Merge pull request #614 from WackerO/umi

Umi test
nf-core · Jul 17, 2022 · e8d484f · e8d484f
2 parents 9683205 + eb07bf0
commit e8d484f
Show file tree

Hide file tree

Showing 8 changed files with 118 additions and 111 deletions.
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
@@ -59,6 +59,7 @@ jobs:
           - "tiddit"
           - "trimming"
           - "tumor_normal_pair"
+          - "umi"
           - "variantcalling_channel"
 
     steps:

diff --git a/conf/modules.config b/conf/modules.config
@@ -224,7 +224,7 @@ process {
 
     withName: 'BAM2FASTQ' {
         ext.args         = '-T RX'
-        ext.when         = { umi_read_structure }
+        ext.when         = { params.umi_read_structure }
     }
 
     withName: 'CALLUMICONSENSUS' {
@@ -238,6 +238,11 @@ process {
             ext.args2        = '-bS'
             ext.prefix       = {"${meta.id}.umi_unsorted"}
         }
+
+        withName: ".*DRAGMAP_ALIGN" {
+            ext.args2        = '-bS'
+            ext.prefix       = {"${meta.id}.umi_unsorted"}
+        }
     }
 
     withName: 'SAMBLASTER' {
@@ -247,17 +252,16 @@ process {
 
 // MAPPING
     if (params.step == 'mapping') {
-        withName: ".*GATK4_MAPPING:BWAMEM1_MEM" {
+        withName: ".*BWAMEM1_MEM" {
             ext.when         = { params.aligner == "bwa-mem" }
         }
-        withName: ".*GATK4_MAPPING:BWAMEM2_MEM" {
+        withName: ".*BWAMEM2_MEM" {
             ext.when         = { params.aligner == "bwa-mem2" }
         }
 
-        withName: ".*GATK4_MAPPING:DRAGMAP_ALIGN" {
+        withName: ".*DRAGMAP_ALIGN" {
             ext.when         = { params.aligner == "dragmap" }
             ext.args         = { "--RGSM ${meta.read_group}" }
-
         }
 
         withName: "NFCORE_SAREK:SAREK:GATK4_MAPPING:(BWAMEM.*_MEM|DRAGMAP_ALIGN)" {

diff --git a/modules.json b/modules.json
@@ -76,13 +76,13 @@
                 "git_sha": "49b18b1639f4f7104187058866a8fab33332bdfe"
             },
             "fgbio/callmolecularconsensusreads": {
-                "git_sha": "e745e167c1020928ef20ea1397b6b4d230681b4d"
+                "git_sha": "6720d88f4e46e67b9a225f06bfb10c8e8ae04a84"
             },
             "fgbio/fastqtobam": {
-                "git_sha": "e745e167c1020928ef20ea1397b6b4d230681b4d"
+                "git_sha": "6720d88f4e46e67b9a225f06bfb10c8e8ae04a84"
             },
             "fgbio/groupreadsbyumi": {
-                "git_sha": "e745e167c1020928ef20ea1397b6b4d230681b4d"
+                "git_sha": "6720d88f4e46e67b9a225f06bfb10c8e8ae04a84"
             },
             "freebayes": {
                 "git_sha": "e745e167c1020928ef20ea1397b6b4d230681b4d"

diff --git a/modules/nf-core/modules/fgbio/callmolecularconsensusreads/main.nf b/modules/nf-core/modules/fgbio/callmolecularconsensusreads/main.nf
diff --git a/modules/nf-core/modules/fgbio/fastqtobam/main.nf b/modules/nf-core/modules/fgbio/fastqtobam/main.nf
diff --git a/modules/nf-core/modules/fgbio/groupreadsbyumi/main.nf b/modules/nf-core/modules/fgbio/groupreadsbyumi/main.nf
diff --git a/tests/test_umi.yml b/tests/test_umi.yml
@@ -1,93 +1,94 @@
-- name: Run UMI Quiaseq test
-  command: nextflow run main.nf -profile test,umi_quiaseq,docker
+- name: Run UMI test
+  command: nextflow run main.nf -profile test,umi,docker
   tags:
-    - umi_quiaseq
     - umi
   files:
-    - path: results/fastqtobam/1234N-SRR7545951_umi_converted.bam
-      md5sum: 1473c98be04aff10f80e2946b111c459
-    - path: results/bam2fastq/1234N-SRR7545951_interleaved.fq.gz
-      md5sum: 17c0cb479e8117016e9f0d835d4df50f
-    - path: results/callumiconsensus/1234N-SRR7545951_umi-consensus.bam
-      md5sum: 3be1adf27d473de12cab895423a5f3ed
-    - path: results/cat/1234N-SRR7545951_1.merged.fastq.gz
-      md5sum: c1ed636a11b3031c72f3d4c84a7a0949
-    - path: results/cat/1234N-SRR7545951_2.merged.fastq.gz
-      md5sum: 0bbef631b8de80f2abe441de713cbb49
-    - path: results/groupreadsbyumi/1234N-SRR7545951_umi-grouped.bam
-      md5sum: e3bb348056513c46d7a60df4d7a8cc97
-    - path: results/groupreadsbyumi/1234N-SRR7545951_umi_histogram.txt
-      md5sum: 2e1e9a774135675a3d1a3b0a88d29c53
-    - path: results/samblaster/1234N-SRR7545951_unsorted_tagged.bam
-      md5sum: daa51939768c256315ce700dabaccad6
-    - path: results/samtools/1234N-SRR7545951.map_map.bam
-      md5sum: 741de784bb83815fab8b2d23d88c74d7
-    - path: results/samtools/1234N-SRR7545951.map_unmap.bam
-      md5sum: 31627b2d163da372584bb78258c046e5
-    - path: results/samtools/1234N-SRR7545951.mapped_1.fq.gz
-      md5sum: 709872fc2910431b1e8b7074bfe38c67
-    - path: results/samtools/1234N-SRR7545951.mapped_2.fq.gz
-      md5sum: 709872fc2910431b1e8b7074bfe38c67
-    - path: results/samtools/1234N-SRR7545951.mapped_other.fq.gz
-      md5sum: 709872fc2910431b1e8b7074bfe38c67
-    - path: results/samtools/1234N-SRR7545951.mapped_singleton.fq.gz
-      md5sum: 709872fc2910431b1e8b7074bfe38c67
-    - path: results/samtools/1234N-SRR7545951.unmap_map.bam
-      md5sum: 42fb9dbb5a157c194c77b1de5b4f3b39
-    - path: results/samtools/1234N-SRR7545951.unmap_unmap.bam
-      md5sum: 2075e48cd007f29c8f6fdc518e587e27
-    - path: results/samtools/1234N-SRR7545951.unmapped_1.fq.gz
-      md5sum: 1f95ae5bbc7345372005b6673884f853
-    - path: results/samtools/1234N-SRR7545951.unmapped_2.fq.gz
-      md5sum: 9902e27a99c2a628d184cb91759d6ff7
-    - path: results/samtools/1234N-SRR7545951.unmapped_other.fq.gz
-      md5sum: 709872fc2910431b1e8b7074bfe38c67
-    - path: results/samtools/1234N-SRR7545951.unmapped_singleton.fq.gz
-      md5sum: 709872fc2910431b1e8b7074bfe38c67
+    - path: results/fastqtobam/test-test_L1_umi_converted.bam
+      md5sum: ce67e3b1f46302127cd07cb3c2d2e739
+    - path: results/bam2fastq/test-test_L1_interleaved.fq.gz
+      md5sum: 7a57a8ab53dd1d799cca67a85c47ccd9
+    - path: results/callumiconsensus/test-test_L1_umi-consensus.bam
+      md5sum: ccd8b0b95119b6f42655800cf9d07975
+    - path: results/cat/test-test_L1_1.merged.fastq.gz
+      md5sum: 401d4e7ccead0696d6d77e0f679fc608
+    - path: results/cat/test-test_L1_2.merged.fastq.gz
+      md5sum: 01e6167468f6c46c6455a3e6bb6c4d1b
+    - path: results/groupreadsbyumi/test-test_L1_umi-grouped.bam
+      md5sum: 7f7775bd971c6c6790642cb14e7140a0
+    - path: results/groupreadsbyumi/test-test_L1_umi_histogram.txt
+      md5sum: f2f29523bb7016e1a78fe634466cbfba
+    - path: results/samblaster/test-test_L1_unsorted_tagged.bam
+      md5sum: b6b55546a1134b1083fd50b8f36d5a13
+    - path: results/samtools/test-test_L1.map_map.bam
+      md5sum: f29dc564311e592941bdf152dd722b4f
+    - path: results/samtools/test-test_L1.map_unmap.bam
+      md5sum: 6518cea4d9c0bdb17af66932b68e0ad9
+    - path: results/samtools/test-test_L1.bam
+    - path: results/samtools/test-test_L1.unmap_map.bam
+      md5sum: 9479aabdbbfdc8f5c79f8daa1011e9fb
+    - path: results/samtools/test-test_L1.unmap_unmap.bam
+      md5sum: a1801ce12b5b4a486c5b4cba49fad0dd
+      # TODO: The following files are not produced anymore
+#    - path: results/samtools/test-test_L1.mapped_1.fq.gz
+#      md5sum: 709872fc2910431b1e8b7074bfe38c67
+#    - path: results/samtools/test-test_L1.mapped_2.fq.gz
+#      md5sum: 709872fc2910431b1e8b7074bfe38c67
+#    - path: results/samtools/test-test_L1.mapped_other.fq.gz
+#      md5sum: 709872fc2910431b1e8b7074bfe38c67
+#    - path: results/samtools/test-test_L1.mapped_singleton.fq.gz
+#      md5sum: 709872fc2910431b1e8b7074bfe38c67
+#    - path: results/samtools/test-test_L1.unmapped_1.fq.gz
+#      md5sum: 1f95ae5bbc7345372005b6673884f853
+#    - path: results/samtools/test-test_L1.unmapped_2.fq.gz
+#      md5sum: 9902e27a99c2a628d184cb91759d6ff7
+#    - path: results/samtools/test-test_L1.unmapped_other.fq.gz
+#      md5sum: 709872fc2910431b1e8b7074bfe38c67
+#    - path: results/samtools/test-test_L1.unmapped_singleton.fq.gz
+#      md5sum: 709872fc2910431b1e8b7074bfe38c67
 
-- name: Run UMI TSO test
-  command: nextflow run main.nf -profile test,umi_tso,docker
-  tags:
-    - umi_tso
-    - umi
-  files:
-    - path: results/fastqtobam/1234N-HT1080_umi_converted.bam
-      md5sum: aaf4b9a80f4b780d93820df8b536c531
-    - path: results/bam2fastq/1234N-HT1080_interleaved.fq.gz
-      md5sum: b037fe8e180e805226485d9bea4a69e5
-    - path: results/callumiconsensus/1234N-HT1080_umi-consensus.bam
-      md5sum: 807bfe6a82eabdf29aaf8bc6a832f0cd
-    - path: results/cat/1234N-HT1080_1.merged.fastq.gz
-      md5sum: c8f0d1308df4c46c0716ccfe76f78fb4
-    - path: results/cat/1234N-HT1080_2.merged.fastq.gz
-      md5sum: 1b42767b2d5dbc99e447c41d6d2418ab
-    - path: results/groupreadsbyumi/1234N-HT1080_umi-grouped.bam
-      md5sum: c9c76b410ba208fea76c79cecfb45241
-    - path: results/groupreadsbyumi/1234N-HT1080_umi_histogram.txt
-      md5sum: 3dec92ae49763c10c0b2e79827d25ffb
-    - path: results/samblaster/1234N-HT1080_unsorted_tagged.bam
-      md5sum: 628bd970fa00615fb9f841bad7c8f597
-    - path: results/samtools/1234N-HT1080.map_map.bam
-      md5sum: 2689e640c3ededf70228a44ee4b1de48
-    - path: results/samtools/1234N-HT1080.map_unmap.bam
-      md5sum: 37300fea69f7c400220a03ca5854d04b
-    - path: results/samtools/1234N-HT1080.mapped_1.fq.gz
-      md5sum: 709872fc2910431b1e8b7074bfe38c67
-    - path: results/samtools/1234N-HT1080.mapped_2.fq.gz
-      md5sum: 709872fc2910431b1e8b7074bfe38c67
-    - path: results/samtools/1234N-HT1080.mapped_other.fq.gz
-      md5sum: 709872fc2910431b1e8b7074bfe38c67
-    - path: results/samtools/1234N-HT1080.mapped_singleton.fq.gz
-      md5sum: 709872fc2910431b1e8b7074bfe38c67
-    - path: results/samtools/1234N-HT1080.unmap_map.bam
-      md5sum: 6267d0ed8274e1fa7aad8d1e71d98696
-    - path: results/samtools/1234N-HT1080.unmap_unmap.bam
-      md5sum: 32ed07aa23900f5722e5983a646a5ac2
-    - path: results/samtools/1234N-HT1080.unmapped_1.fq.gz
-      md5sum: d9a7a279c05563a37d9a5a6938be68fb
-    - path: results/samtools/1234N-HT1080.unmapped_2.fq.gz
-      md5sum: afe191dcd48360f96bdb583fa60ff400
-    - path: results/samtools/1234N-HT1080.unmapped_other.fq.gz
-      md5sum: 709872fc2910431b1e8b7074bfe38c67
-    - path: results/samtools/1234N-HT1080.unmapped_singleton.fq.gz
-      md5sum: 709872fc2910431b1e8b7074bfe38c67
+# - name: Run UMI TSO test
+#   command: nextflow run main.nf -profile test,umi_tso,docker
+#   tags:
+#     - umi_tso
+#     - umi
+#   files:
+#     - path: results/fastqtobam/1234N-HT1080_umi_converted.bam
+#       md5sum: aaf4b9a80f4b780d93820df8b536c531
+#     - path: results/bam2fastq/1234N-HT1080_interleaved.fq.gz
+#       md5sum: b037fe8e180e805226485d9bea4a69e5
+#     - path: results/callumiconsensus/1234N-HT1080_umi-consensus.bam
+#       md5sum: 807bfe6a82eabdf29aaf8bc6a832f0cd
+#     - path: results/cat/1234N-HT1080_1.merged.fastq.gz
+#       md5sum: c8f0d1308df4c46c0716ccfe76f78fb4
+#     - path: results/cat/1234N-HT1080_2.merged.fastq.gz
+#       md5sum: 1b42767b2d5dbc99e447c41d6d2418ab
+#     - path: results/groupreadsbyumi/1234N-HT1080_umi-grouped.bam
+#       md5sum: c9c76b410ba208fea76c79cecfb45241
+#     - path: results/groupreadsbyumi/1234N-HT1080_umi_histogram.txt
+#       md5sum: 3dec92ae49763c10c0b2e79827d25ffb
+#     - path: results/samblaster/1234N-HT1080_unsorted_tagged.bam
+#       md5sum: 628bd970fa00615fb9f841bad7c8f597
+#     - path: results/samtools/1234N-HT1080.map_map.bam
+#       md5sum: 2689e640c3ededf70228a44ee4b1de48
+#     - path: results/samtools/1234N-HT1080.map_unmap.bam
+#       md5sum: 37300fea69f7c400220a03ca5854d04b
+#     - path: results/samtools/1234N-HT1080.mapped_1.fq.gz
+#       md5sum: 709872fc2910431b1e8b7074bfe38c67
+#     - path: results/samtools/1234N-HT1080.mapped_2.fq.gz
+#       md5sum: 709872fc2910431b1e8b7074bfe38c67
+#     - path: results/samtools/1234N-HT1080.mapped_other.fq.gz
+#       md5sum: 709872fc2910431b1e8b7074bfe38c67
+#     - path: results/samtools/1234N-HT1080.mapped_singleton.fq.gz
+#       md5sum: 709872fc2910431b1e8b7074bfe38c67
+#     - path: results/samtools/1234N-HT1080.unmap_map.bam
+#       md5sum: 6267d0ed8274e1fa7aad8d1e71d98696
+#     - path: results/samtools/1234N-HT1080.unmap_unmap.bam
+#       md5sum: 32ed07aa23900f5722e5983a646a5ac2
+#     - path: results/samtools/1234N-HT1080.unmapped_1.fq.gz
+#       md5sum: d9a7a279c05563a37d9a5a6938be68fb
+#     - path: results/samtools/1234N-HT1080.unmapped_2.fq.gz
+#       md5sum: afe191dcd48360f96bdb583fa60ff400
+#     - path: results/samtools/1234N-HT1080.unmapped_other.fq.gz
+#       md5sum: 709872fc2910431b1e8b7074bfe38c67
+#     - path: results/samtools/1234N-HT1080.unmapped_singleton.fq.gz
+#       md5sum: 709872fc2910431b1e8b7074bfe38c67
diff --git a/workflows/sarek.nf b/workflows/sarek.nf
@@ -388,8 +388,10 @@ workflow SAREK {
                 umi_read_structure,
                 params.group_by_umi_strategy)
 
+            bamtofastq = CREATE_UMI_CONSENSUS.out.consensusbam.map{meta, bam -> [meta,bam,[]]}
+
             // convert back to fastq for further preprocessing
-            ALIGNMENT_TO_FASTQ_UMI(CREATE_UMI_CONSENSUS.out.consensusbam, [])
+            ALIGNMENT_TO_FASTQ_UMI(bamtofastq, [])
 
             ch_reads_fastp = ALIGNMENT_TO_FASTQ_UMI.out.reads