Add bamToFastq module

modules/bamToFastq/samtools/main.nf

@@ -0,0 +1,87 @@
+include { parseJSON } from "../../functions"
+
+params.samtoolsVersion = '1.19.2--h50ea8bc_1'
+params.samtoolsContainer = "${params.containerRepo}/samtools:${params.samtoolsVersion}"
+params.bamstatsVersion = '0.3.5--he881be0_0'
+params.bamStatsContainer = "quay.io/biocontainers/bamstats:${params.bamstatsVersion}"
+params.bamStatsMaxBuf = '1000000'
+params.bamStatsLogLevel = 'info'
+
+process getProtocol {
+
+    tag "${sample}"
+    container params.bamStatsContainer
+
+    input:
+    tuple val(sample), val(id), path(bam), val(type), val(view)
+
+
+    output:
+    tuple val(sample), val(id),  stdout
+
+
+    script:
+    prefix = "${sample}"
+
+    """
+    bamstats -c ${task.cpus} \\
+             -i ${bam} \\
+             -n 1000 \\
+             --max-buf ${params.bamStatsMaxBuf} \\
+             --loglevel ${params.bamStatsLogLevel}
+    """
+}
+
+process toFastq {
+
+    tag "${sample}"
+    container params.samtoolsContainer
+
+    input:
+    tuple val(sample), val(id), path(bam), val(type), val(view), val(protocol)
+
+    output:
+    tuple val(sample), val(id), path("${prefix}*.fastq.gz"), val(type), val(view)
+
+    script:
+    prefix = "${id}"
+    type = "fastq"
+    view = "FqRd"
+
+    def cmd = []
+    def inputBam = bam
+    outParams = "-o ${prefix}.fastq.gz"
+    if ( protocol == "PairedEnd" ) {
+	view = ['FqRd1', 'FqRd2']
+	inputBam = "${prefix}.collate.bam"
+	outParams = "-1 ${prefix}_1.fastq.gz -2 ${prefix}_2.fastq.gz"
+	cmd << "mkfifo ${prefix}.collate.bam"
+	cmd << "samtools collate -o ${prefix}.collate.bam --threads ${task.cpus} ${bam} &"
+    }
+
+    cmd << "samtools fastq -N -O -c 9 --threads ${task.cpus} ${outParams} ${inputBam}"
+
+    cmd.join('\n')
+}
+
+workflow bamToFastq {
+    take:
+      genomeAlignments
+
+    main:
+
+    getProtocol( genomeAlignments )
+    getProtocol.out.map {
+	d = parseJSON(it[-1])
+	it[0..1] + [ d.general.protocol ]
+    }.set { protocol }
+
+    genomeAlignments
+	.join( protocol, by: [0,1] )
+	.set { bams }
+
+    toFastq( bams )
+
+    emit:
+      toFastq.out
+}

tests/modules/bamToFastq/samtools/main.getprotocol.nf.test

@@ -0,0 +1,26 @@
+nextflow_process {
+
+    name "Test Process getProtocol"
+    script "modules/bamToFastq/samtools/main.nf"
+    process "getProtocol"
+
+    test("Should get sequencing protocol from BAM file") {
+
+        when {
+            process {
+                """
+                input[0] = Channel.from([
+                    [ "sample3", "test3", file("${baseDir}/data/sample3_m4_n10_toGenome.bam"), "bam", "GenomeAlignments" ]
+                ])
+                """
+            }
+        }
+
+        then {
+            assert process.success
+            assert snapshot(process.out).match()
+        }
+
+    }
+
+}

tests/modules/bamToFastq/samtools/main.getprotocol.nf.test.snap

@@ -0,0 +1,20 @@
+{
+    "Should get sequencing protocol from BAM file": {
+        "content": [
+            {
+                "0": [
+                    [
+                        "sample3",
+                        "test3",
+                        "{\n\t\"general\": {\n\t\t\"protocol\": \"PairedEnd\",\n\t\t\"reads\": {\n\t\t\t\"total\": 1000,\n\t\t\t\"mapped\": {\n\t\t\t\t\"1\": 998,\n\t\t\t\t\"2\": 2\n\t\t\t},\n\t\t\t\"mappings\": {\n\t\t\t\t\"ratio\": 1.002,\n\t\t\t\t\"count\": 1002\n\t\t\t}\n\t\t},\n\t\t\"pairs\": {\n\t\t\t\"total\": 489,\n\t\t\t\"mapped\": {\n\t\t\t\t\"1\": 488,\n\t\t\t\t\"2\": 1\n\t\t\t},\n\t\t\t\"insert_sizes\": {\n\t\t\t\t\"74\": 1,\n\t\t\t\t\"102\": 1,\n\t\t\t\t\"104\": 1,\n\t\t\t\t\"107\": 1,\n\t\t\t\t\"111\": 2,\n\t\t\t\t\"112\": 1,\n\t\t\t\t\"113\": 2,\n\t\t\t\t\"123\": 1,\n\t\t\t\t\"124\": 2,\n\t\t\t\t\"125\": 1,\n\t\t\t\t\"126\": 3,\n\t\t\t\t\"127\": 2,\n\t\t\t\t\"129\": 2,\n\t\t\t\t\"133\": 3,\n\t\t\t\t\"135\": 2,\n\t\t\t\t\"137\": 2,\n\t\t\t\t\"141\": 2,\n\t\t\t\t\"143\": 2,\n\t\t\t\t\"144\": 1,\n\t\t\t\t\"150\": 1,\n\t\t\t\t\"152\": 2,\n\t\t\t\t\"153\": 3,\n\t\t\t\t\"154\": 2,\n\t\t\t\t\"155\": 1,\n\t\t\t\t\"156\": 2,\n\t\t\t\t\"157\": 1,\n\t\t\t\t\"158\": 1,\n\t\t\t\t\"159\": 1,\n\t\t\t\t\"160\": 4,\n\t\t\t\t\"161\": 5,\n\t\t\t\t\"162\": 3,\n\t\t\t\t\"163\": 4,\n\t\t\t\t\"164\": 8,\n\t\t\t\t\"165\": 4,\n\t\t\t\t\"166\": 7,\n\t\t\t\t\"167\": 6,\n\t\t\t\t\"168\": 11,\n\t\t\t\t\"169\": 9,\n\t\t\t\t\"170\": 12,\n\t\t\t\t\"171\": 4,\n\t\t\t\t\"172\": 3,\n\t\t\t\t\"173\": 6,\n\t\t\t\t\"174\": 13,\n\t\t\t\t\"175\": 12,\n\t\t\t\t\"176\": 9,\n\t\t\t\t\"177\": 12,\n\t\t\t\t\"178\": 12,\n\t\t\t\t\"179\": 11,\n\t\t\t\t\"180\": 15,\n\t\t\t\t\"181\": 10,\n\t\t\t\t\"182\": 11,\n\t\t\t\t\"183\": 11,\n\t\t\t\t\"184\": 13,\n\t\t\t\t\"185\": 10,\n\t\t\t\t\"186\": 14,\n\t\t\t\t\"187\": 10,\n\t\t\t\t\"188\": 10,\n\t\t\t\t\"189\": 8,\n\t\t\t\t\"190\": 4,\n\t\t\t\t\"191\": 4,\n\t\t\t\t\"192\": 11,\n\t\t\t\t\"193\": 11,\n\t\t\t\t\"194\": 9,\n\t\t\t\t\"195\": 11,\n\t\t\t\t\"196\": 10,\n\t\t\t\t\"197\": 8,\n\t\t\t\t\"198\": 3,\n\t\t\t\t\"199\": 5,\n\t\t\t\t\"200\": 8,\n\t\t\t\t\"201\": 5,\n\t\t\t\t\"202\": 7,\n\t\t\t\t\"203\": 5,\n\t\t\t\t\"204\": 6,\n\t\t\t\t\"205\": 5,\n\t\t\t\t\"206\": 2,\n\t\t\t\t\"208\": 2,\n\t\t\t\t\"209\": 6,\n\t\t\t\t\"210\": 4,\n\t\t\t\t\"211\": 2,\n\t\t\t\t\"212\": 1,\n\t\t\t\t\"213\": 2,\n\t\t\t\t\"214\": 3,\n\t\t\t\t\"215\": 2,\n\t\t\t\t\"216\": 1,\n\t\t\t\t\"217\": 2,\n\t\t\t\t\"218\": 2,\n\t\t\t\t\"219\": 4,\n\t\t\t\t\"220\": 1,\n\t\t\t\t\"221\": 2,\n\t\t\t\t\"222\": 1,\n\t\t\t\t\"223\": 1,\n\t\t\t\t\"224\": 2,\n\t\t\t\t\"233\": 1,\n\t\t\t\t\"235\": 2,\n\t\t\t\t\"237\": 1,\n\t\t\t\t\"7228\": 1,\n\t\t\t\t\"11975\": 1,\n\t\t\t\t\"20827\": 1,\n\t\t\t\t\"59774\": 1,\n\t\t\t\t\"59790\": 2,\n\t\t\t\t\"59793\": 1,\n\t\t\t\t\"59797\": 1,\n\t\t\t\t\"59799\": 1,\n\t\t\t\t\"59800\": 1,\n\t\t\t\t\"59805\": 1,\n\t\t\t\t\"59810\": 1,\n\t\t\t\t\"59812\": 1,\n\t\t\t\t\"59816\": 1,\n\t\t\t\t\"59817\": 1,\n\t\t\t\t\"59828\": 1,\n\t\t\t\t\"59829\": 1,\n\t\t\t\t\"59830\": 1,\n\t\t\t\t\"68769\": 1,\n\t\t\t\t\"80023\": 1\n\t\t\t}\n\t\t}\n\t}\n}"
+                    ]
+                ]
+            }
+        ],
+        "meta": {
+            "nf-test": "0.8.4",
+            "nextflow": "24.04.2"
+        },
+        "timestamp": "2024-06-21T16:29:27.186894"
+    }
+}

tests/modules/bamToFastq/samtools/main.nf.test

@@ -0,0 +1,26 @@
+nextflow_workflow {
+
+    name "Test Workflow bamToFastq"
+    script "modules/bamToFastq/samtools/main.nf"
+    workflow "bamToFastq"
+
+    test("Should convert BAM to FastQ (detecting protocol)") {
+
+        when {
+            workflow {
+                """
+                input[0] = Channel.from([
+                    [ "sample3", "test3", file("${baseDir}/data/sample3_m4_n10_toGenome.bam"), "bam", "GenomeAlignments" ]
+                ])
+                """
+            }
+        }
+
+        then {
+            assert workflow.success
+            assert snapshot(workflow.out).match()
+        }
+
+    }
+
+}

tests/modules/bamToFastq/samtools/main.nf.test.snap

@@ -0,0 +1,28 @@
+{
+    "Should convert BAM to FastQ (detecting protocol)": {
+        "content": [
+            {
+                "0": [
+                    [
+                        "sample3",
+                        "test3",
+                        [
+                            "test3_1.fastq.gz:md5,acac9b626a3bba40b309c4fb880360f7",
+                            "test3_2.fastq.gz:md5,877a915ea5eb84a315f41a1fa54bf5f1"
+                        ],
+                        "fastq",
+                        [
+                            "FqRd1",
+                            "FqRd2"
+                        ]
+                    ]
+                ]
+            }
+        ],
+        "meta": {
+            "nf-test": "0.8.4",
+            "nextflow": "24.04.2"
+        },
+        "timestamp": "2024-06-21T16:29:39.476232"
+    }
+}

tests/modules/bamToFastq/samtools/main.tofastq.nf.test

@@ -0,0 +1,26 @@
+nextflow_process {
+
+    name "Test Process toFastq"
+    script "modules/bamToFastq/samtools/main.nf"
+    process "toFastq"
+
+    test("Should convert BAM to FastQ") {
+
+        when {
+            process {
+                """
+                input[0] = Channel.from([
+                    [ "sample3", "test3", file("${baseDir}/data/sample3_m4_n10_toGenome.bam"), "bam", "GenomeAlignments", "PairedEnd"]
+                ])
+                """
+            }
+        }
+
+        then {
+            assert process.success
+            assert snapshot(process.out).match()
+        }
+
+    }
+
+}

tests/modules/bamToFastq/samtools/main.tofastq.nf.test.snap

@@ -0,0 +1,28 @@
+{
+    "Should convert BAM to FastQ": {
+        "content": [
+            {
+                "0": [
+                    [
+                        "sample3",
+                        "test3",
+                        [
+                            "test3_1.fastq.gz:md5,acac9b626a3bba40b309c4fb880360f7",
+                            "test3_2.fastq.gz:md5,877a915ea5eb84a315f41a1fa54bf5f1"
+                        ],
+                        "fastq",
+                        [
+                            "FqRd1",
+                            "FqRd2"
+                        ]
+                    ]
+                ]
+            }
+        ],
+        "meta": {
+            "nf-test": "0.8.4",
+            "nextflow": "24.04.2"
+        },
+        "timestamp": "2024-06-21T16:29:17.900642"
+    }
+}