Fare thee well, oh LocusWindowTraversal.

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@3089 348d0f76-0448-11de-a6fe-93d51630548a
This commit is contained in:
ebanks 2010-03-29 13:17:26 +00:00
parent 8ea98faf47
commit 1e8b3ca6ba
4 changed files with 0 additions and 145 deletions

View File

@ -744,17 +744,6 @@ public class GenomeAnalysisEngine {
ShardStrategyFactory.SHATTER_STRATEGY.READS_EXPERIMENTAL,
drivingDataSource.getSequenceDictionary(),
SHARD_SIZE, maxIterations);
} else if (walker instanceof LocusWindowWalker) {
if ((intervals == null || intervals.isEmpty()) && !exclusions.contains(ValidationExclusion.TYPE.ALLOW_EMPTY_INTERVAL_LIST))
Utils.warnUser("walker is of type LocusWindow (which operates over intervals), but no intervals were provided." +
"This may be unintentional, check your command-line arguments.");
shardStrategy = ShardStrategyFactory.shatter(readsDataSource,
referenceDataSource,
ShardStrategyFactory.SHATTER_STRATEGY.INTERVAL,
drivingDataSource.getSequenceDictionary(),
SHARD_SIZE,
intervals, maxIterations);
} else
throw new StingException("Unable to support walker of type" + walker.getClass().getName());

View File

@ -38,7 +38,6 @@ import org.broadinstitute.sting.gatk.iterators.NullSAMIterator;
import org.broadinstitute.sting.gatk.Reads;
import org.broadinstitute.sting.gatk.GenomeAnalysisEngine;
import org.broadinstitute.sting.utils.fasta.IndexedFastaSequenceFile;
import org.broadinstitute.sting.utils.GenomeLocSortedSet;
import java.util.*;
import java.io.File;
@ -107,8 +106,6 @@ public abstract class MicroScheduler {
traversalEngine = new TraverseReads();
} else if (walker instanceof LocusWalker) {
traversalEngine = new TraverseLoci();
} else if (walker instanceof LocusWindowWalker) {
traversalEngine = new TraverseLocusWindows();
} else if (walker instanceof DuplicateWalker) {
traversalEngine = new TraverseDuplicates();
} else if (walker instanceof ReadPairWalker) {

View File

@ -1,106 +0,0 @@
package org.broadinstitute.sting.gatk.traversals;
import net.sf.samtools.SAMRecord;
import org.broadinstitute.sting.gatk.datasources.providers.*;
import org.broadinstitute.sting.gatk.iterators.StingSAMIterator;
import org.broadinstitute.sting.gatk.iterators.LocusIterator;
import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
import org.broadinstitute.sting.gatk.walkers.LocusWindowWalker;
import org.broadinstitute.sting.gatk.contexts.AlignmentContext;
import org.broadinstitute.sting.utils.GenomeLoc;
import org.broadinstitute.sting.utils.GenomeLocParser;
import org.broadinstitute.sting.utils.Pair;
import java.util.*;
/**
* Created by IntelliJ IDEA.
* User: ebanks
* Date: Apr 23, 2009
* Time: 10:26:03 AM
* To change this template use File | Settings | File Templates.
*/
public class TraverseLocusWindows<M,T> extends TraversalEngine<M,T,LocusWindowWalker<M,T>,LocusShardDataProvider> {
/** descriptor of the type */
private static final String LOCUS_WINDOW_STRING = "intervals";
public T traverse( LocusWindowWalker<M,T> walker,
LocusShardDataProvider dataProvider,
T sum ) {
GenomeLoc interval = dataProvider.getLocus();
LocusReferenceView referenceView = new LocusReferenceView( walker, dataProvider );
ReferenceOrderedView referenceOrderedDataView = new ManagingReferenceOrderedView( dataProvider );
Pair<GenomeLoc, List<SAMRecord>> locus = getLocusContext(dataProvider.getLocusIterator(), interval);
// The TraverseByLocusWindow expands intervals to cover all reads in a non-standard way.
// TODO: Convert this approach to the standard.
GenomeLoc expandedInterval = locus.getFirst();
String referenceSubsequence = new String(referenceView.getReferenceBases(expandedInterval));
// Iterate forward to get all reference ordered data covering this interval
final RefMetaDataTracker tracker = referenceOrderedDataView.getReferenceOrderedDataAtLocus(locus.getFirst());
//
// Execute our contract with the walker. Call filter, map, and reduce
//
//final boolean keepMeP = locusWindowWalker.filter(tracker, referenceSubsequence, locus);
//if (keepMeP) {
M x = walker.map(tracker, referenceSubsequence, locus.getFirst(), locus.getSecond());
sum = walker.reduce(x, sum);
//}
printProgress(LOCUS_WINDOW_STRING, locus.getFirst());
return sum;
}
private Pair<GenomeLoc, List<SAMRecord>> getLocusContext(LocusIterator locusIter, GenomeLoc interval) {
List<SAMRecord> reads = new ArrayList<SAMRecord>();
boolean done = false;
long leftmostIndex = interval.getStart(),
rightmostIndex = interval.getStop();
while(locusIter.hasNext() && !done) {
AlignmentContext alignment = locusIter.next();
Iterator<SAMRecord> readIter = alignment.getReads().iterator();
while (readIter.hasNext() && !done) {
TraversalStatistics.nRecords++;
SAMRecord read = readIter.next();
if(reads.contains(read)) continue;
reads.add(read);
if ( read.getAlignmentStart() < leftmostIndex )
leftmostIndex = read.getAlignmentStart();
if ( read.getAlignmentEnd() > rightmostIndex )
rightmostIndex = read.getAlignmentEnd();
if ( this.maximumIterations > 0 && TraversalStatistics.nRecords > this.maximumIterations) {
logger.warn(String.format("Maximum number of reads encountered, terminating traversal " + TraversalStatistics.nRecords));
done = true;
}
}
}
GenomeLoc window = GenomeLocParser.createGenomeLoc(interval.getContig(), leftmostIndex, rightmostIndex);
// AlignmentContext locus = new AlignmentContext(window, reads, null);
// if ( readIter.getSourceInfo().getDownsampleToCoverage() != null )
// locus.downsampleToCoverage(readIter.getSourceInfo().getDownsampleToCoverage());
return new Pair<GenomeLoc, List<SAMRecord>>(window, reads);
}
/**
* Temporary override of printOnTraversalDone.
* TODO: Add some sort of TE.getName() function once all TraversalEngines are ported.
* @param sum Result of the computation.
*/
public void printOnTraversalDone( T sum ) {
printOnTraversalDone(LOCUS_WINDOW_STRING, sum );
}
}

View File

@ -1,25 +0,0 @@
package org.broadinstitute.sting.gatk.walkers;
import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
import org.broadinstitute.sting.gatk.contexts.AlignmentContext;
import org.broadinstitute.sting.utils.GenomeLoc;
import net.sf.samtools.SAMRecord;
import java.util.List;
/**
* Created by IntelliJ IDEA.
* User: ebanks
* Date: Apr 23, 2009
* Time: 2:52:28 PM
* To change this template use File | Settings | File Templates.
*/
@Requires({DataSource.READS,DataSource.REFERENCE, DataSource.REFERENCE_BASES})
public abstract class LocusWindowWalker<MapType, ReduceType> extends Walker<MapType, ReduceType> {
// Map over the org.broadinstitute.sting.gatk.contexts.AlignmentContext
public abstract MapType map(RefMetaDataTracker tracker, String ref, GenomeLoc loc, List<SAMRecord> reads);
// Given result of map function
public abstract ReduceType reduceInit();
public abstract ReduceType reduce(MapType value, ReduceType sum);
}