Add output index type to samtools view

nf-core · Feb 14, 2025 · bb3ec89 · bb3ec89
1 parent 27978ac
commit bb3ec89
Show file tree

Hide file tree

Showing 4 changed files with 867 additions and 145 deletions.
diff --git a/modules/nf-core/samtools/view/main.nf b/modules/nf-core/samtools/view/main.nf
@@ -11,6 +11,7 @@ process SAMTOOLS_VIEW {
     tuple val(meta), path(input), path(index)
     tuple val(meta2), path(fasta)
     path qname
+    val index_format
 
     output:
     tuple val(meta), path("${prefix}.bam"),                                    emit: bam,              optional: true
@@ -20,7 +21,7 @@ process SAMTOOLS_VIEW {
     tuple val(meta), path("${prefix}.${file_type}.csi"),                       emit: csi,              optional: true
     tuple val(meta), path("${prefix}.${file_type}.crai"),                      emit: crai,             optional: true
     tuple val(meta), path("${prefix}.unselected.${file_type}"),                emit: unselected,       optional: true
-    tuple val(meta), path("${prefix}.unselected.${file_type}.{bai,csi,crsi}"), emit: unselected_index, optional: true
+    tuple val(meta), path("${prefix}.unselected.${file_type}.{csi,crai}"),     emit: unselected_index, optional: true
     path  "versions.yml",                                                      emit: versions
 
     when:
@@ -35,16 +36,27 @@ process SAMTOOLS_VIEW {
                 args.contains("--output-fmt bam") ? "bam" :
                 args.contains("--output-fmt cram") ? "cram" :
                 input.getExtension()
+
+    output_file = index_format ? "${prefix}.${file_type}##idx##${prefix}.${file_type}.${index_format}" : "${prefix}.${file_type}"
+    // Can't choose index type of unselected file
     readnames = qname ? "--qname-file ${qname} --output-unselected ${prefix}.unselected.${file_type}": ""
+
     if ("$input" == "${prefix}.${file_type}") error "Input and output names are the same, use \"task.ext.prefix\" to disambiguate!"
+    if (index_format) {
+        if (!index_format.matches('bai|csi|crai')) {
+            error "Index format not one of bai, csi, crai."
+        } else if (file_type == "sam") {
+            error "Indexing not compatible with SAM output"
+        }
+    }
     """
     samtools \\
         view \\
         --threads ${task.cpus-1} \\
         ${reference} \\
         ${readnames} \\
         $args \\
-        -o ${prefix}.${file_type} \\
+        -o ${output_file} \\
         $input \\
         $args2
 
@@ -61,13 +73,27 @@ process SAMTOOLS_VIEW {
                 args.contains("--output-fmt bam") ? "bam" :
                 args.contains("--output-fmt cram") ? "cram" :
                 input.getExtension()
-    if ("$input" == "${prefix}.${file_type}") error "Input and output names are the same, use \"task.ext.prefix\" to disambiguate!"
-
-    index = args.contains("--write-index") ? "touch ${prefix}.${file_type}.csi" : ""
+    default_index_format =
+        file_type == "bam" ? "csi" :
+        file_type == "cram" ? "crai" : ""
+    index = args.contains("--write-index") && index_format ? "touch ${prefix}.${file_type}.${index_format}" : args.contains("--write-index") ? "touch ${prefix}.${file_type}.${default_index_format}" : ""
+    unselected = qname ? "touch ${prefix}.unselected.${file_type}" : ""
+    // Can't choose index type of unselected file
+    unselected_index = qname && args.contains("--write-index") ? "touch ${prefix}.unselected.${file_type}.${default_index_format}" : ""
 
+    if ("$input" == "${prefix}.${file_type}") error "Input and output names are the same, use \"task.ext.prefix\" to disambiguate!"
+    if (index_format) {
+        if (!index_format.matches('bai|csi|crai')) {
+            error "Index format not one of bai, csi, crai."
+        } else if (file_type == "sam") {
+            error "Indexing not compatible with SAM output."
+        }
+    }
     """
     touch ${prefix}.${file_type}
     ${index}
+    ${unselected}
+    ${unselected_index}
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":

diff --git a/modules/nf-core/samtools/view/tests/cram_index.config b/modules/nf-core/samtools/view/tests/cram_index.config
@@ -0,0 +1,3 @@
+process {
+    ext.args = "--output-fmt cram --write-index"
+}
diff --git a/modules/nf-core/samtools/view/tests/main.nf.test b/modules/nf-core/samtools/view/tests/main.nf.test
@@ -21,6 +21,7 @@ nextflow_process {
                 ])
                 input[1] = [[],[]]
                 input[2] = []
+                input[3] = []
                 """
             }
         }
@@ -39,6 +40,135 @@ nextflow_process {
         }
     }
 
+    test("bam_csi_index") {
+
+        config "./bam_index.config"
+
+        when {
+            process {
+                """
+                input[0] = Channel.of([
+                    [ id:'test', single_end:false ], // meta map
+                    file(params.modules_testdata_base_path + 'genomics/sarscov2/illumina/bam/test.paired_end.sorted.bam', checkIfExists: true),
+                    []
+                ])
+                input[1] = [[],[]]
+                input[2] = []
+                input[3] = 'csi'
+                """
+            }
+        }
+
+        then {
+            assertAll(
+                { assert process.success },
+                { assert snapshot(
+                    file(process.out.bam[0][1]).name,
+                    file(process.out.csi[0][1]).name,
+                    process.out.versions).match()
+                }
+            )
+        }
+    }
+
+    test("bam_bai_index") {
+
+        config "./bam_index.config"
+
+        when {
+            process {
+                """
+                input[0] = Channel.of([
+                    [ id:'test', single_end:false ], // meta map
+                    file(params.modules_testdata_base_path + 'genomics/sarscov2/illumina/bam/test.paired_end.sorted.bam', checkIfExists: true),
+                    []
+                ])
+                input[1] = [[],[]]
+                input[2] = []
+                input[3] = 'bai'
+                """
+            }
+        }
+
+        then {
+            assertAll(
+                { assert process.success },
+                { assert snapshot(
+                    file(process.out.bam[0][1]).name,
+                    file(process.out.bai[0][1]).name,
+                    process.out.versions).match() }
+            )
+        }
+    }
+
+    test("bam_bai_index_unselected") {
+
+        config "./bam_index.config"
+
+        when {
+            process {
+                """
+                input[0] = Channel.of([
+                    [ id:'test', single_end:false ], // meta map
+                    file(params.modules_testdata_base_path + 'genomics/sarscov2/illumina/bam/test.paired_end.sorted.bam', checkIfExists: true),
+                    []
+                ])
+                input[1] = [[],[]]
+                input[2] = Channel.of('testN:1')
+                    .collectFile(name: 'selected_reads.txt')
+                input[3] = 'bai'
+                """
+            }
+        }
+
+        then {
+            assertAll(
+                { assert process.success },
+                { assert snapshot(
+                    file(process.out.bam[0][1]).name,
+                    file(process.out.bai[0][1]).name,
+                    file(process.out.unselected[0][1]).name,
+                    file(process.out.unselected_index[0][1]).name,
+                    process.out.versions).match()
+                }
+            )
+        }
+    }
+
+    test("cram_crai_index_unselected") {
+
+        config "./cram_index.config"
+
+        when {
+            process {
+                """
+                input[0] = Channel.of([
+                    [ id:'test', single_end:false ], // meta map
+                    file(params.modules_testdata_base_path + 'genomics/sarscov2/illumina/bam/test.paired_end.sorted.bam', checkIfExists: true),
+                    []
+                ])
+                input[1] = [[],[]]
+                input[2] = Channel.of('testN:1')
+                    .collectFile(name: 'selected_reads.txt')
+                input[3] = 'crai'
+                """
+            }
+        }
+
+        then {
+            assertAll(
+                { assert process.success },
+                { assert snapshot(
+                    file(process.out.cram[0][1]).name,
+                    file(process.out.crai[0][1]).name,
+                    file(process.out.unselected[0][1]).name,
+                    file(process.out.unselected_index[0][1]).name,
+                    process.out.versions).match()
+                }
+            )
+        }
+    }
+
     test("cram") {
 
         when {
@@ -54,6 +184,7 @@ nextflow_process {
                     file(params.modules_testdata_base_path + 'genomics/homo_sapiens/genome/genome.fasta', checkIfExists: true)
                 ])
                 input[2] = []
+                input[3] = []
                 """
             }
         }
@@ -89,6 +220,7 @@ nextflow_process {
                     file(params.modules_testdata_base_path + 'genomics/homo_sapiens/genome/genome.fasta', checkIfExists: true)
                 ])
                 input[2] = []
+                input[3] = []
                 """
             }
         }
@@ -124,6 +256,7 @@ nextflow_process {
                     file(params.modules_testdata_base_path + 'genomics/homo_sapiens/genome/genome.fasta', checkIfExists: true)
                 ])
                 input[2] = []
+                input[3] = []
                 """
             }
         }
@@ -159,6 +292,7 @@ nextflow_process {
                     file(params.modules_testdata_base_path + 'genomics/homo_sapiens/genome/genome.fasta', checkIfExists: true)
                 ])
                 input[2] = Channel.of("testN:2817", "testN:2814").collectFile(name: "readnames.list", newLine: true)
+                input[3] = []
                 """
             }
         }
@@ -194,6 +328,7 @@ nextflow_process {
                 ])
                 input[1] = [[],[]]
                 input[2] = []
+                input[3] = []
                 """
             }
         }
@@ -211,4 +346,118 @@ nextflow_process {
             )
         }
     }
+
+    test("bam_csi_index - stub") {
+
+        options "-stub"
+        config "./bam_index.config"
+
+        when {
+            process {
+                """
+                input[0] = Channel.of([
+                    [ id:'test', single_end:false ], // meta map
+                    file(params.modules_testdata_base_path + 'genomics/sarscov2/illumina/bam/test.paired_end.sorted.bam', checkIfExists: true),
+                    []
+                ])
+                input[1] = [[],[]]
+                input[2] = []
+                input[3] = 'csi'
+                """
+            }
+        }
+
+        then {
+            assertAll(
+                { assert process.success },
+                { assert snapshot(process.out).match() }
+            )
+        }
+    }
+
+    test("bam_bai_index - stub") {
+
+        options "-stub"
+        config "./bam_index.config"
+
+        when {
+            process {
+                """
+                input[0] = Channel.of([
+                    [ id:'test', single_end:false ], // meta map
+                    file(params.modules_testdata_base_path + 'genomics/sarscov2/illumina/bam/test.paired_end.sorted.bam', checkIfExists: true),
+                    []
+                ])
+                input[1] = [[],[]]
+                input[2] = []
+                input[3] = 'bai'
+                """
+            }
+        }
+
+        then {
+            assertAll(
+                { assert process.success },
+                { assert snapshot(process.out).match() }
+            )
+        }
+    }
+
+    test("bam_bai_index_uselected - stub") {
+
+        options "-stub"
+        config "./bam_index.config"
+
+        when {
+            process {
+                """
+                input[0] = Channel.of([
+                    [ id:'test', single_end:false ], // meta map
+                    file(params.modules_testdata_base_path + 'genomics/sarscov2/illumina/bam/test.paired_end.sorted.bam', checkIfExists: true),
+                    []
+                ])
+                input[1] = [[],[]]
+                input[2] = Channel.of('testN:1')
+                    .collectFile(name: 'selected_reads.txt')
+                input[3] = 'bai'
+                """
+            }
+        }
+
+        then {
+            assertAll(
+                { assert process.success },
+                { assert snapshot(process.out).match() }
+            )
+        }
+    }
+
+    test("cram_crai_index_unselected - stub") {
+
+        options "-stub"
+        config "./cram_index.config"
+
+        when {
+            process {
+                """
+                input[0] = Channel.of([
+                    [ id:'test', single_end:false ], // meta map
+                    file(params.modules_testdata_base_path + 'genomics/sarscov2/illumina/bam/test.paired_end.sorted.bam', checkIfExists: true),
+                    []
+                ])
+                input[1] = [[],[]]
+                input[2] = Channel.of('testN:1')
+                    .collectFile(name: 'selected_reads.txt')
+                input[3] = 'crai'
+                """
+            }
+        }
+
+        then {
+            assertAll(
+                { assert process.success },
+                { assert snapshot(process.out).match() }
+            )
+        }
+    }
 }