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

Merge branch 'master' of ssh://nickel.broadinstitute.org/humgen/gsa-scr1/gsa-engineering/git/unstable

This commit is contained in:
Ryan Poplin 2012-03-05 08:29:08 -05:00
parent d6871967ae 3b5a7c34d7
commit f879daa7d0
5 changed files with 67 additions and 47 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
public/R/src/org/broadinstitute/sting/utils/R/gsalib/R/gsa.read.gatkreport.R
View File

@ -4,11 +4,9 @@
colnames(d) = tableHeader; colnames(d) = tableHeader;
for (i in 1:ncol(d)) { for (i in 1:ncol(d)) {
v = suppressWarnings(as.numeric(d[,i])); # use the general type.convert infrastructure of read.table to convert column data to R types
v = type.convert(d[,i])
if (length(na.omit(as.numeric(v))) == length(d[,i])) { d[,i] = v;
d[,i] = v;
}
} }
usedNames = ls(envir=tableEnv, pattern=tableName); usedNames = ls(envir=tableEnv, pattern=tableName);

8
public/java/src/org/broadinstitute/sting/gatk/walkers/annotator/RankSumTest.java
View File

@ -43,15 +43,15 @@ public abstract class RankSumTest extends InfoFieldAnnotation implements Standar
final ArrayList<Double> altQuals = new ArrayList<Double>(); final ArrayList<Double> altQuals = new ArrayList<Double>();
if ( vc.isSNP() ) { if ( vc.isSNP() ) {
final List<Byte> altAlleles = new ArrayList<Byte>();
for ( final Allele a : vc.getAlternateAlleles() )
altAlleles.add(a.getBases()[0]);
for ( final Genotype genotype : genotypes.iterateInSampleNameOrder() ) { for ( final Genotype genotype : genotypes.iterateInSampleNameOrder() ) {
final AlignmentContext context = stratifiedContexts.get(genotype.getSampleName()); final AlignmentContext context = stratifiedContexts.get(genotype.getSampleName());
if ( context == null ) if ( context == null )
continue; continue;
final List<Byte> altAlleles = new ArrayList<Byte>();
for ( final Allele a : vc.getAlternateAlleles() )
altAlleles.add(a.getBases()[0]);
fillQualsFromPileup(ref.getBase(), altAlleles, context.getBasePileup(), refQuals, altQuals); fillQualsFromPileup(ref.getBase(), altAlleles, context.getBasePileup(), refQuals, altQuals);
} }
} else if ( vc.isIndel() || vc.isMixed() ) { } else if ( vc.isIndel() || vc.isMixed() ) {

73
public/java/src/org/broadinstitute/sting/gatk/walkers/diagnostics/ReadGroupProperties.java
View File

@ -36,13 +36,14 @@ import org.broadinstitute.sting.utils.Median;
import org.broadinstitute.sting.utils.sam.GATKSAMRecord; import org.broadinstitute.sting.utils.sam.GATKSAMRecord;
import java.io.PrintStream; import java.io.PrintStream;
import java.text.DateFormat;
import java.util.HashMap; import java.util.HashMap;
import java.util.Map; import java.util.Map;
/** /**
* Emits a GATKReport containing read group, sample, library, platform, center, paired end status, * Emits a GATKReport containing read group, sample, library, platform, center, sequencing data,
* simple read type name (e.g. 2x76) median insert size and median read length for each read group * paired end status, simple read type name (e.g. 2x76) median insert size and median read length
* in every provided BAM file * for each read group in every provided BAM file
* *
* Note that this walker stops when all read groups have been observed at least a few thousand times so that * Note that this walker stops when all read groups have been observed at least a few thousand times so that
* the median statistics are well determined. It is safe to run it WG and it'll finish in an appropriate * the median statistics are well determined. It is safe to run it WG and it'll finish in an appropriate
@ -61,23 +62,23 @@ import java.util.Map;
* *
* <pre> * <pre>
* ##:GATKReport.v0.2 ReadGroupProperties : Table of read group properties * ##:GATKReport.v0.2 ReadGroupProperties : Table of read group properties
* readgroup sample library platform center has.any.reads is.paired.end n.reads.analyzed simple.read.type median.read.length median.insert.size * readgroup sample library platform center date has.any.reads is.paired.end n.reads.analyzed simple.read.type median.read.length median.insert.size
* 20FUK.1 NA12878 Solexa-18483 illumina BI true true 10100 2x101 101 387 * 20FUK.1 NA12878 Solexa-18483 illumina BI 2/2/10 true true 498 2x101 101 386
* 20FUK.2 NA12878 Solexa-18484 illumina BI true true 10115 2x101 101 415 * 20FUK.2 NA12878 Solexa-18484 illumina BI 2/2/10 true true 476 2x101 101 417
* 20FUK.3 NA12878 Solexa-18483 illumina BI true true 10090 2x101 101 388 * 20FUK.3 NA12878 Solexa-18483 illumina BI 2/2/10 true true 407 2x101 101 387
* 20FUK.4 NA12878 Solexa-18484 illumina BI true true 10081 2x101 101 415 * 20FUK.4 NA12878 Solexa-18484 illumina BI 2/2/10 true true 389 2x101 101 415
* 20FUK.5 NA12878 Solexa-18483 illumina BI true true 10078 2x101 101 387 * 20FUK.5 NA12878 Solexa-18483 illumina BI 2/2/10 true true 433 2x101 101 386
* 20FUK.6 NA12878 Solexa-18484 illumina BI true true 10072 2x101 101 415 * 20FUK.6 NA12878 Solexa-18484 illumina BI 2/2/10 true true 480 2x101 101 418
* 20FUK.7 NA12878 Solexa-18483 illumina BI true true 10086 2x101 101 388 * 20FUK.7 NA12878 Solexa-18483 illumina BI 2/2/10 true true 450 2x101 101 386
* 20FUK.8 NA12878 Solexa-18484 illumina BI true true 10097 2x101 101 415 * 20FUK.8 NA12878 Solexa-18484 illumina BI 2/2/10 true true 438 2x101 101 418
* 20GAV.1 NA12878 Solexa-18483 illumina BI true true 10135 2x101 101 388 * 20GAV.1 NA12878 Solexa-18483 illumina BI 1/26/10 true true 490 2x101 101 391
* 20GAV.2 NA12878 Solexa-18484 illumina BI true true 10172 2x101 101 415 * 20GAV.2 NA12878 Solexa-18484 illumina BI 1/26/10 true true 485 2x101 101 417
* 20GAV.3 NA12878 Solexa-18483 illumina BI true true 10141 2x101 101 388 * 20GAV.3 NA12878 Solexa-18483 illumina BI 1/26/10 true true 460 2x101 101 392
* 20GAV.4 NA12878 Solexa-18484 illumina BI true true 10251 2x101 101 416 * 20GAV.4 NA12878 Solexa-18484 illumina BI 1/26/10 true true 434 2x101 101 415
* 20GAV.5 NA12878 Solexa-18483 illumina BI true true 10145 2x101 101 388 * 20GAV.5 NA12878 Solexa-18483 illumina BI 1/26/10 true true 479 2x101 101 389
* 20GAV.6 NA12878 Solexa-18484 illumina BI true true 10184 2x101 101 415 * 20GAV.6 NA12878 Solexa-18484 illumina BI 1/26/10 true true 461 2x101 101 416
* 20GAV.7 NA12878 Solexa-18483 illumina BI true true 10174 2x101 101 387 * 20GAV.7 NA12878 Solexa-18483 illumina BI 1/26/10 true true 509 2x101 101 386
* 20GAV.8 NA12878 Solexa-18484 illumina BI true true 10141 2x101 101 414 * 20GAV.8 NA12878 Solexa-18484 illumina BI 1/26/10 true true 476 2x101 101 410 101 414
* </pre> * </pre>
* </p> * </p>
* *
@ -172,6 +173,7 @@ public class ReadGroupProperties extends ReadWalker<Integer, Integer> {
final GATKReport report = new GATKReport(); final GATKReport report = new GATKReport();
report.addTable(TABLE_NAME, "Table of read group properties"); report.addTable(TABLE_NAME, "Table of read group properties");
GATKReportTable table = report.getTable(TABLE_NAME); GATKReportTable table = report.getTable(TABLE_NAME);
DateFormat dateFormatter = DateFormat.getDateInstance(DateFormat.SHORT);
table.addPrimaryKey("readgroup"); table.addPrimaryKey("readgroup");
//* Emits a GATKReport containing read group, sample, library, platform, center, median insert size and //* Emits a GATKReport containing read group, sample, library, platform, center, median insert size and
@ -180,6 +182,7 @@ public class ReadGroupProperties extends ReadWalker<Integer, Integer> {
table.addColumn("library", "NA"); table.addColumn("library", "NA");
table.addColumn("platform", "NA"); table.addColumn("platform", "NA");
table.addColumn("center", "NA"); table.addColumn("center", "NA");
table.addColumn("date", "NA");
table.addColumn("has.any.reads", "false"); table.addColumn("has.any.reads", "false");
table.addColumn("is.paired.end", "false"); table.addColumn("is.paired.end", "false");
table.addColumn("n.reads.analyzed", "NA"); table.addColumn("n.reads.analyzed", "NA");
@ -196,18 +199,28 @@ public class ReadGroupProperties extends ReadWalker<Integer, Integer> {
final boolean hasAnyReads = info.nReadsSeen > 0; final boolean hasAnyReads = info.nReadsSeen > 0;
final int readLength = info.readLength.getMedian(0); final int readLength = info.readLength.getMedian(0);
table.set(rgID, "sample", rg.getSample()); setTableValue(table, rgID, "sample", rg.getSample());
table.set(rgID, "library", rg.getLibrary()); setTableValue(table, rgID, "library", rg.getLibrary());
table.set(rgID, "platform", rg.getPlatform()); setTableValue(table, rgID, "platform", rg.getPlatform());
table.set(rgID, "center", rg.getSequencingCenter()); setTableValue(table, rgID, "center", rg.getSequencingCenter());
table.set(rgID, "has.any.reads", hasAnyReads); try {
table.set(rgID, "is.paired.end", isPaired); setTableValue(table, rgID, "date", rg.getRunDate() != null ? dateFormatter.format(rg.getRunDate()) : "NA");
table.set(rgID, "n.reads.analyzed", info.nReadsSeen); } catch ( NullPointerException e ) {
table.set(rgID, "simple.read.type", hasAnyReads ? String.format("%dx%d", isPaired ? 2 : 1, readLength) : "NA"); // TODO: remove me when bug in Picard is fixed that causes NPE when date isn't present
table.set(rgID, "median.read.length", hasAnyReads ? readLength : "NA" ); setTableValue(table, rgID, "date", "NA");
table.set(rgID, "median.insert.size", hasAnyReads && isPaired ? info.insertSize.getMedian(0) : "NA" ); }
setTableValue(table, rgID, "has.any.reads", hasAnyReads);
setTableValue(table, rgID, "is.paired.end", isPaired);
setTableValue(table, rgID, "n.reads.analyzed", info.nReadsSeen);
setTableValue(table, rgID, "simple.read.type", hasAnyReads ? String.format("%dx%d", isPaired ? 2 : 1, readLength) : "NA");
setTableValue(table, rgID, "median.read.length", hasAnyReads ? readLength : "NA" );
setTableValue(table, rgID, "median.insert.size", hasAnyReads && isPaired ? info.insertSize.getMedian(0) : "NA" );
} }
report.print(out); report.print(out);
} }
private final void setTableValue(GATKReportTable table, final String rgID, final String key, final Object value) {
table.set(rgID, key, value == null ? "NA" : value);
}
} }

23
public/java/src/org/broadinstitute/sting/gatk/walkers/validation/ValidationAmplicons.java
View File

@ -110,6 +110,13 @@ public class ValidationAmplicons extends RodWalker<Integer,Integer> {
@Argument(doc="Lower case SNPs rather than replacing with 'N'",fullName="lowerCaseSNPs",required=false) @Argument(doc="Lower case SNPs rather than replacing with 'N'",fullName="lowerCaseSNPs",required=false)
boolean lowerCaseSNPs = false; boolean lowerCaseSNPs = false;
/**
* If onlyOutputValidAmplicons is true, the output fasta file will contain only valid sequences.
* Useful for producing delivery-ready files.
*/
@Argument(doc="Only output valid sequences.",fullName="onlyOutputValidAmplicons",required=false)
boolean onlyOutputValidAmplicons = false;
/** /**
* BWA single-end alignment is used as a primer specificity proxy. Low-complexity regions (that don't align back to themselves as a best hit) are lowercased. * BWA single-end alignment is used as a primer specificity proxy. Low-complexity regions (that don't align back to themselves as a best hit) are lowercased.
* This changes the size of the k-mer used for alignment. * This changes the size of the k-mer used for alignment.
@ -486,14 +493,16 @@ public class ValidationAmplicons extends RodWalker<Integer,Integer> {
valid = "Valid"; valid = "Valid";
} }
String seqIdentity = sequence.toString().replace('n', 'N').replace('i', 'I').replace('d', 'D');
if (!sequenomOutput) if (!onlyOutputValidAmplicons || !sequenceInvalid) {
out.printf(">%s %s %s%n%s%n", allelePos != null ? allelePos.toString() : "multiple", valid, probeName, seqIdentity); String seqIdentity = sequence.toString().replace('n', 'N').replace('i', 'I').replace('d', 'D');
else { if (!sequenomOutput)
seqIdentity = seqIdentity.replace("*",""); // identifier < 20 letters long, no * in ref allele, one line per record out.printf(">%s %s %s%n%s%n", allelePos != null ? allelePos.toString() : "multiple", valid, probeName, seqIdentity);
probeName = probeName.replace("amplicon_","a"); else {
out.printf("%s_%s %s%n", allelePos != null ? allelePos.toString() : "multiple", probeName, seqIdentity); seqIdentity = seqIdentity.replace("*",""); // identifier < 20 letters long, no * in ref allele, one line per record
probeName = probeName.replace("amplicon_","a");
out.printf("%s_%s %s%n", allelePos != null ? allelePos.toString() : "multiple", probeName, seqIdentity);
}
} }
} }
} }

2
public/java/test/org/broadinstitute/sting/gatk/walkers/diagnostics/ReadGroupPropertiesIntegrationTest.java
View File

@ -38,7 +38,7 @@ public class ReadGroupPropertiesIntegrationTest extends WalkerTest {
WalkerTestSpec spec = new WalkerTestSpec( WalkerTestSpec spec = new WalkerTestSpec(
"-T ReadGroupProperties -R " + b37KGReference + " -I " + b37GoodBAM + " -L 20:10,000,000-11,000,000 -o %s", "-T ReadGroupProperties -R " + b37KGReference + " -I " + b37GoodBAM + " -L 20:10,000,000-11,000,000 -o %s",
1, 1,
Arrays.asList("8e4d09665c0b65c971bd5dead24f95fe")); Arrays.asList("6b8cce223af28cbadcfe87a3b841fc56"));
executeTest("ReadGroupProperties:", spec); executeTest("ReadGroupProperties:", spec);
} }
} }