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

Sped up parseGenomeLoc(..) by replacing regexp with String.indexOf(..) 

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@3522 348d0f76-0448-11de-a6fe-93d51630548a
This commit is contained in:
weisburd 2010-06-09 18:11:43 +00:00
parent 20167fd411
commit adc4c4e577
1 changed files with 63 additions and 48 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

111
java/src/org/broadinstitute/sting/utils/GenomeLocParser.java
View File

@ -25,25 +25,24 @@
package org.broadinstitute.sting.utils; package org.broadinstitute.sting.utils;
import net.sf.picard.util.IntervalList;
import net.sf.picard.util.Interval;
import net.sf.picard.reference.ReferenceSequenceFile;
import net.sf.samtools.SAMRecord;
import net.sf.samtools.SAMSequenceDictionary;
import net.sf.samtools.SAMSequenceRecord;
import org.apache.log4j.Logger;
import org.broadinstitute.sting.gatk.GenomeAnalysisEngine;
import org.broadinstitute.sting.utils.interval.IntervalMergingRule;
import org.broadinstitute.sting.gatk.arguments.ValidationExclusion;
import org.broadinstitute.sting.utils.bed.BedParser;
import org.broadinstitute.sting.utils.text.XReadLines;
import java.io.File; import java.io.File;
import java.util.ArrayList; import java.util.ArrayList;
import java.util.Iterator; import java.util.Iterator;
import java.util.List; import java.util.List;
import java.util.regex.Matcher;
import java.util.regex.Pattern; import net.sf.picard.reference.ReferenceSequenceFile;
import net.sf.picard.util.Interval;
import net.sf.picard.util.IntervalList;
import net.sf.samtools.SAMRecord;
import net.sf.samtools.SAMSequenceDictionary;
import net.sf.samtools.SAMSequenceRecord;
import org.apache.log4j.Logger;
import org.broadinstitute.sting.gatk.GenomeAnalysisEngine;
import org.broadinstitute.sting.gatk.arguments.ValidationExclusion;
import org.broadinstitute.sting.utils.bed.BedParser;
import org.broadinstitute.sting.utils.interval.IntervalMergingRule;
import org.broadinstitute.sting.utils.text.XReadLines;
/** /**
* Created by IntelliJ IDEA. * Created by IntelliJ IDEA.
@ -55,7 +54,7 @@ import java.util.regex.Pattern;
public class GenomeLocParser { public class GenomeLocParser {
private static Logger logger = Logger.getLogger(GenomeLocParser.class); private static Logger logger = Logger.getLogger(GenomeLocParser.class);
private static final Pattern mPattern = Pattern.compile("([\\p{Print}&&[^:]]+):*([\\d,]+)?([\\+-])?([\\d,]+)?$"); // matches case 3 //private static final Pattern mPattern = Pattern.compile("([\\p{Print}&&[^:]]+):*([\\d,]+)?([\\+-])?([\\d,]+)?$"); // matches case 3
// -------------------------------------------------------------------------------------------------------------- // --------------------------------------------------------------------------------------------------------------
@ -137,7 +136,7 @@ public class GenomeLocParser {
/** /**
* parse a genome interval, from a location string * parse a genome interval, from a location string
* *
* Performs interval-style validation: * Performs interval-style validation:
* *
* contig is valid; start and stop less than the end; start <= sto * contig is valid; start and stop less than the end; start <= sto
* @param str the string to parse * @param str the string to parse
@ -154,8 +153,8 @@ public class GenomeLocParser {
/** /**
* parse a genome location, from a location string * parse a genome location, from a location string
* *
* Performs read-style validation: * Performs read-style validation:
* checks that start and stop are positive, start < stop, and the contig is valid * checks that start and stop are positive, start < stop, and the contig is valid
* does not check that genomeLoc is actually on the contig * does not check that genomeLoc is actually on the contig
* *
@ -167,40 +166,41 @@ public class GenomeLocParser {
public static GenomeLoc parseGenomeLoc(final String str) { public static GenomeLoc parseGenomeLoc(final String str) {
// 'chr2', 'chr2:1000000' or 'chr2:1,000,000-2,000,000' // 'chr2', 'chr2:1000000' or 'chr2:1,000,000-2,000,000'
//System.out.printf("Parsing location '%s'%n", str); //System.out.printf("Parsing location '%s'%n", str);
String contig = null; String contig = null;
long start = 1; long start = 1;
long stop = -1; long stop = -1;
boolean bad = false;
Matcher match = mPattern.matcher(str); final int colonIndex = str.indexOf(":");
try { if(colonIndex == -1) {
if (match.matches() && match.groupCount() == 4) { contig = str.substring(0, str.length()); // chr1
if (match.group(1) != null) contig = match.group(1); stop = Integer.MAX_VALUE;
if (match.group(2) != null) start = parsePosition(match.group(2)); } else {
if ((match.group(3) != null && match.group(3).equals("+")) || // chr:1+ contig = str.substring(0, colonIndex);
(match.group(3) == null && match.group(4) == null && match.group(2) == null)) // chr1 final int dashIndex = str.indexOf('-', colonIndex);
stop = Integer.MAX_VALUE; try {
else if (match.group(3) != null && match.group(3).equals("-")) // chr1:1-1 if(dashIndex == -1) {
stop = parsePosition(match.group(4)); if(str.charAt(str.length() - 1) == '+') {
else if (match.group(3) == null && match.group(4) == null) // chr1:1 start = parsePosition(str.substring(colonIndex + 1, str.length() - 1)); // chr:1+
stop = start; stop = Integer.MAX_VALUE;
else { } else {
bad = true; start = parsePosition(str.substring(colonIndex + 1)); // chr1:1
stop = start;
}
} else {
start = parsePosition(str.substring(colonIndex + 1, dashIndex)); // chr1:1-1
stop = parsePosition(str.substring(dashIndex + 1));
} }
} catch(Exception e) {
throw new StingException("Failed to parse Genome Location string: " + str, e);
} }
} catch (Exception e) {
bad = true;
} }
if (bad)
throw new StingException("Failed to parse Genome Location string: " + str);
// is the contig valid? // is the contig valid?
if (!isContigValid(contig)) if (!isContigValid(contig))
throw new StingException("Contig " + contig + " does not match any contig in the GATK sequence dictionary derived from the reference; are you sure you are using the correct reference fasta file?"); throw new StingException("Contig '" + contig + "' does not match any contig in the GATK sequence dictionary derived from the reference; are you sure you are using the correct reference fasta file?");
if (stop == Integer.MAX_VALUE && hasKnownContigOrdering()) if (stop == Integer.MAX_VALUE && hasKnownContigOrdering())
// lookup the actually stop position! // lookup the actually stop position!
stop = getContigInfo(contig).getSequenceLength(); stop = getContigInfo(contig).getSequenceLength();
@ -214,9 +214,24 @@ public class GenomeLocParser {
// Parsing string representations // Parsing string representations
// //
// -------------------------------------------------------------------------------------------------------------- // --------------------------------------------------------------------------------------------------------------
/**
* Parses a number like 1,000,000 into a long.
* @param pos
*/
private static long parsePosition(final String pos) { private static long parsePosition(final String pos) {
String x = pos.replaceAll(",", ""); //String x = pos.replaceAll(",", ""); - this was replaced because it uses regexps
return Long.parseLong(x); if(pos.indexOf(',') != -1) {
final StringBuilder buffer = new StringBuilder();
for(int i = 0; i < pos.length(); i++) {
final char c = pos.charAt(i);
if(c != ',') {
buffer.append(c);
}
}
return Long.parseLong(buffer.toString());
} else {
return Long.parseLong(pos);
}
} }
@ -457,11 +472,11 @@ public class GenomeLocParser {
/** /**
* verify the specified genome loc is valid, if it's not, throw an exception * verify the specified genome loc is valid, if it's not, throw an exception
* Will not verify the location against contig bounds. * Will not verify the location against contig bounds.
*
* *
* Validation: *
* checks that start and stop are positive, start < stop, and the contig is valid * Validation:
* does not check that genomeLoc is actually on the contig, so start could be > end of contig * checks that start and stop are positive, start < stop, and the contig is valid
* does not check that genomeLoc is actually on the contig, so start could be > end of contig
* *
* @param toReturn the genome loc we're about to return * @param toReturn the genome loc we're about to return
* *