) to the genomic site we are currently looking at, and return the list
return new RODRecordList(name,records, GenomeLocParser.createGenomeLoc(curr_contig,curr_position));
return new RODRecordListImpl(name,records, GenomeLocParser.createGenomeLoc(curr_contig,curr_position));
}
/**
@ -200,22 +201,9 @@ public class SeekableRODIterator implements Iterator<List<ReferenceOrderedDatum>
* method.
*/
public void remove() {
throw new UnsupportedOperationException("SeekableRODIterator does not implement remove() operation");
throw new UnsupportedOperationException("LocationAwareSeekableRODIterator does not implement remove() operation");
}
/**
*
*/
public GenomeLoc lastQueryLocation() {
if ( curr_contig < 0 ) return null;
if ( curr_query_end > curr_position ) {
return GenomeLocParser.createGenomeLoc(curr_contig,curr_position,curr_query_end);
}
else {
return GenomeLocParser.createGenomeLoc(curr_contig,curr_position);
}
}
/**
* Returns the current "position" (not location!! ;) ) of this iterator. This method is used by the sharding
@ -322,7 +310,7 @@ public class SeekableRODIterator implements Iterator<List<ReferenceOrderedDatum>
}
if ( records.size() > 0 ) {
return new RODRecordList(name,records,interval.clone());
return new RODRecordListImpl(name,records,interval.clone());
} else {
return null;
}

1
java/src/org/broadinstitute/sting/gatk/refdata/rodDbSNP.java
View File

@ -1,6 +1,7 @@
package org.broadinstitute.sting.gatk.refdata;
import net.sf.samtools.util.SequenceUtil;
import org.broadinstitute.sting.gatk.refdata.utils.RODRecordList;
import org.broadinstitute.sting.utils.*;
import java.util.ArrayList;

1
java/src/org/broadinstitute/sting/gatk/refdata/rodRefSeq.java
View File

@ -1,5 +1,6 @@
package org.broadinstitute.sting.gatk.refdata;
import org.broadinstitute.sting.gatk.refdata.utils.RODRecordList;
import org.broadinstitute.sting.utils.GenomeLoc;
import org.broadinstitute.sting.utils.StingException;
import org.broadinstitute.sting.utils.GenomeLocParser;

7
java/src/org/broadinstitute/sting/gatk/refdata/tracks/RODRMDTrack.java
View File

@ -23,10 +23,11 @@
package org.broadinstitute.sting.gatk.refdata.tracks;
import org.broadinstitute.sting.gatk.refdata.RODRecordList;
import org.broadinstitute.sting.gatk.refdata.ReferenceOrderedData;
import org.broadinstitute.sting.gatk.refdata.SeekableRODIterator;
import org.broadinstitute.sting.gatk.refdata.utils.GATKFeature;
import org.broadinstitute.sting.gatk.refdata.utils.LocationAwareSeekableRODIterator;
import org.broadinstitute.sting.gatk.refdata.utils.RODRecordList;
import java.io.File;
import java.util.Iterator;
@ -72,9 +73,9 @@ public class RODRMDTrack extends RMDTrack {
class SRIToIterator implements Iterator<GATKFeature> {
private RODRecordList list = null;
private SeekableRODIterator iterator = null;
private LocationAwareSeekableRODIterator iterator = null;
SRIToIterator(SeekableRODIterator iter) {
SRIToIterator(LocationAwareSeekableRODIterator iter) {
iterator = iter;
}

127
java/src/org/broadinstitute/sting/gatk/refdata/utils/FlashBackIterator.java 100644
View File

@ -0,0 +1,127 @@
package org.broadinstitute.sting.gatk.refdata.utils;
import org.broadinstitute.sting.gatk.refdata.ReferenceOrderedDatum;
import org.broadinstitute.sting.utils.GenomeLoc;
import java.util.Comparator;
import java.util.LinkedList;
import java.util.List;
/**
*
* @author aaron
*
* Class FlashBackIterator
*
* better than acid washed jeans...more like a Delorean that flies through time
*
* This iterator buffers a certain amount of ROD data to 'flash back' to. This
* is needed for using ROD's in read traversals, because between shards we sometimes
* (actually often) need to go back to before the current iterators location and
* get RODs that overlap the current read.
*/
public class FlashBackIterator implements LocationAwareSeekableRODIterator {
private LocationAwareSeekableRODIterator iterator;
private LinkedList<ComparableList> list = new LinkedList<ComparableList>();
private int MAX_QUEUE = 5000;
private boolean usingQueue = false;
public FlashBackIterator(LocationAwareSeekableRODIterator iterator) {
this.iterator = iterator;
}
@Override
public GenomeLoc peekNextLocation() {
return iterator.peekNextLocation();
}
@Override
public GenomeLoc position() {
return (usingQueue) ? list.getFirst().getLocation() : iterator.position();
}
@Override
public RODRecordList seekForward(GenomeLoc interval) {
RODRecordList lt = iterator.seekForward(interval);
if (lt != null) list.addLast(new ComparableList(lt));
return lt;
}
@Override
public boolean hasNext() {
if (usingQueue) return (list.size() > 0 || iterator.hasNext());
return iterator.hasNext();
}
@Override
public RODRecordList next() {
RODRecordList ret;
if (!usingQueue || list.size() < 1) {
usingQueue = false;
ret = iterator.next();
list.addLast(new ComparableList(ret));
if (list.size() > MAX_QUEUE) list.removeFirst();
} else {
ret = list.getFirst().getList();
list.removeFirst();
}
return ret;
}
@Override
public void remove() {
throw new UnsupportedOperationException("We don't support remove");
}
/**
* can we flash back to the specified location?
*
* @param location the location to try and flash back to
*
* @return true if we can, false otherwise
*/
public boolean canFlashBackTo(GenomeLoc location) {
GenomeLoc farthestBack = (list.size() > 0) ? list.getFirst().getLocation() : iterator.peekNextLocation();
System.err.println("farthestBack = " + farthestBack + " loc = " + location);
return (!farthestBack.isPast(location));
}
/**
* flashback! Throws an unsupported operation exception
*
* @param location where to flash back to
*/
public void flashBackTo(GenomeLoc location) {
if (!canFlashBackTo(location)) throw new UnsupportedOperationException("we can't flash back to " + location);
if (list.size() > 0 && !list.getLast().getLocation().isBefore(location))
usingQueue = true;
}
}
class ComparableList implements Comparator<ComparableList> {
private RODRecordList list;
private GenomeLoc location = null;
public ComparableList(RODRecordList list) {
this.list = list;
if (list != null && list.size() != 0) location = list.get(0).getLocation();
else throw new IllegalStateException("Bad voodoo!");
}
@Override
public int compare(ComparableList list1, ComparableList list2) {
if (list1.location == null && list2.location == null)
return 0;
if (list1.location == null) return 1;
if (list2.location == null) return -1;
return (list1.location.compareTo(list2.location));
}
public GenomeLoc getLocation() {
return location;
}
public RODRecordList getList() {
return list;
}
}

23
java/src/org/broadinstitute/sting/gatk/refdata/utils/LocationAwareSeekableRODIterator.java 100644
View File

@ -0,0 +1,23 @@
package org.broadinstitute.sting.gatk.refdata.utils;
import org.broadinstitute.sting.gatk.refdata.ReferenceOrderedDatum;
import org.broadinstitute.sting.utils.GenomeLoc;
import java.util.Iterator;
import java.util.List;
/**
* @author aaron
* <p/>
* Interface LocationAwareSeekableRODIterator
* <p/>
* combine iteration with a position aware interface
*/
public interface LocationAwareSeekableRODIterator extends Iterator<RODRecordList> {
public GenomeLoc peekNextLocation();
public GenomeLoc position();
public RODRecordList seekForward(GenomeLoc interval);
}

20
java/src/org/broadinstitute/sting/gatk/refdata/utils/RODRecordList.java 100644
View File

@ -0,0 +1,20 @@
package org.broadinstitute.sting.gatk.refdata.utils;
import org.broadinstitute.sting.gatk.refdata.ReferenceOrderedDatum;
import org.broadinstitute.sting.utils.GenomeLoc;
import java.util.List;
/**
* @author aaron
* <p/>
* Class RODRecordList
* <p/>
* make the RODRecord list an interface, so we can stub in other implementations
* during testing.
*/
public interface RODRecordList extends List<ReferenceOrderedDatum>, Comparable<RODRecordList> {
public GenomeLoc getLocation();
public String getName();
}

5
java/src/org/broadinstitute/sting/gatk/walkers/annotator/VariantAnnotator.java
View File

@ -2,6 +2,7 @@ package org.broadinstitute.sting.gatk.walkers.annotator;
import org.broadinstitute.sting.gatk.contexts.*;
import org.broadinstitute.sting.gatk.refdata.*;
import org.broadinstitute.sting.gatk.refdata.utils.RODRecordList;
import org.broadinstitute.sting.gatk.walkers.*;
import org.broadinstitute.sting.gatk.datasources.simpleDataSources.ReferenceOrderedDataSource;
import org.broadinstitute.sting.utils.*;
@ -187,11 +188,11 @@ public class VariantAnnotator extends LocusWalker<Integer, Integer> {
RODRecordList rods = tracker.getTrackData("variant", null);
// ignore places where we don't have a variant
if ( rods == null || rods.getRecords().size() == 0 )
if ( rods == null || rods.size() == 0 )
return 0;
Map<String, String> annotations = new HashMap<String, String>();
VariationRod variant = (VariationRod)rods.getRecords().get(0);
VariationRod variant = (VariationRod)rods.get(0);
// if the reference base is not ambiguous, we can annotate
if ( BaseUtils.simpleBaseToBaseIndex(ref.getBase()) != -1 ) {

4
java/src/org/broadinstitute/sting/gatk/walkers/fasta/PickSequenomProbes.java
View File

@ -4,6 +4,8 @@ import org.broadinstitute.sting.gatk.GenomeAnalysisEngine;
import org.broadinstitute.sting.gatk.contexts.AlignmentContext;
import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
import org.broadinstitute.sting.gatk.refdata.*;
import org.broadinstitute.sting.gatk.refdata.utils.LocationAwareSeekableRODIterator;
import org.broadinstitute.sting.gatk.refdata.utils.RODRecordList;
import org.broadinstitute.sting.gatk.walkers.*;
import org.broadinstitute.sting.utils.*;
import org.broadinstitute.sting.utils.cmdLine.Argument;
@ -26,7 +28,7 @@ public class PickSequenomProbes extends RefWalker<String, String> {
String project_id = null;
private byte [] maskFlags = new byte[401];
private SeekableRODIterator snpMaskIterator=null;
private LocationAwareSeekableRODIterator snpMaskIterator=null;
public void initialize() {
if ( SNP_MASK != null ) {

7
java/src/org/broadinstitute/sting/gatk/walkers/filters/VariantFiltrationWalker.java
View File

@ -2,6 +2,7 @@ package org.broadinstitute.sting.gatk.walkers.filters;
import org.broadinstitute.sting.gatk.contexts.*;
import org.broadinstitute.sting.gatk.refdata.*;
import org.broadinstitute.sting.gatk.refdata.utils.RODRecordList;
import org.broadinstitute.sting.gatk.walkers.*;
import org.broadinstitute.sting.gatk.datasources.simpleDataSources.ReferenceOrderedDataSource;
import org.broadinstitute.sting.utils.*;
@ -117,10 +118,10 @@ public class VariantFiltrationWalker extends RodWalker<Integer, Integer> {
RODRecordList rods = tracker.getTrackData("variant", null);
// ignore places where we don't have a variant
if ( rods == null || rods.getRecords().size() == 0 )
if ( rods == null || rods.size() == 0 )
return 0;
RodVCF variant = (RodVCF)rods.getRecords().get(0);
RodVCF variant = (RodVCF)rods.get(0);
Pair<RefMetaDataTracker, RodVCF> varContext = new Pair<RefMetaDataTracker, RodVCF>(tracker, variant);
// if we're still initializing the context, do so
@ -148,7 +149,7 @@ public class VariantFiltrationWalker extends RodWalker<Integer, Integer> {
// test for SNP mask, if present
RODRecordList mask = context.first.getTrackData("mask", null);
if ( mask != null && mask.getRecords().size() > 0 )
if ( mask != null && mask.size() > 0 )
addFilter(filterString, MASK_NAME);
// test for clustered SNPs if requested

6
java/src/org/broadinstitute/sting/gatk/walkers/indels/IndelGenotyperV2Walker.java
View File

@ -1,5 +1,7 @@
package org.broadinstitute.sting.gatk.walkers.indels;
import org.broadinstitute.sting.gatk.refdata.utils.LocationAwareSeekableRODIterator;
import org.broadinstitute.sting.gatk.refdata.utils.RODRecordList;
import org.broadinstitute.sting.gatk.walkers.ReadFilters;
import org.broadinstitute.sting.gatk.walkers.ReadWalker;
import org.broadinstitute.sting.gatk.filters.Platform454Filter;
@ -80,7 +82,7 @@ public class IndelGenotyperV2Walker extends ReadWalker<Integer,Integer> {
boolean outOfContigUserWarned = false;
private SeekableRODIterator refseqIterator=null;
private LocationAwareSeekableRODIterator refseqIterator=null;
private Set<String> normalReadGroups; // we are going to remember which read groups are normals and which are tumors in order to be able
private Set<String> tumorReadGroups ; // to properly assign the reads coming from a merged stream
@ -581,7 +583,7 @@ public class IndelGenotyperV2Walker extends ReadWalker<Integer,Integer> {
else b.append(annUnknown); // we have no idea what this is. this may actually happen when we have a fully non-coding exon...
}
b.append('\t');
b.append(((Transcript)ann.getRecords().get(0)).getGeneName()); // there is at least one transcript in the list, guaranteed
b.append(((Transcript)ann.get(0)).getGeneName()); // there is at least one transcript in the list, guaranteed
// while ( it.hasNext() ) { //
// t.getGeneName()
// }

3
java/src/org/broadinstitute/sting/gatk/walkers/qc/CountRodWalker.java
View File

@ -3,8 +3,7 @@ package org.broadinstitute.sting.gatk.walkers.qc;
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.refdata.ReferenceOrderedDatum;
import org.broadinstitute.sting.gatk.refdata.RODRecordList;
import org.broadinstitute.sting.gatk.refdata.utils.RODRecordList;
import org.broadinstitute.sting.gatk.walkers.RodWalker;
import org.broadinstitute.sting.utils.*;
import org.broadinstitute.sting.utils.cmdLine.Argument;

5
java/src/org/broadinstitute/sting/gatk/walkers/qc/VCFValidator.java
View File

@ -1,13 +1,12 @@
package org.broadinstitute.sting.gatk.walkers.qc;
import org.broadinstitute.sting.gatk.refdata.utils.RODRecordList;
import org.broadinstitute.sting.gatk.walkers.RodWalker;
import org.broadinstitute.sting.gatk.walkers.Requires;
import org.broadinstitute.sting.gatk.walkers.RMD;
import org.broadinstitute.sting.gatk.refdata.RodVCF;
import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
import org.broadinstitute.sting.gatk.refdata.RODRecordList;
import org.broadinstitute.sting.gatk.refdata.ReferenceOrderedDatum;
import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
import org.broadinstitute.sting.gatk.contexts.AlignmentContext;
import org.broadinstitute.sting.utils.StingException;
@ -27,7 +26,7 @@ public class VCFValidator extends RodWalker<Integer, Integer> {
if ( tracker != null ) {
RODRecordList rodlist = tracker.getTrackData("vcf", null);
if ( rodlist != null ) {
RodVCF rod = (RodVCF)rodlist.getRecords().get(0);
RodVCF rod = (RodVCF)rodlist.get(0);
if ( (rod.isSNP() || rod.isReference()) && Character.toUpperCase(rod.getReference().charAt(0)) != Character.toUpperCase(ref.getBase()) )
throw new StingException("The reference base (" + ref.getBase() + ") does not match the base from the VCF record (" + rod.getReference() + ")");
}

3
java/src/org/broadinstitute/sting/gatk/walkers/varianteval/ChipConcordance.java
View File

@ -1,5 +1,6 @@
package org.broadinstitute.sting.gatk.walkers.varianteval;
import org.broadinstitute.sting.gatk.refdata.utils.RODRecordList;
import org.broadinstitute.sting.utils.genotype.Variation;
import org.broadinstitute.sting.utils.genotype.Genotype;
import org.broadinstitute.sting.utils.genotype.VariantBackedByGenotype;
@ -79,7 +80,7 @@ public abstract class ChipConcordance extends BasicVariantAnalysis {
HashMap<String, Genotype> chips = new HashMap<String, Genotype>();
for ( String name : rodNames ) {
RODRecordList rods = tracker.getTrackData(name, null);
Variation chip = (rods == null ? null : (Variation)rods.getRecords().get(0));
Variation chip = (rods == null ? null : (Variation)rods.get(0));
if ( chip != null ) {
// chips must be Genotypes
if ( !(chip instanceof VariantBackedByGenotype) )

1
java/src/org/broadinstitute/sting/gatk/walkers/varianteval/VariantDBCoverage.java
View File

@ -2,6 +2,7 @@ package org.broadinstitute.sting.gatk.walkers.varianteval;
import org.broadinstitute.sting.gatk.contexts.AlignmentContext;
import org.broadinstitute.sting.gatk.refdata.*;
import org.broadinstitute.sting.gatk.refdata.utils.RODRecordList;
import org.broadinstitute.sting.utils.Utils;
import org.broadinstitute.sting.utils.genotype.Variation;

1
java/src/org/broadinstitute/sting/gatk/walkers/varianteval/VariantEvalWalker.java
View File

@ -2,6 +2,7 @@ package org.broadinstitute.sting.gatk.walkers.varianteval;
import org.broadinstitute.sting.gatk.contexts.AlignmentContext;
import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
import org.broadinstitute.sting.gatk.refdata.utils.RODRecordList;
import org.broadinstitute.sting.gatk.walkers.*;
import org.broadinstitute.sting.gatk.refdata.*;
import org.broadinstitute.sting.utils.StingException;

5
java/src/org/broadinstitute/sting/oneoffprojects/walkers/HapmapPoolAllelicInfoWalker.java
View File

@ -1,9 +1,8 @@
package org.broadinstitute.sting.oneoffprojects.walkers;
import org.broadinstitute.sting.gatk.refdata.utils.RODRecordList;
import org.broadinstitute.sting.gatk.walkers.LocusWalker;
import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
import org.broadinstitute.sting.gatk.refdata.ReferenceOrderedDatum;
import org.broadinstitute.sting.gatk.refdata.RODRecordList;
import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
import org.broadinstitute.sting.gatk.contexts.AlignmentContext;
import org.broadinstitute.sting.utils.cmdLine.Argument;
@ -136,7 +135,7 @@ public class HapmapPoolAllelicInfoWalker extends LocusWalker<String, PrintWriter
List<Pair<Genotype, Genotype>> chips = new ArrayList <Pair<Genotype,Genotype>>(rodNames.length);
for ( String name : rodNames ) {
RODRecordList rods = tracker.getTrackData(name, null);
Variation chip = (rods == null ? null : (Variation)rods.getRecords().get(0));
Variation chip = (rods == null ? null : (Variation)rods.get(0));
if ( chip != null ) {
// chips must be Genotypes
if ( !(chip instanceof VariantBackedByGenotype) )

9
java/src/org/broadinstitute/sting/oneoffprojects/walkers/annotator/ThousandGenomesAnnotator.java
View File

@ -2,16 +2,13 @@ package org.broadinstitute.sting.oneoffprojects.walkers.annotator;
import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
import org.broadinstitute.sting.gatk.contexts.StratifiedAlignmentContext;
import org.broadinstitute.sting.gatk.refdata.RODRecordList;
import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
import org.broadinstitute.sting.gatk.refdata.ReferenceOrderedDatum;
import org.broadinstitute.sting.gatk.refdata.utils.RODRecordList;
import org.broadinstitute.sting.gatk.walkers.annotator.VariantAnnotation;
import org.broadinstitute.sting.oneoffprojects.refdata.HapmapVCFROD;
import org.broadinstitute.sting.utils.genotype.Variation;
import org.broadinstitute.sting.utils.genotype.vcf.VCFInfoHeaderLine;
import org.broadinstitute.sting.utils.genotype.vcf.VCFRecord;
import java.util.List;
import java.util.Map;
/**
@ -42,9 +39,9 @@ public class ThousandGenomesAnnotator implements VariantAnnotation {
if ( pilot1 == null && pilot2 == null) {
return "0";
} else {
if ( pilot1 != null && ! ( (HapmapVCFROD) pilot1.getRecords().get(0)).getRecord().isFiltered() ) {
if ( pilot1 != null && ! ( (HapmapVCFROD) pilot1.get(0)).getRecord().isFiltered() ) {
return "1";
} else if ( pilot2 != null && ! ( (HapmapVCFROD) pilot2.getRecords().get(0)).getRecord().isFiltered() ) {
} else if ( pilot2 != null && ! ( (HapmapVCFROD) pilot2.get(0)).getRecord().isFiltered() ) {
return "1";
} else {
return "0";

1
java/src/org/broadinstitute/sting/playground/gatk/walkers/diagnostics/SNPDensity.java
View File

@ -1,5 +1,6 @@
package org.broadinstitute.sting.playground.gatk.walkers.diagnostics;
import org.broadinstitute.sting.gatk.refdata.utils.RODRecordList;
import org.broadinstitute.sting.gatk.walkers.*;
import org.broadinstitute.sting.gatk.refdata.*;
import org.broadinstitute.sting.gatk.contexts.ReferenceContext;

4
java/src/org/broadinstitute/sting/playground/gatk/walkers/hybridselection/HybSelPerformanceWalker.java
View File

@ -2,6 +2,8 @@ package org.broadinstitute.sting.playground.gatk.walkers.hybridselection;
import org.broadinstitute.sting.gatk.contexts.AlignmentContext;
import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
import org.broadinstitute.sting.gatk.refdata.utils.LocationAwareSeekableRODIterator;
import org.broadinstitute.sting.gatk.refdata.utils.RODRecordList;
import org.broadinstitute.sting.gatk.walkers.LocusWalker;
import org.broadinstitute.sting.gatk.walkers.By;
import org.broadinstitute.sting.gatk.walkers.DataSource;
@ -56,7 +58,7 @@ public class HybSelPerformanceWalker extends LocusWalker<Integer, HybSelPerforma
doc="Name of RefSeq transcript annotation file. If specified, intervals will be specified with gene names", required=false)
String REFSEQ_FILE = null;
private SeekableRODIterator refseqIterator=null;
private LocationAwareSeekableRODIterator refseqIterator=null;
public static class TargetInfo {
public int counts = 0;

5
java/src/org/broadinstitute/sting/playground/gatk/walkers/vcftools/VCFSelectWalker.java
View File

@ -2,6 +2,7 @@ package org.broadinstitute.sting.playground.gatk.walkers.vcftools;
import org.broadinstitute.sting.gatk.contexts.*;
import org.broadinstitute.sting.gatk.refdata.*;
import org.broadinstitute.sting.gatk.refdata.utils.RODRecordList;
import org.broadinstitute.sting.gatk.walkers.*;
import org.broadinstitute.sting.gatk.walkers.filters.ClusteredSnps;
import org.broadinstitute.sting.gatk.walkers.filters.VariantContextWindow;
@ -81,10 +82,10 @@ public class VCFSelectWalker extends RodWalker<Integer, Integer> {
RODRecordList rods = tracker.getTrackData("variant", null);
// ignore places where we don't have a variant
if ( rods == null || rods.getRecords().size() == 0 )
if ( rods == null || rods.size() == 0 )
return 0;
RodVCF variant = (RodVCF)rods.getRecords().get(0);
RodVCF variant = (RodVCF)rods.get(0);
boolean someoneMatched = false;
for ( MatchExp exp : matchExpressions ) {
Map<String, String> infoMap = new HashMap<String, String>(variant.mCurrentRecord.getInfoValues());

24
java/src/org/broadinstitute/sting/utils/MergingIterator.java
View File

@ -1,10 +1,8 @@
package org.broadinstitute.sting.utils;
import org.broadinstitute.sting.gatk.iterators.PeekingIterator;
import org.broadinstitute.sting.gatk.iterators.PushbackIterator;
import org.broadinstitute.sting.gatk.refdata.ReferenceOrderedDatum;
import org.broadinstitute.sting.gatk.refdata.SeekableRODIterator;
import org.broadinstitute.sting.gatk.refdata.RODRecordList;
import org.broadinstitute.sting.gatk.refdata.utils.FlashBackIterator;
import org.broadinstitute.sting.gatk.refdata.utils.RODRecordList;
import java.util.*;
@ -12,17 +10,17 @@ public class MergingIterator implements Iterator<RODRecordList>, Iterable<RODRec
PriorityQueue<Element> queue = new PriorityQueue<Element>();
private class Element implements Comparable<Element> {
public SeekableRODIterator it = null;
public FlashBackIterator it = null;
//public E value = null;
public GenomeLoc nextLoc = null;
public Element(Iterator<List<ReferenceOrderedDatum>> it) {
if ( it instanceof SeekableRODIterator ) {
this.it = (SeekableRODIterator)it;
public Element(Iterator<RODRecordList> it) {
if ( it instanceof FlashBackIterator) {
this.it = (FlashBackIterator)it;
if ( ! it.hasNext() ) throw new StingException("Iterator is empty");
update();
} else {
throw new StingException("Iterator passed to MergingIterator is not SeekableRODIterator");
throw new StingException("Iterator passed to MergingIterator is not LocationAwareSeekableRODIterator");
}
}
@ -57,12 +55,12 @@ public class MergingIterator implements Iterator<RODRecordList>, Iterable<RODRec
;
}
public MergingIterator(Iterator<List<ReferenceOrderedDatum>> it) {
public MergingIterator(Iterator<RODRecordList> it) {
add(it);
}
public MergingIterator(Collection<Iterator<List<ReferenceOrderedDatum>>> its) {
for ( Iterator<List<ReferenceOrderedDatum>> it : its ) {
public MergingIterator(Collection<Iterator<RODRecordList>> its) {
for ( Iterator<RODRecordList> it : its ) {
add(it);
}
}
@ -71,7 +69,7 @@ public class MergingIterator implements Iterator<RODRecordList>, Iterable<RODRec
* will be after a call to next() is peeked into and cached as queue's priority value.
* @param it
*/
public void add(Iterator<List<ReferenceOrderedDatum>> it) {
public void add(Iterator<RODRecordList> it) {
if ( it.hasNext() )
queue.add(new Element(it));
}

29
java/test/org/broadinstitute/sting/gatk/datasources/simpleDataSources/ReferenceOrderedDataPoolTest.java
View File

@ -1,5 +1,6 @@
package org.broadinstitute.sting.gatk.datasources.simpleDataSources;
import org.broadinstitute.sting.gatk.refdata.utils.FlashBackIterator;
import org.junit.Before;
import org.junit.BeforeClass;
import org.junit.Test;
@ -55,12 +56,12 @@ public class ReferenceOrderedDataPoolTest extends BaseTest {
@Test
public void testCreateSingleIterator() {
ResourcePool iteratorPool = new ReferenceOrderedDataPool(rod);
SeekableRODIterator iterator = (SeekableRODIterator)iteratorPool.iterator( new MappedStreamSegment(testSite1) );
FlashBackIterator iterator = (FlashBackIterator)iteratorPool.iterator( new MappedStreamSegment(testSite1) );
Assert.assertEquals("Number of iterators in the pool is incorrect", 1, iteratorPool.numIterators());
Assert.assertEquals("Number of available iterators in the pool is incorrect", 0, iteratorPool.numAvailableIterators());
TabularROD datum = (TabularROD)iterator.next().getRecords().get(0);
TabularROD datum = (TabularROD)iterator.next().get(0);
assertTrue(datum.getLocation().equals(testSite1));
assertTrue(datum.get("COL1").equals("A"));
@ -76,36 +77,36 @@ public class ReferenceOrderedDataPoolTest extends BaseTest {
@Test
public void testCreateMultipleIterators() {
ReferenceOrderedDataPool iteratorPool = new ReferenceOrderedDataPool(rod);
SeekableRODIterator iterator1 = iteratorPool.iterator( new MappedStreamSegment(testSite1) );
FlashBackIterator iterator1 = iteratorPool.iterator( new MappedStreamSegment(testSite1) );
// Create a new iterator at position 2.
SeekableRODIterator iterator2 = iteratorPool.iterator( new MappedStreamSegment(testSite2) );
FlashBackIterator iterator2 = iteratorPool.iterator( new MappedStreamSegment(testSite2) );
Assert.assertEquals("Number of iterators in the pool is incorrect", 2, iteratorPool.numIterators());
Assert.assertEquals("Number of available iterators in the pool is incorrect", 0, iteratorPool.numAvailableIterators());
// Test out-of-order access: first iterator2, then iterator1.
// Ugh...first call to a region needs to be a seek.
TabularROD datum = (TabularROD)iterator2.seekForward(testSite2).getRecords().get(0);
TabularROD datum = (TabularROD)iterator2.seekForward(testSite2).get(0);
assertTrue(datum.getLocation().equals(testSite2));
assertTrue(datum.get("COL1").equals("C"));
assertTrue(datum.get("COL2").equals("D"));
assertTrue(datum.get("COL3").equals("E"));
datum = (TabularROD)iterator1.next().getRecords().get(0);
datum = (TabularROD)iterator1.next().get(0);
assertTrue(datum.getLocation().equals(testSite1));
assertTrue(datum.get("COL1").equals("A"));
assertTrue(datum.get("COL2").equals("B"));
assertTrue(datum.get("COL3").equals("C"));
// Advance iterator2, and make sure both iterator's contents are still correct.
datum = (TabularROD)iterator2.next().getRecords().get(0);
datum = (TabularROD)iterator2.next().get(0);
assertTrue(datum.getLocation().equals(testSite3));
assertTrue(datum.get("COL1").equals("F"));
assertTrue(datum.get("COL2").equals("G"));
assertTrue(datum.get("COL3").equals("H"));
datum = (TabularROD)iterator1.next().getRecords().get(0);
datum = (TabularROD)iterator1.next().get(0);
assertTrue(datum.getLocation().equals(testSite2));
assertTrue(datum.get("COL1").equals("C"));
assertTrue(datum.get("COL2").equals("D"));
@ -126,12 +127,12 @@ public class ReferenceOrderedDataPoolTest extends BaseTest {
@Test
public void testIteratorConservation() {
ReferenceOrderedDataPool iteratorPool = new ReferenceOrderedDataPool(rod);
SeekableRODIterator iterator = iteratorPool.iterator( new MappedStreamSegment(testSite1) );
FlashBackIterator iterator = iteratorPool.iterator( new MappedStreamSegment(testSite1) );
Assert.assertEquals("Number of iterators in the pool is incorrect", 1, iteratorPool.numIterators());
Assert.assertEquals("Number of available iterators in the pool is incorrect", 0, iteratorPool.numAvailableIterators());
TabularROD datum = (TabularROD)iterator.next().getRecords().get(0);
TabularROD datum = (TabularROD)iterator.next().get(0);
assertTrue(datum.getLocation().equals(testSite1));
assertTrue(datum.get("COL1").equals("A"));
assertTrue(datum.get("COL2").equals("B"));
@ -146,7 +147,7 @@ public class ReferenceOrderedDataPoolTest extends BaseTest {
Assert.assertEquals("Number of iterators in the pool is incorrect", 1, iteratorPool.numIterators());
Assert.assertEquals("Number of available iterators in the pool is incorrect", 0, iteratorPool.numAvailableIterators());
datum = (TabularROD)iterator.seekForward(testSite3).getRecords().get(0);
datum = (TabularROD)iterator.seekForward(testSite3).get(0);
assertTrue(datum.getLocation().equals(testSite3));
assertTrue(datum.get("COL1").equals("F"));
assertTrue(datum.get("COL2").equals("G"));
@ -161,12 +162,12 @@ public class ReferenceOrderedDataPoolTest extends BaseTest {
@Test
public void testIteratorCreation() {
ReferenceOrderedDataPool iteratorPool = new ReferenceOrderedDataPool(rod);
SeekableRODIterator iterator = iteratorPool.iterator( new MappedStreamSegment(testSite3) );
FlashBackIterator iterator = iteratorPool.iterator( new MappedStreamSegment(testSite3) );
Assert.assertEquals("Number of iterators in the pool is incorrect", 1, iteratorPool.numIterators());
Assert.assertEquals("Number of available iterators in the pool is incorrect", 0, iteratorPool.numAvailableIterators());
TabularROD datum = (TabularROD)iterator.seekForward(testSite3).getRecords().get(0);
TabularROD datum = (TabularROD)iterator.seekForward(testSite3).get(0);
assertTrue(datum.getLocation().equals(testSite3));
assertTrue(datum.get("COL1").equals("F"));
assertTrue(datum.get("COL2").equals("G"));
@ -181,7 +182,7 @@ public class ReferenceOrderedDataPoolTest extends BaseTest {
Assert.assertEquals("Number of iterators in the pool is incorrect", 2, iteratorPool.numIterators());
Assert.assertEquals("Number of available iterators in the pool is incorrect", 1, iteratorPool.numAvailableIterators());
datum = (TabularROD)iterator.next().getRecords().get(0);
datum = (TabularROD)iterator.next().get(0);
assertTrue(datum.getLocation().equals(testSite1));
assertTrue(datum.get("COL1").equals("A"));
assertTrue(datum.get("COL2").equals("B"));

40
java/test/org/broadinstitute/sting/gatk/refdata/TabularRODTest.java
View File

@ -4,6 +4,8 @@ package org.broadinstitute.sting.gatk.refdata;
// the imports for unit testing.
import org.broadinstitute.sting.gatk.refdata.utils.LocationAwareSeekableRODIterator;
import org.broadinstitute.sting.gatk.refdata.utils.RODRecordList;
import org.junit.*;
import static org.junit.Assert.assertTrue;
import org.broadinstitute.sting.BaseTest;
@ -26,7 +28,7 @@ import net.sf.picard.reference.ReferenceSequenceFile;
public class TabularRODTest extends BaseTest {
private static ReferenceSequenceFile seq;
private ReferenceOrderedData ROD;
private SeekableRODIterator iter;
private LocationAwareSeekableRODIterator iter;
@BeforeClass
@ -49,7 +51,7 @@ public class TabularRODTest extends BaseTest {
public void test1() {
logger.warn("Executing test1");
RODRecordList oneList = iter.next();
TabularROD one = (TabularROD)oneList.getRecords().get(0);
TabularROD one = (TabularROD)oneList.get(0);
assertTrue(one.size() == 4);
assertTrue(one.getLocation().equals(GenomeLocParser.createGenomeLoc("chrM", 10)));
assertTrue(one.get("COL1").equals("A"));
@ -62,8 +64,8 @@ public class TabularRODTest extends BaseTest {
logger.warn("Executing test2");
RODRecordList oneList = iter.next();
RODRecordList twoList = iter.next();
TabularROD one = (TabularROD)oneList.getRecords().get(0);
TabularROD two = (TabularROD)twoList.getRecords().get(0);
TabularROD one = (TabularROD)oneList.get(0);
TabularROD two = (TabularROD)twoList.get(0);
assertTrue(two.size() == 4);
assertTrue(two.getLocation().equals(GenomeLocParser.createGenomeLoc("chrM", 20)));
assertTrue(two.get("COL1").equals("C"));
@ -77,9 +79,9 @@ public class TabularRODTest extends BaseTest {
RODRecordList oneList = iter.next();
RODRecordList twoList = iter.next();
RODRecordList threeList = iter.next();
TabularROD one = (TabularROD)oneList.getRecords().get(0);
TabularROD two = (TabularROD)twoList.getRecords().get(0);
TabularROD three = (TabularROD)threeList.getRecords().get(0);
TabularROD one = (TabularROD)oneList.get(0);
TabularROD two = (TabularROD)twoList.get(0);
TabularROD three = (TabularROD)threeList.get(0);
assertTrue(three.size() == 4);
assertTrue(three.getLocation().equals(GenomeLocParser.createGenomeLoc("chrM", 30)));
assertTrue(three.get("COL1").equals("F"));
@ -93,9 +95,9 @@ public class TabularRODTest extends BaseTest {
RODRecordList oneList = iter.next();
RODRecordList twoList = iter.next();
RODRecordList threeList = iter.next();
TabularROD one = (TabularROD)oneList.getRecords().get(0);
TabularROD two = (TabularROD)twoList.getRecords().get(0);
TabularROD three = (TabularROD)threeList.getRecords().get(0);
TabularROD one = (TabularROD)oneList.get(0);
TabularROD two = (TabularROD)twoList.get(0);
TabularROD three = (TabularROD)threeList.get(0);
assertTrue(!iter.hasNext());
}
@ -103,7 +105,7 @@ public class TabularRODTest extends BaseTest {
public void testSeek() {
logger.warn("Executing testSeek");
RODRecordList twoList = iter.seekForward(GenomeLocParser.createGenomeLoc("chrM", 20));
TabularROD two = (TabularROD)twoList.getRecords().get(0);
TabularROD two = (TabularROD)twoList.get(0);
assertTrue(two.size() == 4);
assertTrue(two.getLocation().equals(GenomeLocParser.createGenomeLoc("chrM", 20)));
assertTrue(two.get("COL1").equals("C"));
@ -115,7 +117,7 @@ public class TabularRODTest extends BaseTest {
public void testToString() {
logger.warn("Executing testToString");
RODRecordList oneList = iter.next();
TabularROD one = (TabularROD)oneList.getRecords().get(0);
TabularROD one = (TabularROD)oneList.get(0);
assertTrue(one.toString().equals("chrM:10\tA\tB\tC"));
}
@ -124,11 +126,11 @@ public class TabularRODTest extends BaseTest {
public void testDelim1() {
File file2 = new File(testDir + "TabularDataTest2.dat");
ReferenceOrderedData ROD_commas = new ReferenceOrderedData("tableTest", file2, TabularROD.class);
SeekableRODIterator iter_commas = ROD_commas.iterator();
LocationAwareSeekableRODIterator iter_commas = ROD_commas.iterator();
logger.warn("Executing testDelim1");
RODRecordList one2List = iter_commas.next();
TabularROD one2 = (TabularROD)one2List.getRecords().get(0);
TabularROD one2 = (TabularROD)one2List.get(0);
assertTrue(one2.size() == 5);
assertTrue(one2.getLocation().equals(GenomeLocParser.createGenomeLoc("chrM", 10)));
assertTrue(one2.get("COL1").equals("A"));
@ -142,11 +144,11 @@ public class TabularRODTest extends BaseTest {
TabularROD.setDelimiter(",",",");
File file2 = new File(testDir + "TabularDataTest2.dat");
ReferenceOrderedData ROD_commas = new ReferenceOrderedData("tableTest", file2, TabularROD.class);
SeekableRODIterator iter_commas = ROD_commas.iterator();
LocationAwareSeekableRODIterator iter_commas = ROD_commas.iterator();
logger.warn("Executing testDelim1");
RODRecordList one2List = iter_commas.next();
TabularROD one2 = (TabularROD)one2List.getRecords().get(0);
TabularROD one2 = (TabularROD)one2List.get(0);
assertTrue(one2.size() == 5);
assertTrue(one2.getLocation().equals(GenomeLocParser.createGenomeLoc("chrM", 10)));
assertTrue(one2.get("COL1").equals("A"));
@ -187,10 +189,10 @@ public class TabularRODTest extends BaseTest {
out.println(row.toString());
ReferenceOrderedData ROD_commas = new ReferenceOrderedData("tableTest", outputFile, TabularROD.class);
SeekableRODIterator iter_commas = ROD_commas.iterator();
LocationAwareSeekableRODIterator iter_commas = ROD_commas.iterator();
RODRecordList oneList = iter_commas.next();
TabularROD one = (TabularROD)oneList.getRecords().get(0);
TabularROD one = (TabularROD)oneList.get(0);
assertTrue(one.size() == 4);
assertTrue(one.getLocation().equals(GenomeLocParser.createGenomeLoc("chrM", 1)));
assertTrue(one.get("col1").equals("1"));
@ -198,7 +200,7 @@ public class TabularRODTest extends BaseTest {
assertTrue(one.get("col3").equals("3"));
RODRecordList twoList = iter_commas.next();
TabularROD two = (TabularROD)twoList.getRecords().get(0);
TabularROD two = (TabularROD)twoList.get(0);
assertTrue(two.size() == 4);
assertTrue(two.getLocation().equals(GenomeLocParser.createGenomeLoc("chrM", 2)));
assertTrue(two.get("col1").equals("3"));

239
java/test/org/broadinstitute/sting/gatk/refdata/utils/FlashBackIteratorTest.java 100644
View File

@ -0,0 +1,239 @@
package org.broadinstitute.sting.gatk.refdata.utils;
import net.sf.samtools.SAMFileHeader;
import org.broadinstitute.sting.BaseTest;
import org.broadinstitute.sting.gatk.refdata.ReferenceOrderedDatum;
import org.broadinstitute.sting.utils.GenomeLoc;
import org.broadinstitute.sting.utils.GenomeLocParser;
import org.broadinstitute.sting.utils.GenomeLocSortedSet;
import org.broadinstitute.sting.utils.sam.ArtificialSAMUtils;
import org.junit.Assert;
import org.junit.Before;
import org.junit.Test;
import java.io.File;
import java.io.FileNotFoundException;
import java.io.IOException;
import java.util.AbstractList;
import java.util.ArrayList;
import java.util.List;
/**
* @author aaron
* <p/>
* Class FlashBackIteratorTest
* <p/>
* just like a greatful dead show...this will be prone to flashbacks
*/
public class FlashBackIteratorTest extends BaseTest {
private SAMFileHeader header = ArtificialSAMUtils.createArtificialSamHeader(NUMBER_OF_CHROMOSOMES, STARTING_CHROMOSOME, CHROMOSOME_SIZE);
private static final int NUMBER_OF_CHROMOSOMES = 5;
private static final int STARTING_CHROMOSOME = 1;
private static final int CHROMOSOME_SIZE = 1000;
@Before
public void setup() {
GenomeLocParser.setupRefContigOrdering(header.getSequenceDictionary());
}
@Test
public void testBasicIteration() {
GenomeLoc loc = GenomeLocParser.createGenomeLoc(0,0,0);
FlashBackIterator iter = new FlashBackIterator(new FakeSeekableRODIterator(loc));
GenomeLoc lastLocation = null;
for (int x = 0; x < 10; x++) {
iter.next();
GenomeLoc cur = iter.position();
if (lastLocation != null) {
Assert.assertTrue(lastLocation.isBefore(cur));
}
lastLocation = cur;
}
}
@Test
public void testBasicIterationThenFlashBack() {
GenomeLoc loc = GenomeLocParser.createGenomeLoc(0,0,0);
FlashBackIterator iter = new FlashBackIterator(new FakeSeekableRODIterator(loc));
GenomeLoc lastLocation = null;
for (int x = 0; x < 10; x++) {
iter.next();
GenomeLoc cur = iter.position();
if (lastLocation != null) {
Assert.assertTrue(lastLocation.isBefore(cur));
}
lastLocation = cur;
}
iter.flashBackTo(GenomeLocParser.createGenomeLoc(0,2));
}
@Test
public void testBasicIterationThenFlashBackThenIterate() {
GenomeLoc loc = GenomeLocParser.createGenomeLoc(0,0,0);
FlashBackIterator iter = new FlashBackIterator(new FakeSeekableRODIterator(loc));
GenomeLoc lastLocation = null;
for (int x = 0; x < 10; x++) {
iter.next();
GenomeLoc cur = iter.position();
if (lastLocation != null) {
Assert.assertTrue(lastLocation.isBefore(cur));
}
lastLocation = cur;
}
iter.flashBackTo(GenomeLocParser.createGenomeLoc(0,1));
int count = 0;
while (iter.hasNext()) {
count++;
iter.next();
}
Assert.assertEquals(10,count);
}
}
class FakeSeekableRODIterator implements LocationAwareSeekableRODIterator {
// current location
private GenomeLoc location;
private FakeRODatum curROD;
private int recordCount = 10;
public FakeSeekableRODIterator(GenomeLoc startingLoc) {
this.location = GenomeLocParser.createGenomeLoc(startingLoc.getContigIndex(),startingLoc.getStart()+1,startingLoc.getStop()+1);;
}
@Override
public GenomeLoc peekNextLocation() {
System.err.println("Peek Next -> " + location);
return location;
}
@Override
public GenomeLoc position() {
return location;
}
@Override
public RODRecordList seekForward(GenomeLoc interval) {
this.location = interval;
return next();
}
@Override
public boolean hasNext() {
return (recordCount > 0);
}
@Override
public RODRecordList next() {
RODRecordList list = new FakeRODRecordList();
curROD = new FakeRODatum(location);
location = GenomeLocParser.createGenomeLoc(location.getContigIndex(),location.getStart()+1,location.getStop()+1);
list.add(curROD);
recordCount--;
return list;
}
@Override
public void remove() {
throw new IllegalStateException("GRRR");
}
}
/** for testing only */
class FakeRODatum implements ReferenceOrderedDatum {
final GenomeLoc location;
public FakeRODatum(GenomeLoc location) {
this.location = location;
}
@Override
public String getName() {
return "false";
}
@Override
public boolean parseLine(Object header, String[] parts) throws IOException {
return false;
}
@Override
public String toSimpleString() {
return "";
}
@Override
public String repl() {
return "";
}
/**
* Used by the ROD system to determine how to split input lines
*
* @return Regex string delimiter separating fields
*/
@Override
public String delimiterRegex() {
return "";
}
@Override
public GenomeLoc getLocation() {
return location;
}
@Override
public int compareTo(ReferenceOrderedDatum that) {
return location.compareTo(that.getLocation());
}
/**
* Backdoor hook to read header, meta-data, etc. associated with the file. Will be
* called by the ROD system before streaming starts
*
* @param source source data file on disk from which this rod stream will be pulled
*
* @return a header object that will be passed to parseLine command
*/
@Override
public Object initialize(File source) throws FileNotFoundException {
return null;
}
}
class FakeRODRecordList extends AbstractList<ReferenceOrderedDatum> implements RODRecordList {
private final List<ReferenceOrderedDatum> list = new ArrayList<ReferenceOrderedDatum>();
public boolean add(ReferenceOrderedDatum data) {
return list.add(data);
}
@Override
public ReferenceOrderedDatum get(int i) {
return list.get(i);
}
@Override
public int size() {
return list.size();
}
@Override
public GenomeLoc getLocation() {
return list.get(0).getLocation();
}
@Override
public String getName() {
return "test";
}
@Override
public int compareTo(RODRecordList rodRecordList) {
return this.list.get(0).getLocation().compareTo(rodRecordList.getLocation());
}
}