From 6e46b3696eb7e64bb15cc81620a414d19d2d5a95 Mon Sep 17 00:00:00 2001 From: Ron Levine Date: Mon, 6 Jul 2015 12:30:53 -0400 Subject: [PATCH] Merge contiguous intervals properly --- ...astaAlternateReferenceIntegrationTest.java | 53 +++++++++++++++++-- .../walkers/fasta/FastaReferenceMaker.java | 18 +++++-- 2 files changed, 63 insertions(+), 8 deletions(-) diff --git a/protected/gatk-tools-protected/src/test/java/org/broadinstitute/gatk/tools/walkers/fasta/FastaAlternateReferenceIntegrationTest.java b/protected/gatk-tools-protected/src/test/java/org/broadinstitute/gatk/tools/walkers/fasta/FastaAlternateReferenceIntegrationTest.java index 6d1c4b0e3..765deb0e5 100644 --- a/protected/gatk-tools-protected/src/test/java/org/broadinstitute/gatk/tools/walkers/fasta/FastaAlternateReferenceIntegrationTest.java +++ b/protected/gatk-tools-protected/src/test/java/org/broadinstitute/gatk/tools/walkers/fasta/FastaAlternateReferenceIntegrationTest.java @@ -59,23 +59,68 @@ import java.util.Arrays; public class FastaAlternateReferenceIntegrationTest extends WalkerTest { + private static String CONTIGUOUS_INTERVAL_SAME_CONTIG_MD5 = "e1f4b93f9071d158d94dc4fb25e07702"; + private static String CONTIGUOUS_INTERVAL_DIFF_CONTIG_MD5 = "dfca4e0b0fe0cb18596ec51af541a69e"; + @Test public void testReferenceOnly() { WalkerTestSpec spec = new WalkerTestSpec( "-T FastaReferenceMaker -R " + b36KGReference + " -L 1:10,000,100-10,000,500 -L 1:10,100,000-10,101,000 -L 1:10,900,000-10,900,001 -o %s", 1, - Arrays.asList("328d2d52cedfdc52da7d1abff487633d")); + Arrays.asList("75d4d352a9ce4fae22fd7924a42c800a")); executeTest("test FastaReference", spec); } + @Test + public void testReferenceOnlyContiguousSameContig() { + + WalkerTestSpec spec = new WalkerTestSpec( + "-T FastaReferenceMaker -R " + b36KGReference + " -L 1:10,000,100-10,000,200 -L 1:10,000,201-10,000,301 -o %s", + 1, + Arrays.asList(CONTIGUOUS_INTERVAL_SAME_CONTIG_MD5)); + executeTest("test FastaReference with contiguous intervals, same contig", spec); + } + + @Test + public void testReferenceOnlyContiguousDiffContigs() { + + WalkerTestSpec spec = new WalkerTestSpec( + "-T FastaReferenceMaker -R " + b36KGReference + " -L 1:10,000,100-10,000,200 -L 2:10,000,201-10,000,301 -o %s", + 1, + Arrays.asList(CONTIGUOUS_INTERVAL_DIFF_CONTIG_MD5)); + executeTest("test FastaReference with contiguous intervals, different contigs", spec); + } + + @Test + public void testAlternateReferenceContiguousSameContig() { + // Show that FastaAlternateReferenceMaker behaves the same as FastaReferenceMaker across contiguous intervals on the same contig. + // Note that there are variant locations in this interval. + WalkerTestSpec spec = new WalkerTestSpec( + "-T FastaAlternateReferenceMaker -R " + b36KGReference + " -V " + validationDataLocation + "NA12878.chr1_10mb_11mb.slx.indels.vcf4 -L 1:10,000,100-10,000,200 -L 1:10,000,201-10,000,301 -o %s", + 1, + Arrays.asList(CONTIGUOUS_INTERVAL_SAME_CONTIG_MD5)); + executeTest("test Alternate FastaReference with contiguous intervals, same contig", spec); + } + + @Test + public void testAlternateReferenceContiguousDiffContigs() { + // Show that FastaAlternateReferenceMaker behaves the same as FastaReferenceMaker across contiguous intervals on different contigs. + // Note that there are variant locations in this interval. + WalkerTestSpec spec = new WalkerTestSpec( + "-T FastaAlternateReferenceMaker -R " + b36KGReference + " -V " + validationDataLocation + "NA12878.chr1_10mb_11mb.slx.indels.vcf4 -L 1:10,000,100-10,000,200 -L 2:10,000,201-10,000,301 -o %s", + 1, + Arrays.asList(CONTIGUOUS_INTERVAL_DIFF_CONTIG_MD5)); + executeTest("test Alternate FastaReference with contiguous intervals, different contigs", spec); + } + @Test public void testIndelsAndSnpMask() { WalkerTestSpec spec = new WalkerTestSpec( "-T FastaAlternateReferenceMaker -R " + b36KGReference + " -V " + validationDataLocation + "NA12878.chr1_10mb_11mb.slx.indels.vcf4 --snpmask:vcf " + b36dbSNP129 + " -L 1:10,075,000-10,075,380 -L 1:10,093,447-10,093,847 -L 1:10,271,252-10,271,452 -o %s", 1, - Arrays.asList("ef481be9962e21d09847b8a1d4a4ff65")); + Arrays.asList("375efb2feb017f01339f680fdffac6cd")); executeTest("test indels", spec); } @@ -85,7 +130,7 @@ public class FastaAlternateReferenceIntegrationTest extends WalkerTest { WalkerTestSpec spec = new WalkerTestSpec( "-T FastaAlternateReferenceMaker -R " + b36KGReference + " -V " + GATKDataLocation + "dbsnp_129_b36.vcf -L 1:10,023,400-10,023,500 -L 1:10,029,200-10,029,500 -o %s", 1, - Arrays.asList("8b6cd2e20c381f9819aab2d270f5e641")); + Arrays.asList("81e30f0ab92684c496343c8ea51a393e")); executeTest("test SNPs", spec); } @@ -108,7 +153,7 @@ public class FastaAlternateReferenceIntegrationTest extends WalkerTest { WalkerTestSpec spec = new WalkerTestSpec( "-T FastaAlternateReferenceMaker -R " + b37KGReference + " --use_IUPAC_sample NA12878 -V " + privateTestDir + "NA12878.WGS.b37.chr20.firstMB.vcf -L 20:61050-66380 -o %s", 1, - Arrays.asList("5feb2a576ff2ed1745a007eaa36448b3")); + Arrays.asList("8fd887bca9f3949f2c23c3565f7dcc1b")); executeTest("test iupac", spec); } } diff --git a/public/gatk-tools-public/src/main/java/org/broadinstitute/gatk/tools/walkers/fasta/FastaReferenceMaker.java b/public/gatk-tools-public/src/main/java/org/broadinstitute/gatk/tools/walkers/fasta/FastaReferenceMaker.java index 08ab3019a..9df4579e2 100644 --- a/public/gatk-tools-public/src/main/java/org/broadinstitute/gatk/tools/walkers/fasta/FastaReferenceMaker.java +++ b/public/gatk-tools-public/src/main/java/org/broadinstitute/gatk/tools/walkers/fasta/FastaReferenceMaker.java @@ -55,7 +55,15 @@ import java.io.PrintStream; * *

Output

*

- * A fasta file representing the requested intervals. + * A fasta file representing the requested intervals. Each interval has a description line starting with a greater-than (">") symbol followed by sequence data. + * The description begins with the contig name followed by the beginning position on the contig. + * 

+ * For example, the fasta file for contig 1 and intervals 1:3-1:4 and 1:6-1:9
+ * >1 1:3
+ * AT
+ * >1 1:6
+ * GGGG
+ *
*

* *

Usage example

@@ -104,18 +112,20 @@ public class FastaReferenceMaker extends RefWalker, Geno // if there is no interval to the left, then this is the first one if ( sum == null ) { sum = value.first; + fasta.setName(fasta.getName() + " " + sum.toString()); fasta.append(value.second); } - // if the intervals don't overlap, print out the leftmost one and start a new one + // if the intervals are not contiguous, print out the leftmost one and start a new one // (end of contig or new interval) - else if ( value.first.getStart() != sum.getStop() + 1 ) { + else if ( value.first.getStart() != sum.getStop() + 1 || ! value.first.getContig().equals(sum.getContig()) ) { fasta.flush(); sum = value.first; + fasta.setName(fasta.getName() + " " + sum.toString()); fasta.append(value.second); } // otherwise, merge them else { - sum = getToolkit().getGenomeLocParser().setStop(sum, value.first.getStop()); + sum = sum.setStop(sum, value.first.getStop()); fasta.append(value.second); } return sum;