changed intervals and also removed ememoty dbsnp from test

nf-core · Nov 26, 2024 · c1aa757 · c1aa757
1 parent 13ad117
commit c1aa757
Show file tree

Hide file tree

Showing 50 changed files with 54,254 additions and 5 deletions.
diff --git a/conf/test.config b/conf/test.config
@@ -31,6 +31,7 @@ params {
     // Additional flags - interval can be null
     barcodes                    = params.pipelines_testdata_base_path + 'refs/heads/pacvar/pacbio_data/barcodes_D01.fasta'
     intervals                   = params.pipelines_testdata_base_path + 'refs/heads/pacvar/pacbio_data/c9orf72-short.bed'
+
     workflow                    = 'repeat'
     repeat_id                          = 'C9ORF72'
 

diff --git a/conf/test_wgs_deepvariant.config b/conf/test_wgs_deepvariant.config
@@ -28,8 +28,6 @@ params {
     fasta_fai                   = params.pipelines_testdata_base_path + 'refs/heads/pacvar/pacbio_data/C9ORF72-12.fa.fai'
     genome                      = 'GATK.GRCh38'
     dict                        = params.pipelines_testdata_base_path + 'refs/heads/pacvar/pacbio_data/C9ORF72-12.dict'
-    dbsnp                       = ''
-    dbsnp_tbi                   = ''
     barcodes                    = params.pipelines_testdata_base_path + 'refs/heads/pacvar/pacbio_data/barcodes_D01.fasta'
     snv_caller                  = 'deepvariant'
     intervals                   = params.pipelines_testdata_base_path + 'refs/heads/pacvar/pacbio_data/c9orf72-short.bed'

diff --git a/conf/test_wgs_gatk.config b/conf/test_wgs_gatk.config
@@ -28,8 +28,6 @@ params {
     fasta_fai                   = params.pipelines_testdata_base_path + 'refs/heads/pacvar/pacbio_data/C9ORF72-12.fa.fai'
     genome                      = 'GATK.GRCh38'
     dict                        = params.pipelines_testdata_base_path + 'refs/heads/pacvar/pacbio_data/C9ORF72-12.dict'
-    dbsnp                       = ''
-    dbsnp_tbi                   = ''
     barcodes                    = params.pipelines_testdata_base_path + 'refs/heads/pacvar/pacbio_data/barcodes_D01.fasta'
     snv_caller                  = 'gatk4'
     intervals                   = params.pipelines_testdata_base_path + 'refs/heads/pacvar/pacbio_data/c9orf72-short.bed'

diff --git a/nextflow.config b/nextflow.config
@@ -27,7 +27,7 @@ params {
 
     workflow                   = 'wgs'
     snv_caller                 = 'deepvariant'
-    intervals                  = []
+    intervals                  =  null
 
     // References
     genome                     = null

diff --git a/out/lima/sample1.bc2026--bc2026.bam b/out/lima/sample1.bc2026--bc2026.bam
diff --git a/out/lima/sample1.bc2026--bc2026.bam.pbi b/out/lima/sample1.bc2026--bc2026.bam.pbi
diff --git a/out/lima/sample1.bc2026--bc2026.consensusreadset.xml b/out/lima/sample1.bc2026--bc2026.consensusreadset.xml
@@ -0,0 +1,14 @@
+<?xml version="1.0" encoding="utf-8"?>
+<pbds:ConsensusReadSet CreatedAt="2024-11-26T23:31:14.33Z" MetaType="PacBio.DataSet.ConsensusReadSet" Name="sample1.bc2026--bc2026 (bc2026--bc2026)" Tags="barcoded" TimeStampedName="sample1.bc2026--bc2026 (bc2026--bc2026)-241126_23311435" UniqueId="d37de0ef-4929-4029-887e-91d4812bcff6" Version="3.0.1" xmlns:pbbase="http://pacificbiosciences.com/PacBioBaseDataModel.xsd" xmlns:pbdm="http://pacificbiosciences.com/PacBioDataModel.xsd" xmlns:pbmeta="http://pacificbiosciences.com/PacBioCollectionMetadata.xsd" xmlns:pbpn="http://pacificbiosciences.com/PacBioPartNumbers.xsd" xmlns:pbrk="http://pacificbiosciences.com/PacBioReagentKit.xsd" xmlns:pbsample="http://pacificbiosciences.com/PacBioSampleInfo.xsd" xmlns="http://pacificbiosciences.com/PacBioDatasets.xsd" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://pacificbiosciences.com/PacBioDatasets.xsd" xmlns:pbds="http://pacificbiosciences.com/PacBioDatasets.xsd">
+	<pbds:DataSetMetadata>
+		<pbds:TotalLength>31966671</pbds:TotalLength>
+		<pbds:NumRecords>5725</pbds:NumRecords>
+	</pbds:DataSetMetadata>
+	<pbbase:ExternalResources>
+		<pbbase:ExternalResource CreatedAt="2024-11-26T23:31:14.34Z" MetaType="PacBio.ConsensusReadFile.ConsensusReadBamFile" ResourceId="sample1.bc2026--bc2026.bam" TimeStampedName="pacbio_consensusreadfile_consensusreadbamfile-241126_23311434" UniqueId="745700d3-2154-478b-a446-ea7aea1990f5" Version="3.0.1">
+			<pbbase:FileIndices>
+				<pbbase:FileIndex CreatedAt="2024-11-26T23:31:14.34Z" MetaType="PacBio.Index.PacBioIndex" ResourceId="sample1.bc2026--bc2026.bam.pbi" TimeStampedName="pacbio_index_pacbioindex-241126_23311434" UniqueId="7c94d293-de4a-4609-aa0c-d7f1cb70a971" Version="3.0.1" />
+			</pbbase:FileIndices>
+		</pbbase:ExternalResource>
+	</pbbase:ExternalResources>
+</pbds:ConsensusReadSet>
diff --git a/out/lima/sample1.bc2058--bc2058.bam b/out/lima/sample1.bc2058--bc2058.bam
diff --git a/out/lima/sample1.bc2058--bc2058.bam.pbi b/out/lima/sample1.bc2058--bc2058.bam.pbi
diff --git a/out/lima/sample1.bc2058--bc2058.consensusreadset.xml b/out/lima/sample1.bc2058--bc2058.consensusreadset.xml
@@ -0,0 +1,14 @@
+<?xml version="1.0" encoding="utf-8"?>
+<pbds:ConsensusReadSet CreatedAt="2024-11-26T23:31:14.58Z" MetaType="PacBio.DataSet.ConsensusReadSet" Name="sample1.bc2058--bc2058 (bc2058--bc2058)" Tags="barcoded" TimeStampedName="sample1.bc2058--bc2058 (bc2058--bc2058)-241126_23311458" UniqueId="6c38803d-0e02-4593-b036-913aa8e05052" Version="3.0.1" xmlns:pbbase="http://pacificbiosciences.com/PacBioBaseDataModel.xsd" xmlns:pbdm="http://pacificbiosciences.com/PacBioDataModel.xsd" xmlns:pbmeta="http://pacificbiosciences.com/PacBioCollectionMetadata.xsd" xmlns:pbpn="http://pacificbiosciences.com/PacBioPartNumbers.xsd" xmlns:pbrk="http://pacificbiosciences.com/PacBioReagentKit.xsd" xmlns:pbsample="http://pacificbiosciences.com/PacBioSampleInfo.xsd" xmlns="http://pacificbiosciences.com/PacBioDatasets.xsd" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://pacificbiosciences.com/PacBioDatasets.xsd" xmlns:pbds="http://pacificbiosciences.com/PacBioDatasets.xsd">
+	<pbds:DataSetMetadata>
+		<pbds:TotalLength>42003795</pbds:TotalLength>
+		<pbds:NumRecords>7106</pbds:NumRecords>
+	</pbds:DataSetMetadata>
+	<pbbase:ExternalResources>
+		<pbbase:ExternalResource CreatedAt="2024-11-26T23:31:14.58Z" MetaType="PacBio.ConsensusReadFile.ConsensusReadBamFile" ResourceId="sample1.bc2058--bc2058.bam" TimeStampedName="pacbio_consensusreadfile_consensusreadbamfile-241126_23311458" UniqueId="f699df53-1ba5-4f5a-b6c6-5ffc6a2d76e4" Version="3.0.1">
+			<pbbase:FileIndices>
+				<pbbase:FileIndex CreatedAt="2024-11-26T23:31:14.58Z" MetaType="PacBio.Index.PacBioIndex" ResourceId="sample1.bc2058--bc2058.bam.pbi" TimeStampedName="pacbio_index_pacbioindex-241126_23311458" UniqueId="e8d1f3ac-e40e-4266-9804-706a1502c45a" Version="3.0.1" />
+			</pbbase:FileIndices>
+		</pbbase:ExternalResource>
+	</pbbase:ExternalResources>
+</pbds:ConsensusReadSet>
diff --git a/out/lima/sample1.consensusreadset.xml b/out/lima/sample1.consensusreadset.xml
@@ -0,0 +1,31 @@
+<?xml version="1.0" encoding="utf-8"?>
+<pbds:ConsensusReadSet CreatedAt="2024-11-26T23:31:14.61Z" MetaType="PacBio.DataSet.ConsensusReadSet" Name="sample1 (all samples)" Tags="barcoded" TimeStampedName="sample1 (all samples)-241126_23311463" UniqueId="7d17b9ae-fcfc-4e0b-8625-57d73263d8d6" Version="3.0.1" xmlns:pbbase="http://pacificbiosciences.com/PacBioBaseDataModel.xsd" xmlns:pbdm="http://pacificbiosciences.com/PacBioDataModel.xsd" xmlns:pbmeta="http://pacificbiosciences.com/PacBioCollectionMetadata.xsd" xmlns:pbpn="http://pacificbiosciences.com/PacBioPartNumbers.xsd" xmlns:pbrk="http://pacificbiosciences.com/PacBioReagentKit.xsd" xmlns:pbsample="http://pacificbiosciences.com/PacBioSampleInfo.xsd" xmlns="http://pacificbiosciences.com/PacBioDatasets.xsd" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://pacificbiosciences.com/PacBioDatasets.xsd" xmlns:pbds="http://pacificbiosciences.com/PacBioDatasets.xsd">
+	<pbbase:ExternalResources>
+		<pbbase:ExternalResource CreatedAt="2024-11-26T23:31:14.62Z" MetaType="PacBio.ConsensusReadFile.ConsensusReadBamFile" ResourceId="sample1.bc2026--bc2026.bam" TimeStampedName="pacbio_consensusreadfile_consensusreadbamfile-241126_23311462" UniqueId="29e50918-2bc8-445c-a012-4936b64d936f" Version="3.0.1">
+			<pbbase:FileIndices>
+				<pbbase:FileIndex CreatedAt="2024-11-26T23:31:14.62Z" MetaType="PacBio.Index.PacBioIndex" ResourceId="sample1.bc2026--bc2026.bam.pbi" TimeStampedName="pacbio_index_pacbioindex-241126_23311462" UniqueId="2f8060b9-8747-4be4-b8d2-73e2e04c2dea" Version="3.0.1" />
+			</pbbase:FileIndices>
+			<pbbase:ExternalResources>
+				<pbbase:ExternalResource CreatedAt="2024-11-26T23:31:14.62Z" MetaType="PacBio.DataSet.ConsensusReadSet" ResourceId="sample1.bc2026--bc2026.consensusreadset.xml" TimeStampedName="pacbio_dataset_consensusreadset-241126_23311462" UniqueId="84191489-145e-48a3-b063-9075bf85682b" Version="3.0.1" />
+			</pbbase:ExternalResources>
+		</pbbase:ExternalResource>
+		<pbbase:ExternalResource CreatedAt="2024-11-26T23:31:14.62Z" MetaType="PacBio.ConsensusReadFile.ConsensusReadBamFile" ResourceId="sample1.bc2058--bc2058.bam" TimeStampedName="pacbio_consensusreadfile_consensusreadbamfile-241126_23311462" UniqueId="7903b7c7-dd52-4d04-891a-17dcb3c8f15c" Version="3.0.1">
+			<pbbase:FileIndices>
+				<pbbase:FileIndex CreatedAt="2024-11-26T23:31:14.62Z" MetaType="PacBio.Index.PacBioIndex" ResourceId="sample1.bc2058--bc2058.bam.pbi" TimeStampedName="pacbio_index_pacbioindex-241126_23311462" UniqueId="f7744734-cace-479f-9716-6af09d4b1563" Version="3.0.1" />
+			</pbbase:FileIndices>
+			<pbbase:ExternalResources>
+				<pbbase:ExternalResource CreatedAt="2024-11-26T23:31:14.62Z" MetaType="PacBio.DataSet.ConsensusReadSet" ResourceId="sample1.bc2058--bc2058.consensusreadset.xml" TimeStampedName="pacbio_dataset_consensusreadset-241126_23311462" UniqueId="3b0c5135-4004-4d35-831c-cebe84889b4d" Version="3.0.1" />
+			</pbbase:ExternalResources>
+		</pbbase:ExternalResource>
+	</pbbase:ExternalResources>
+	<pbbase:SupplementalResources>
+		<pbbase:ExternalResource CreatedAt="2024-11-26T23:31:14.62Z" Description="Points to the barcodes.fasta file." MetaType="PacBio.BarcodeFile.BarcodeFastaFile" Name="Barcodes FASTA" ResourceId="barcodes_D01.fasta" TimeStampedName="pacbio_barcodefile_barcodefastafile-241126_23311462" UniqueId="e98a02d7-b273-4b52-a2c9-17e52d5def97" Version="3.0.1" />
+		<pbbase:ExternalResource CreatedAt="2024-11-26T23:31:14.62Z" MetaType="PacBio.FileTypes.txt" ResourceId="sample1.lima.counts" TimeStampedName="pacbio_filetypes_txt-241126_23311462" UniqueId="847bdaa5-f4d9-44e5-af3e-a64203c0d4a6" Version="3.0.1" />
+		<pbbase:ExternalResource CreatedAt="2024-11-26T23:31:14.62Z" MetaType="PacBio.FileTypes.txt" ResourceId="sample1.lima.report" TimeStampedName="pacbio_filetypes_txt-241126_23311462" UniqueId="effc21c3-b56f-43aa-96ed-244609037fbb" Version="3.0.1" />
+		<pbbase:ExternalResource CreatedAt="2024-11-26T23:31:14.62Z" MetaType="PacBio.FileTypes.txt" ResourceId="sample1.lima.summary" TimeStampedName="pacbio_filetypes_txt-241126_23311462" UniqueId="6f650b69-532d-4fcc-a6df-36f82a0f3e22" Version="3.0.1" />
+	</pbbase:SupplementalResources>
+	<pbds:DataSetMetadata>
+		<pbds:TotalLength>73970466</pbds:TotalLength>
+		<pbds:NumRecords>12831</pbds:NumRecords>
+	</pbds:DataSetMetadata>
+</pbds:ConsensusReadSet>
diff --git a/out/lima/sample1.json b/out/lima/sample1.json
@@ -0,0 +1,31 @@
+{
+  "createdAt": "2024-11-26T23:31:14.59Z",
+  "files": [
+    {
+      "createdAt": "2024-11-26T23:31:14.37Z",
+      "description": "Split BAM file by lima",
+      "fileSize": 11761071,
+      "fileTypeId": "PacBio.DataSet.ConsensusReadSet",
+      "isChunked": false,
+      "modifiedAt": "2024-11-26T23:31:14.38Z",
+      "name": "Barcodes bc2026--bc2026",
+      "path": "sample1.bc2026--bc2026.consensusreadset.xml",
+      "sourceId": "barcoding.tasks.lima-0",
+      "uniqueId": "d37de0ef-4929-4029-887e-91d4812bcff6"
+    },
+    {
+      "createdAt": "2024-11-26T23:31:14.59Z",
+      "description": "Split BAM file by lima",
+      "fileSize": 15873268,
+      "fileTypeId": "PacBio.DataSet.ConsensusReadSet",
+      "isChunked": false,
+      "modifiedAt": "2024-11-26T23:31:14.59Z",
+      "name": "Barcodes bc2058--bc2058",
+      "path": "sample1.bc2058--bc2058.consensusreadset.xml",
+      "sourceId": "barcoding.tasks.lima-0",
+      "uniqueId": "6c38803d-0e02-4593-b036-913aa8e05052"
+    }
+  ],
+  "updatedAt": "2024-11-26T23:31:14.59Z",
+  "version": "0.2.2"
+}
diff --git a/out/lima/sample1.lima.counts b/out/lima/sample1.lima.counts
@@ -0,0 +1,4 @@
+IdxFirst	IdxCombined	IdxFirstNamed	IdxCombinedNamed	Counts	MeanScore
+-1	-1	Not Barcoded	Not Barcoded	29	49
+0	0	bc2026	bc2026	5725	50
+1	1	bc2058	bc2058	7106	41