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

removing some files from the old approach to dataSource 

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@303 348d0f76-0448-11de-a6fe-93d51630548a
This commit is contained in:
aaron 2009-04-06 21:57:34 +00:00
parent 5feb7ee627
commit 0fc8a90553
10 changed files with 0 additions and 576 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

21
java/src/org/broadinstitute/sting/gatk/dataSources/DataSource.java
View File

@ -1,21 +0,0 @@
package org.broadinstitute.sting.gatk.dataSources;
import org.broadinstitute.sting.gatk.dataSources.shards.DataShard;
/**
* User: aaron
* Date: Mar 25, 2009
* Time: 6:20:00 PM
* <p/>
* The Broad Institute
* SOFTWARE COPYRIGHT NOTICE AGREEMENT
* This software and its documentation are copyright 2009 by the
* Broad Institute/Massachusetts Institute of Technology. All rights are reserved.
* <p/>
* This software is supplied without any warranty or guaranteed support whatsoever. Neither
* the Broad Institute nor MIT can be responsible for its use, misuse, or functionality.
*/
public interface DataSource {
public DataShard toChunk(int chunkCount);
}

74
java/src/org/broadinstitute/sting/gatk/dataSources/DataSourceBuilder.java
View File

@ -1,74 +0,0 @@
package org.broadinstitute.sting.gatk.dataSources;
import org.broadinstitute.sting.gatk.walkers.Walker;
import org.broadinstitute.sting.gatk.walkers.ReadWalker;
import java.util.ArrayList;
import java.io.File;
/**
* User: aaron
* Date: Mar 25, 2009
* Time: 4:51:39 PM
* <p/>
* The Broad Institute
* SOFTWARE COPYRIGHT NOTICE AGREEMENT
* This software and its documentation are copyright 2009 by the
* Broad Institute/Massachusetts Institute of Technology. All rights are reserved.
* <p/>
* This software is supplied without any warranty or guaranteed support whatsoever. Neither
* the Broad Institute nor MIT can be responsible for its use, misuse, or functionality.
*/
public class DataSourceBuilder {
// storage for the passed file
ArrayList<File> passFiles = new ArrayList<File>();
public DataSourceBuilder() {
}
/**
* add a file used to generate the data sources
*
* @param fileName the filename that should be used
*/
public void addDataFile(String fileName) {
// for now, just add it to the internal file list
passFiles.add(new File(fileName));
}
/**
* add a file used to generate the data sources
*
* @param file the filename that should be used
*/
public void addDataFile(File file) {
// for now, just add it to the internal file list
passFiles.add(file);
}
public DataSource build(Walker inputWalker) {
if (inputWalker instanceof ReadWalker) {
}
return null;
}
/**
* this section contains the private methods to create data sources
* based on the type of walker we're passed in.
*/
/**
* we know we have a read data source, let's get the
* @return
*/
//private ReadDataSource generateReadDataSource() {
//
//}
}

35
java/src/org/broadinstitute/sting/gatk/dataSources/datum/Datum.java
View File

@ -1,35 +0,0 @@
package org.broadinstitute.sting.gatk.dataSources.datum;
import org.broadinstitute.sting.utils.GenomeLoc;
import java.io.Serializable;
/**
*
* User: aaron
* Date: Mar 30, 2009
* Time: 1:32:34 PM
*
* The Broad Institute
* SOFTWARE COPYRIGHT NOTICE AGREEMENT
* This software and its documentation are copyright 2009 by the
* Broad Institute/Massachusetts Institute of Technology. All rights are reserved.
*
* This software is supplied without any warranty or guaranteed support whatsoever. Neither
* the Broad Institute nor MIT can be responsible for its use, misuse, or functionality.
*
*/
/**
* @author aaron
* @version 1.0
* @date Mar 30, 2009
* <p/>
* interface Datum
* <p/>
* The interface for all Datum Types.
*/
public interface Datum extends Serializable {
// this function is used for tracking where we are in a genome
public GenomeLoc getSequenceLocation();
}

94
java/src/org/broadinstitute/sting/gatk/dataSources/datum/LocusDatum.java
View File

@ -1,94 +0,0 @@
package org.broadinstitute.sting.gatk.dataSources.datum;
import org.broadinstitute.sting.gatk.LocusContext;
import org.broadinstitute.sting.gatk.refdata.ReferenceOrderedDatum;
import org.broadinstitute.sting.utils.GenomeLoc;
import java.util.List;
/**
*
* User: aaron
* Date: Mar 30, 2009
* Time: 3:08:28 PM
*
* The Broad Institute
* SOFTWARE COPYRIGHT NOTICE AGREEMENT
* This software and its documentation are copyright 2009 by the
* Broad Institute/Massachusetts Institute of Technology. All rights are reserved.
*
* This software is supplied without any warranty or guaranteed support whatsoever. Neither
* the Broad Institute nor MIT can be responsible for its use, misuse, or functionality.
*
*/
/**
* @author aaron
* @version 1.0
* @date Mar 30, 2009
* <p/>
* Class LocusDatum
* <p/>
* The datum for loci. It contains the reference base, locusContext,
* and the reference order data.
*/
public class LocusDatum implements Datum {
// our reference order data
private final List<ReferenceOrderedDatum> rodData;
// our seq base
private final char ref;
// our locus context
private final LocusContext context;
/**
* the locus dataum constructor
*
* @param rodData our reference data
* @param ref our reference sequence base position
* @param context the genome context we're in
*/
public LocusDatum(List<ReferenceOrderedDatum> rodData, char ref, LocusContext context) {
this.rodData = rodData;
this.ref = ref;
this.context = context;
}
/**
* return the Reference order data for this position
*
* @return
*/
public List<ReferenceOrderedDatum> getRodData() {
return rodData;
}
/**
* return the reference base
*
* @return a character representing the reference base
*/
public char getRef() {
return ref;
}
/**
* get the locus context at the current position
*
* @return
*/
public LocusContext getContext() {
return context;
}
/**
* gets the current postion in the sequence, which comes
* free from underlying data types
*
* @return our current GenomeLocation
*/
public GenomeLoc getSequenceLocation() {
return this.context.getLocation();
}
}

65
java/src/org/broadinstitute/sting/gatk/dataSources/datum/ReadDatum.java
View File

@ -1,65 +0,0 @@
package org.broadinstitute.sting.gatk.dataSources.datum;
import net.sf.samtools.SAMRecord;
import org.broadinstitute.sting.gatk.LocusContext;
import org.broadinstitute.sting.utils.GenomeLoc;
import org.broadinstitute.sting.utils.Utils;
/**
*
* User: aaron
* Date: Mar 30, 2009
* Time: 2:53:37 PM
*
* The Broad Institute
* SOFTWARE COPYRIGHT NOTICE AGREEMENT
* This software and its documentation are copyright 2009 by the
* Broad Institute/Massachusetts Institute of Technology. All rights are reserved.
*
* This software is supplied without any warranty or guaranteed support whatsoever. Neither
* the Broad Institute nor MIT can be responsible for its use, misuse, or functionality.
*
*/
/**
* @author aaron
* @version 1.0
* @date Mar 30, 2009
* <p/>
* Class ReadDatum
* <p/>
* The base read datum class.
*/
public class ReadDatum implements Datum {
// our SAM record
final private SAMRecord sam;
// our locus context
final private LocusContext locus;
// the constructor, taking a sam read and a locus
public ReadDatum(SAMRecord r, LocusContext locus) {
this.sam = r;
this.locus = locus;
}
// get the SAMRecord
public SAMRecord getRead() {
return this.sam;
}
// get the locus context
public LocusContext getLocus() {
return this.locus;
}
/**
* gets the region that our read spans
*
* @return a genome loc that details the region that our read spans.
*/
public GenomeLoc getSequenceLocation() {
return GenomeLoc.genomicLocationOf(sam);
}
}

36
java/src/org/broadinstitute/sting/gatk/dataSources/shards/BasicDataShard.java
View File

@ -1,36 +0,0 @@
package org.broadinstitute.sting.gatk.dataSources.shards;
import java.util.ArrayList;
import java.util.List;
/**
* Created by IntelliJ IDEA.
* User: aaronmckenna
* Date: Mar 29, 2009
* Time: 8:35:16 PM
* To change this template use File | Settings | File Templates.
*/
public class BasicDataShard<T> implements DataShard {
List<T> list = new ArrayList<T>();
int index = 0;
public BasicDataShard(List<T> list) {
this.list = list;
}
public boolean hasNext() {
if (list.size() > index) {
return true;
}
return false;
}
public T next() {
return list.get(index);
}
public void remove() {
list.remove(index);
}
}

19
java/src/org/broadinstitute/sting/gatk/dataSources/shards/DataShard.java
View File

@ -1,19 +0,0 @@
package org.broadinstitute.sting.gatk.dataSources.shards;
import java.util.Iterator;
/**
* User: aaron
* Date: Mar 26, 2009
* Time: 2:43:04 PM
* <p/>
* The Broad Institute
* SOFTWARE COPYRIGHT NOTICE AGREEMENT
* This software and its documentation are copyright 2009 by the
* Broad Institute/Massachusetts Institute of Technology. All rights are reserved.
* <p/>
* This software is supplied without any warranty or guaranteed support whatsoever. Neither
* the Broad Institute nor MIT can be responsible for its use, misuse, or functionality.
*/
public interface DataShard extends Iterator {
}

99
java/src/org/broadinstitute/sting/gatk/dataSources/shards/LociShard.java
View File

@ -1,99 +0,0 @@
package org.broadinstitute.sting.gatk.dataSources.shards;
import org.broadinstitute.sting.gatk.LocusContext;
import org.broadinstitute.sting.gatk.dataSources.datum.LocusDatum;
import org.broadinstitute.sting.gatk.iterators.LocusIterator;
import org.broadinstitute.sting.gatk.iterators.ReferenceIterator;
import org.broadinstitute.sting.gatk.refdata.ReferenceOrderedData;
import org.broadinstitute.sting.gatk.refdata.ReferenceOrderedDatum;
import org.broadinstitute.sting.utils.GenomeLoc;
import java.util.ArrayList;
import java.util.List;
/**
*
* User: aaron
* Date: Mar 30, 2009
* Time: 7:01:56 PM
*
* The Broad Institute
* SOFTWARE COPYRIGHT NOTICE AGREEMENT
* This software and its documentation are copyright 2009 by the
* Broad Institute/Massachusetts Institute of Technology. All rights are reserved.
*
* This software is supplied without any warranty or guaranteed support whatsoever. Neither
* the Broad Institute nor MIT can be responsible for its use, misuse, or functionality.
*
*/
/**
* @author aaron
* @version 1.0
* @date Mar 30, 2009
* <p/>
* Class LociShard
* <p/>
* This is the loci shard, which are collectively made when a shatter call is made to
* a data source.
*/
public class LociShard implements DataShard {
// our locusIterator
private final LocusIterator locusIterator;
// our reference locusIterator
private final ReferenceIterator refIterator;
// Iterator over rods
private final List<ReferenceOrderedData.RODIterator> rodIters;
// the max number of iterations
private final int maxCount;
// how many iterations we've had
private int iterCount = 0;
public LociShard(LocusIterator locusIterator, ReferenceIterator refIterator, List<ReferenceOrderedData.RODIterator> rodIters, int maxCount) {
this.locusIterator = locusIterator;
this.maxCount = maxCount;
this.refIterator = refIterator;
this.rodIters = rodIters;
}
public boolean hasNext() {
return locusIterator.hasNext() && maxCount > iterCount;
}
public LocusDatum next() {
LocusContext locus = locusIterator.next();
ReferenceIterator refSite = refIterator.seekForward(locus.getLocation());
locus.setReferenceContig(refSite.getCurrentContig());
// Iterate forward to get all reference ordered data covering this locus
final List<ReferenceOrderedDatum> rodData = getReferenceOrderedDataAtLocus(rodIters, locus.getLocation());
return new LocusDatum(rodData, refSite.getBaseAsChar(), locus);
}
public void remove() {
locusIterator.remove();
}
/**
* Builds a list of the reference ordered datum at loc from each of the iterators. This function
* assumes you are accessing the data in order. You can't use this function for random access. Each
* successive call moves you along the file, consuming all data before loc.
*
* @param rodIters Iterators to access the RODs
* @param loc The location to get the rods at
* @return A list of ReferenceOrderDatum at loc. ROD without a datum at loc will be null in the list
*/
protected List<ReferenceOrderedDatum> getReferenceOrderedDataAtLocus(List<ReferenceOrderedData.RODIterator> rodIters,
final GenomeLoc loc) {
List<ReferenceOrderedDatum> data = new ArrayList<ReferenceOrderedDatum>();
for (ReferenceOrderedData.RODIterator iter : rodIters) {
data.add(iter.seekForward(loc));
}
return data;
}
}

85
java/src/org/broadinstitute/sting/gatk/dataSources/shards/ReadShard.java
View File

@ -1,85 +0,0 @@
package org.broadinstitute.sting.gatk.dataSources.shards;
import edu.mit.broad.picard.sam.MergingSamRecordIterator;
import org.broadinstitute.sting.gatk.dataSources.datum.ReadDatum;
import org.broadinstitute.sting.gatk.LocusContext;
import org.broadinstitute.sting.utils.GenomeLoc;
import org.broadinstitute.sting.utils.Utils;
import net.sf.samtools.SAMRecord;
import java.util.List;
import java.util.Arrays;
/**
*
* User: aaron
* Date: Mar 30, 2009
* Time: 5:45:51 PM
*
* The Broad Institute
* SOFTWARE COPYRIGHT NOTICE AGREEMENT
* This software and its documentation are copyright 2009 by the
* Broad Institute/Massachusetts Institute of Technology. All rights are reserved.
*
* This software is supplied without any warranty or guaranteed support whatsoever. Neither
* the Broad Institute nor MIT can be responsible for its use, misuse, or functionality.
*
*/
/**
* @author aaron
* @version 1.0
* @date Mar 30, 2009
* <p/>
* Class ReadShard
* <p/>
* A read data shard.
*/
public class ReadShard implements DataShard {
private MergingSamRecordIterator iterator;
/**
* create the data chunk with an iterator, and a limiter
*
* @param samIterator
*/
public ReadShard(MergingSamRecordIterator samIterator) {
this.iterator = samIterator;
}
/**
* do we have a next data point
*
* @return true if we have a data point
*/
public boolean hasNext() {
return iterator.hasNext();
}
public ReadDatum next() {
// get the read
final SAMRecord read = iterator.next();
// put the read into a list
final List<SAMRecord> reads = Arrays.asList(read);
// put together the genome location
final GenomeLoc loc = GenomeLoc.genomicLocationOf(read);
// Offset of a single read is always 0
List<Integer> offsets = Arrays.asList(0);
// create the locus
final LocusContext locus = new LocusContext(loc, reads, offsets);
// return the read datum
return new ReadDatum(read, locus);
}
/** remove the current pointed to data source */
public void remove() {
iterator.remove();
}
}

48
java/src/org/broadinstitute/sting/gatk/dataSources/shards/SAMDataShard.java
View File

@ -1,48 +0,0 @@
package org.broadinstitute.sting.gatk.dataSources.shards;
import edu.mit.broad.picard.sam.MergingSamRecordIterator;
import net.sf.samtools.SAMRecord;
/**
* Created by IntelliJ IDEA.
* User: aaronmckenna
* Date: Mar 29, 2009
* Time: 8:47:50 PM
* To change this template use File | Settings | File Templates.
*/
public class SAMDataShard implements DataShard {
// our iterator
final private MergingSamRecordIterator iterator;
// divide by reads or by loci
private boolean byReads = true;
// iterator bounds limiter
private int lengthCount = 0;
private final int limiter;
public SAMDataShard(MergingSamRecordIterator iterator, int limiter) {
this.iterator = iterator;
this.limiter = limiter;
}
public SAMDataShard(MergingSamRecordIterator iterator) {
this.iterator = iterator;
limiter = Integer.MAX_VALUE;
}
public boolean hasNext() {
return iterator.hasNext() && lengthCount > limiter;
}
public SAMRecord next() {
++lengthCount;
return iterator.next();
}
public void remove() {
iterator.remove();
}
}