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

cleanup of the traversal output code 

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2026 348d0f76-0448-11de-a6fe-93d51630548a
This commit is contained in:
aaron 2009-11-12 06:18:10 +00:00
parent 0922400ca9
commit c3c001e02e
8 changed files with 32 additions and 35 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
java/src/org/broadinstitute/sting/gatk/GenomeAnalysisEngine.java
View File

@ -309,7 +309,8 @@ public class GenomeAnalysisEngine {
* Returns sets of samples present in the (merged) input SAM stream, grouped by readers (i.e. underlying
* individual bam files). For instance: if GATK is run with three input bam files (three -I arguments), then the list
* returned by this method will contain 3 elements (one for each reader), with each element being a set of sample names
* found in the corresponding bam file.
* found in the corresponding bam file. The sample names returned will be in order of the files passed in the input
* parameter list of the GATK.
*
* @return
*/

22
java/src/org/broadinstitute/sting/gatk/traversals/TraversalEngine.java
View File

@ -21,9 +21,6 @@ public abstract class TraversalEngine {
/** our log, which we want to capture anything from this class */
protected static Logger logger = Logger.getLogger(TraversalEngine.class);
/** what kind of traversal we're undertaking. This allows us to format output correctly */
public enum TRAVERSAL_TYPE { READ, LOCUS, LOCUS_WINDOW, DUPLICATE };
/**
* set the max number of iterations
* @param maximumIterations the number of iterations
@ -47,7 +44,7 @@ public abstract class TraversalEngine {
* @param type the TRAVERSAL_TYPE of the traversal
* @param loc the location
*/
public void printProgress(final TRAVERSAL_TYPE type, GenomeLoc loc) {
public void printProgress(final String type, GenomeLoc loc) {
printProgress(false, type, loc);
}
@ -59,7 +56,7 @@ public abstract class TraversalEngine {
* @param type String to print out describing our atomic traversal type ("read", "locus", etc)
* @param loc Current location
*/
private void printProgress(boolean mustPrint, final TRAVERSAL_TYPE type, GenomeLoc loc) {
private void printProgress(boolean mustPrint, final String type, GenomeLoc loc) {
final long nRecords = TraversalStatistics.nRecords;
final long curTime = System.currentTimeMillis();
final double elapsed = (curTime - startTime) / 1000.0;
@ -68,17 +65,10 @@ public abstract class TraversalEngine {
if (mustPrint || nRecords == 1 || nRecords % N_RECORDS_TO_PRINT == 0 || maxElapsedIntervalForPrinting(curTime)) {
this.lastProgressPrintTime = curTime;
final double secsPer1MReads = (elapsed * 1000000.0) / nRecords;
String typeString = "loci";
switch (type) {
case LOCUS: typeString = "loci"; break;
case READ: typeString = "reads"; break;
case DUPLICATE: typeString = "dups"; break;
case LOCUS_WINDOW: typeString = "interval"; break;
}
if (loc != null)
logger.info(String.format("[PROGRESS] Traversed to %s, processing %,d %s in %.2f secs (%.2f secs per 1M %s)", loc, nRecords, typeString, elapsed, secsPer1MReads, typeString));
logger.info(String.format("[PROGRESS] Traversed to %s, processing %,d %s in %.2f secs (%.2f secs per 1M %s)", loc, nRecords, type, elapsed, secsPer1MReads, type));
else
logger.info(String.format("[PROGRESS] Traversed %,d %s in %.2f secs (%.2f secs per 1M %s)", nRecords, typeString, elapsed, secsPer1MReads, typeString));
logger.info(String.format("[PROGRESS] Traversed %,d %s in %.2f secs (%.2f secs per 1M %s)", nRecords, type, elapsed, secsPer1MReads, type));
}
}
@ -93,11 +83,11 @@ public abstract class TraversalEngine {
/**
* Called after a traversal to print out information about the traversal process
*
* @param type TRAVERSAL_TYPE describing this type of traversal
* @param type describing this type of traversal
* @param sum The reduce result of the traversal
* @param <T> ReduceType of the traversal
*/
protected <T> void printOnTraversalDone(final TRAVERSAL_TYPE type, T sum) {
protected <T> void printOnTraversalDone(final String type, T sum) {
printProgress(true, type, null);
logger.info("Traversal reduce result is " + sum);
final long curTime = System.currentTimeMillis();

7
java/src/org/broadinstitute/sting/gatk/traversals/TraverseDuplicates.java
View File

@ -66,6 +66,9 @@ public class TraverseDuplicates extends TraversalEngine {
/** our log, which we want to capture anything from this class */
protected static Logger logger = Logger.getLogger(TraverseDuplicates.class);
/** descriptor of the type */
private static final String DUPS_STRING = "dups";
private final boolean DEBUG = false;
private List<SAMRecord> readsAtLoc(final SAMRecord read, PushbackIterator<SAMRecord> iter) {
@ -203,7 +206,7 @@ public class TraverseDuplicates extends TraversalEngine {
if (duplicateReads.size() > 0)
sum = mapOne(dupWalker, uniqueReads, duplicateReads, site, refBases, locus, sum);
printProgress(TRAVERSAL_TYPE.DUPLICATE, site);
printProgress(DUPS_STRING, site);
if (this.maximumIterations > 0 && TraversalStatistics.nRecords > this.maximumIterations) {
logger.warn(String.format(("Maximum number of duplicate sets encountered, terminating traversal " + TraversalStatistics.nRecords)));
@ -309,6 +312,6 @@ public class TraverseDuplicates extends TraversalEngine {
* @param <T> Type of the result.
*/
public <T> void printOnTraversalDone(T sum) {
printOnTraversalDone(TRAVERSAL_TYPE.DUPLICATE, sum);
printOnTraversalDone(DUPS_STRING, sum);
}
}

9
java/src/org/broadinstitute/sting/gatk/traversals/TraverseLoci.java
View File

@ -1,8 +1,8 @@
package org.broadinstitute.sting.gatk.traversals;
import org.apache.log4j.Logger;
import org.broadinstitute.sting.gatk.contexts.AlignmentContext;
import org.broadinstitute.sting.gatk.WalkerManager;
import org.broadinstitute.sting.gatk.contexts.AlignmentContext;
import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
import org.broadinstitute.sting.gatk.datasources.providers.*;
import org.broadinstitute.sting.gatk.datasources.shards.Shard;
@ -10,7 +10,6 @@ import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
import org.broadinstitute.sting.gatk.walkers.DataSource;
import org.broadinstitute.sting.gatk.walkers.LocusWalker;
import org.broadinstitute.sting.gatk.walkers.Walker;
import org.broadinstitute.sting.gatk.walkers.RodWalker;
import org.broadinstitute.sting.utils.GenomeLoc;
import org.broadinstitute.sting.utils.Utils;
@ -20,7 +19,7 @@ import java.util.ArrayList;
* A simple solution to iterating over all reference positions over a series of genomic locations.
*/
public class TraverseLoci extends TraversalEngine {
final private static String UNIT_STRING = "sites";
final private static String LOCI_STRING = "sites";
final private static boolean ENABLE_ROD_TRAVERSAL = false;
@ -86,7 +85,7 @@ public class TraverseLoci extends TraversalEngine {
break;
}
printProgress(TRAVERSAL_TYPE.LOCUS, locus.getLocation());
printProgress(LOCI_STRING, locus.getLocation());
}
}
@ -113,7 +112,7 @@ public class TraverseLoci extends TraversalEngine {
* @param <T> Type of the result.
*/
public <T> void printOnTraversalDone( T sum ) {
printOnTraversalDone( TRAVERSAL_TYPE.LOCUS, sum );
printOnTraversalDone(LOCI_STRING, sum );
}
/**

8
java/src/org/broadinstitute/sting/gatk/traversals/TraverseLocusWindows.java
View File

@ -15,12 +15,14 @@ import java.util.ArrayList;
/**
* Created by IntelliJ IDEA.
* User: ebanks
* User: amckenna
* Date: Apr 23, 2009
* Time: 10:26:03 AM
* To change this template use File | Settings | File Templates.
*/
public class TraverseLocusWindows extends TraversalEngine {
/** descriptor of the type */
private static final String LOCUS_WINDOW_STRING = "intervals";
public <M,T> T traverse( Walker<M,T> walker,
Shard shard,
@ -58,7 +60,7 @@ public class TraverseLocusWindows extends TraversalEngine {
sum = locusWindowWalker.reduce(x, sum);
}
printProgress(TRAVERSAL_TYPE.LOCUS_WINDOW, locus.getLocation());
printProgress(LOCUS_WINDOW_STRING, locus.getLocation());
return sum;
}
@ -98,7 +100,7 @@ public class TraverseLocusWindows extends TraversalEngine {
* @param <T> Type of the result.
*/
public <T> void printOnTraversalDone( T sum ) {
printOnTraversalDone( TRAVERSAL_TYPE.LOCUS_WINDOW, sum );
printOnTraversalDone(LOCUS_WINDOW_STRING, sum );
}
}

7
java/src/org/broadinstitute/sting/gatk/traversals/TraverseReads.java
View File

@ -56,6 +56,9 @@ public class TraverseReads extends TraversalEngine {
/** our log, which we want to capture anything from this class */
protected static Logger logger = Logger.getLogger(TraverseReads.class);
/** descriptor of the type */
private static final String READS_STRING = "reads";
/**
* Traverse by reads, given the data and the walker
*
@ -121,7 +124,7 @@ public class TraverseReads extends TraversalEngine {
sum = readWalker.reduce(x, sum);
}
printProgress(TRAVERSAL_TYPE.READ,
printProgress(READS_STRING,
(read.getReferenceIndex() == SAMRecord.NO_ALIGNMENT_REFERENCE_INDEX) ?
null :
GenomeLocParser.createGenomeLoc(read.getReferenceIndex(),read.getAlignmentStart()));
@ -136,6 +139,6 @@ public class TraverseReads extends TraversalEngine {
* @param <T> Type of the result.
*/
public <T> void printOnTraversalDone( T sum ) {
printOnTraversalDone(TRAVERSAL_TYPE.READ, sum );
printOnTraversalDone(READS_STRING, sum );
}
}

4
java/src/org/broadinstitute/sting/utils/sam/ArtificialReadsTraversal.java
View File

@ -118,7 +118,7 @@ public class ArtificialReadsTraversal extends TraversalEngine {
sum = readWalker.reduce(x, sum);
}
if (alignment != null) { printProgress(TRAVERSAL_TYPE.READ, alignment.getLocation()); }
if (alignment != null) { printProgress("reads", alignment.getLocation()); }
}
return sum;
}
@ -131,7 +131,7 @@ public class ArtificialReadsTraversal extends TraversalEngine {
* @param <T> Type of the result.
*/
public <T> void printOnTraversalDone( T sum ) {
printOnTraversalDone(TRAVERSAL_TYPE.READ, sum);
printOnTraversalDone("reads", sum);
}

7
java/test/org/broadinstitute/sting/gatk/traversals/TraverseReadsTest.java
View File

@ -1,5 +1,6 @@
package org.broadinstitute.sting.gatk.traversals;
import net.sf.picard.reference.ReferenceSequenceFile;
import org.broadinstitute.sting.BaseTest;
import org.broadinstitute.sting.gatk.Reads;
import org.broadinstitute.sting.gatk.datasources.providers.ShardDataProvider;
@ -23,8 +24,6 @@ import java.lang.reflect.Field;
import java.util.ArrayList;
import java.util.List;
import net.sf.picard.reference.ReferenceSequenceFile;
/**
*
* User: aaron
@ -139,7 +138,7 @@ public class TraverseReadsTest extends BaseTest {
}
traversalEngine.printOnTraversalDone(TraversalEngine.TRAVERSAL_TYPE.READ, accumulator);
traversalEngine.printOnTraversalDone("reads", accumulator);
countReadWalker.onTraversalDone(accumulator);
if (!(accumulator instanceof Integer)) {
@ -185,7 +184,7 @@ public class TraverseReadsTest extends BaseTest {
dataProvider.close();
}
traversalEngine.printOnTraversalDone(TraversalEngine.TRAVERSAL_TYPE.READ, accumulator);
traversalEngine.printOnTraversalDone("reads", accumulator);
countReadWalker.onTraversalDone(accumulator);
if (!(accumulator instanceof Integer)) {