Add stringtie merge module (from nanoseq modules) (#475)

* add stringtie merge module

* add md5sum and path for stringtie.merged.gtf

* fix errors

* try fixing stringtie check error

* add tag

* remove unreproducible md5sum

* address PR suggestions

* hopefully fix linting error

software/stringtie/merge/functions.nf

@@ -0,0 +1,60 @@
+/*
+ * -----------------------------------------------------
+ *  Utility functions used in nf-core DSL2 module files
+ * -----------------------------------------------------
+ */
+
+/*
+ * Extract name of software tool from process name using $task.process
+ */
+def getSoftwareName(task_process) {
+    return task_process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()
+}
+
+/*
+ * Function to initialise default values and to generate a Groovy Map of available options for nf-core modules
+ */
+def initOptions(Map args) {
+    def Map options = [:]
+    options.args          = args.args ?: ''
+    options.args2         = args.args2 ?: ''
+    options.args3         = args.args3 ?: ''
+    options.publish_by_id = args.publish_by_id ?: false
+    options.publish_dir   = args.publish_dir ?: ''
+    options.publish_files = args.publish_files
+    options.suffix        = args.suffix ?: ''
+    return options
+}
+
+/*
+ * Tidy up and join elements of a list to return a path string
+ */
+def getPathFromList(path_list) {
+    def paths = path_list.findAll { item -> !item?.trim().isEmpty() }  // Remove empty entries
+    paths = paths.collect { it.trim().replaceAll("^[/]+|[/]+\$", "") } // Trim whitespace and trailing slashes
+    return paths.join('/')
+}
+
+/*
+ * Function to save/publish module results
+ */
+def saveFiles(Map args) {
+    if (!args.filename.endsWith('.version.txt')) {
+        def ioptions = initOptions(args.options)
+        def path_list = [ ioptions.publish_dir ?: args.publish_dir ]
+        if (ioptions.publish_by_id) {
+            path_list.add(args.publish_id)
+        }
+        if (ioptions.publish_files instanceof Map) {
+            for (ext in ioptions.publish_files) {
+                if (args.filename.endsWith(ext.key)) {
+                    def ext_list = path_list.collect()
+                    ext_list.add(ext.value)
+                    return "${getPathFromList(ext_list)}/$args.filename"
+                }
+            }
+        } else if (ioptions.publish_files == null) {
+            return "${getPathFromList(path_list)}/$args.filename"
+        }
+    }
+}

software/stringtie/merge/main.nf

@@ -0,0 +1,35 @@
+// Import generic module functions
+include { initOptions; saveFiles; getSoftwareName } from './functions'
+
+params.options = [:]
+def options    = initOptions(params.options)
+
+process STRINGTIE_MERGE {
+    label 'process_medium'
+    publishDir "${params.outdir}",
+        mode: params.publish_dir_mode,
+        saveAs: { filename -> saveFiles(filename:filename, options:params.options, publish_dir:getSoftwareName(task.process), meta:meta, publish_by_meta:['']) }
+
+    // Note: 2.7X indices incompatible with AWS iGenomes.
+    conda     (params.enable_conda ? "bioconda::stringtie=2.1.4" : null)
+    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+        container "https://depot.galaxyproject.org/singularity/stringtie:2.1.4--h7e0af3c_0"
+    } else {
+        container "quay.io/biocontainers/stringtie:2.1.4--h7e0af3c_0"
+    }
+
+    input:
+    path  stringtie_gtf
+    path  annotation_gtf
+
+    output:
+    path "stringtie.merged.gtf"   , emit: merged_gtf
+
+    script:
+    """
+    stringtie \\
+        --merge $stringtie_gtf \\
+        -G $annotation_gtf \\
+        -o stringtie.merged.gtf
+    """
+}

software/stringtie/merge/meta.yml

@@ -0,0 +1,31 @@
+name: stringtie_merge
+description: Merges the annotation gtf file and the stringtie output gtf files
+keywords:
+    - merge
+    - gtf
+    - reference
+tools:
+    - stringtie2:
+        description: |
+            Transcript assembly and quantification for RNA-Seq
+        homepage: https://ccb.jhu.edu/software/stringtie/index.shtml
+        documentation: https://ccb.jhu.edu/software/stringtie/index.shtml?t=manual
+input:
+    - stringtie_gtf:
+        type: file
+        description: |
+            Stringtie transcript gtf output(s).
+        pattern: "*.gtf"
+    - annotation_gtf:
+        type: file
+        description: |
+            Annotation gtf file.
+        pattern: "*.gtf"
+output:
+    - merged_gtf:
+        type: map
+        description: |
+            Merged gtf from annotation and stringtie output gtfs.
+        pattern: "*.gtf"
+authors:
+    - "@yuukiiwa"

software/stringtie/stringtie/meta.yml

@@ -0,0 +1,56 @@
+name: stringtie
+description: Transcript assembly and quantification for RNA-Se
+keywords:
+    - transcript
+    - assembly
+    - quantification
+    - gtf
+
+tools:
+    - stringtie2:
+        description: |
+            Transcript assembly and quantification for RNA-Seq
+        homepage: https://ccb.jhu.edu/software/stringtie/index.shtml
+        documentation: https://ccb.jhu.edu/software/stringtie/index.shtml?t=manual
+input:
+    - meta:
+        type: map
+        description: |
+            Groovy Map containing sample information
+            e.g. [ id:'test', single_end:false ]
+    - bam:
+        type: file
+        description: |
+            Stringtie transcript gtf output(s).
+    - gtf:
+        type: file
+        description: |
+            Annotation gtf file.
+output:
+    - meta:
+        type: map
+        description: |
+            Groovy Map containing sample information
+            e.g. [ id:'test', single_end:false ]
+    - transcript_gtf:
+        type: file
+        description: transcript gtf
+        pattern: "*.{transcripts.gtf}"
+    - coverage_gtf:
+        type: file
+        description: coverage gtf
+        pattern: "*.{coverage.gtf}"
+    - abudance:
+        type: file
+        description: abudance
+        pattern: "*.{abudance.txt}"
+    - ballgown:
+        type: file
+        description: for running ballgown
+        pattern: "*.{ballgown}"
+    - version:
+        type: file
+        description: File containing software version
+        pattern: "*.{version.txt}"
+authors:
+    - "@drpatelh"

tests/config/pytest_software.yml

@@ -579,9 +579,13 @@ strelka/germline:
   - software/strelka/germline/**
   - tests/software/strelka/germline/**
 
-stringtie:
-  - software/stringtie/**
-  - tests/software/stringtie/**
+stringtie/merge:
+  - software/stringtie/merge/**
+  - tests/software/stringtie/merge/**
+
+stringtie/stringtie:
+  - software/stringtie/stringtie/**
+  - tests/software/stringtie/stringtie/**
 
 subread/featurecounts:
   - software/subread/featurecounts/**

tests/software/stringtie/main.nf

@@ -1,29 +0,0 @@
-#!/usr/bin/env nextflow
-
-nextflow.enable.dsl = 2
-
-include { STRINGTIE as STRINGTIE_FORWARD } from '../../../software/stringtie/main.nf'  addParams( options: [ publish_dir:'test_stringtie_forward' ] )
-include { STRINGTIE as STRINGTIE_REVERSE } from '../../../software/stringtie/main.nf'  addParams( options: [ publish_dir:'test_stringtie_reverse' ] )
-
-/*
- * Test with forward strandedness
- */
-workflow test_stringtie_forward {
-    input = [ [ id:'test', strandedness:'forward' ], // meta map
-              [ file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true) ] ]
-    gtf   = file(params.test_data['sarscov2']['genome']['genome_gtf'], checkIfExists: true)
-    
-    STRINGTIE_FORWARD ( input, gtf )
-}
-
-/*
- * Test with reverse strandedness
- */
-workflow test_stringtie_reverse {
-    input = [ [ id:'test', strandedness:'reverse' ], // meta map
-              [ file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true) ] 
-            ]
-    gtf   = file(params.test_data['sarscov2']['genome']['genome_gtf'], checkIfExists: true)
-    
-    STRINGTIE_REVERSE ( input, gtf )
-}

tests/software/stringtie/merge/main.nf

@@ -0,0 +1,38 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { STRINGTIE as STRINGTIE_FORWARD            } from '../../../../software/stringtie/stringtie/main.nf'  addParams( options: [ publish_dir:'test_stringtie_forward' ] )
+include { STRINGTIE as STRINGTIE_REVERSE            } from '../../../../software/stringtie/stringtie/main.nf'  addParams( options: [ publish_dir:'test_stringtie_reverse' ] )
+include { STRINGTIE_MERGE as STRINGTIE_FORWARD_MERGE} from '../../../../software/stringtie/merge/main.nf'      addParams( options: [ publish_dir:'test_stringtie_forward_merge'] )
+include { STRINGTIE_MERGE as STRINGTIE_REVERSE_MERGE} from '../../../../software/stringtie/merge/main.nf'      addParams( options: [ publish_dir:'test_stringtie_reverse_merge'] )
+/*
+ * Test with forward strandedness
+ */
+workflow test_stringtie_forward_merge {
+    input = [ [ id:'test', strandedness:'forward' ], // meta map
+              [ file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true) ] ]
+    annotation_gtf   = file(params.test_data['homo_sapiens']['genome']['genome_gtf'], checkIfExists: true)
+    
+    STRINGTIE_FORWARD ( input, annotation_gtf )
+    STRINGTIE_FORWARD.out.transcript_gtf
+       .map { it -> it[1] }
+       .set { stringtie_gtf }
+    STRINGTIE_FORWARD_MERGE ( stringtie_gtf, annotation_gtf )
+}
+
+/*
+ * Test with reverse strandedness
+ */
+workflow test_stringtie_reverse_merge {
+    input = [ [ id:'test', strandedness:'reverse' ], // meta map
+              [ file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true) ] 
+            ]
+    annotation_gtf   = file(params.test_data['homo_sapiens']['genome']['genome_gtf'], checkIfExists: true)
+    
+    STRINGTIE_REVERSE ( input, annotation_gtf )
+    STRINGTIE_REVERSE.out.transcript_gtf
+       .map { it -> it[1] }
+       .set { stringtie_gtf }
+    STRINGTIE_REVERSE_MERGE ( stringtie_gtf, annotation_gtf )
+}

tests/software/stringtie/merge/test.yml

@@ -0,0 +1,17 @@
+- name: stringtie forward-strand merge
+  command: nextflow run ./tests/software/stringtie/merge/ -entry test_stringtie_forward_merge -c tests/config/nextflow.config
+  tags:
+    - stringtie
+    - stringtie/merge
+  files:
+    - path: ./output/test_stringtie_forward_merge/stringtie.merged.gtf
+      md5sum: 676aa20a2d7a3db18136cdc7ba183099
+
+- name: stringtie reverse-strand merge
+  command: nextflow run ./tests/software/stringtie/merge/ -entry test_stringtie_reverse_merge -c tests/config/nextflow.config
+  tags:
+    - stringtie
+    - stringtie/merge
+  files:
+    - path: ./output/test_stringtie_reverse_merge/stringtie.merged.gtf
+      md5sum: 67e5102722ecaeea1fb44d1ec0953474      

tests/software/stringtie/stringtie/main.nf

@@ -0,0 +1,28 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { STRINGTIE as STRINGTIE_FORWARD            } from '../../../../software/stringtie/stringtie/main.nf'  addParams( options: [ publish_dir:'test_stringtie_forward' ] )
+include { STRINGTIE as STRINGTIE_REVERSE            } from '../../../../software/stringtie/stringtie/main.nf'  addParams( options: [ publish_dir:'test_stringtie_reverse' ] )
+
+/*
+ * Test with forward strandedness
+ */
+workflow test_stringtie_forward {
+    input = [ [ id:'test', strandedness:'forward' ], // meta map
+              [ file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true) ] ]
+    annotation_gtf   = file(params.test_data['sarscov2']['genome']['genome_gtf'], checkIfExists: true)
+    
+    STRINGTIE_FORWARD ( input, annotation_gtf )
+}
+
+/*
+ * Test with reverse strandedness
+ */
+workflow test_stringtie_reverse {
+    input = [ [ id:'test', strandedness:'reverse' ], // meta map
+              [ file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true) ] ]
+    annotation_gtf   = file(params.test_data['sarscov2']['genome']['genome_gtf'], checkIfExists: true)
+    
+    STRINGTIE_REVERSE ( input, annotation_gtf )
+}

tests/software/stringtie/stringtie/test.yml

@@ -1,10 +1,10 @@
-- name: stringtie forward-strand
-  command: nextflow run ./tests/software/stringtie/ -entry test_stringtie_forward -c tests/config/nextflow.config
+- name: test_stringtie_forward
+  command: nextflow run ./tests/software/stringtie/stringtie/ -entry test_stringtie_forward -c tests/config/nextflow.config
   tags:
     - stringtie
+    - stringtie/stringtie
   files:
     - path: ./output/test_stringtie_forward/test.transcripts.gtf
-      md5sum: 8abb6a87581bd7a61af74085b5a6daec
     - path: ./output/test_stringtie_forward/test.gene.abundance.txt
       md5sum: 7d8bce7f2a922e367cedccae7267c22e
     - path: ./output/test_stringtie_forward/test.coverage.gtf
@@ -20,13 +20,13 @@
     - path: ./output/test_stringtie_forward/test.ballgown/e2t.ctab
       md5sum: e981c0038295ae54b63cedb1083f1540
 
-- name: stringtie reverse-strand
-  command: nextflow run ./tests/software/stringtie/ -entry test_stringtie_reverse -c tests/config/nextflow.config
+- name: test_stringtie_reverse
+  command: nextflow run ./tests/software/stringtie/stringtie/ -entry test_stringtie_reverse -c tests/config/nextflow.config
   tags:
     - stringtie
+    - stringtie/stringtie
   files:
     - path: ./output/test_stringtie_reverse/test.transcripts.gtf
-      md5sum: 8bdcda06f39183fa4312644bd0e41bb5
     - path: ./output/test_stringtie_reverse/test.gene.abundance.txt
       md5sum: 7385b870b955dae2c2ab78a70cf05cce
     - path: ./output/test_stringtie_reverse/test.coverage.gtf