add ctatsplicing/prepgenomelib and ctatsplicing/startocancerintrons

.github/skip_nf_test.json

@@ -28,6 +28,8 @@
         "modules/nf-core/cellrangeratac/count",
         "modules/nf-core/checkqc",
         "modules/nf-core/coreograph",
+        "modules/nf-core/ctatsplicing/prepgenomelib",
+        "modules/nf-core/ctatsplicing/startocancerintrons",
         "modules/nf-core/custom/dumpsoftwareversions",
         "modules/nf-core/deepcell/mesmer",
         "modules/nf-core/deepsomatic",

modules/nf-core/ctatsplicing/prepgenomelib/main.nf

@@ -0,0 +1,129 @@
+process CTATSPLICING_PREPGENOMELIB {
+    tag "$meta.id"
+    label 'process_single'
+    stageInMode 'copy'
+
+    container "nf-core/ctatsplicing:0.0.3"
+
+    input:
+    tuple val(meta), path(genome_lib)
+    path(cancer_intron_tsv)
+
+    output:
+    tuple val(meta), path(genome_lib), emit: reference
+    path "versions.yml"              , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    // Exit if running this module with -profile conda / -profile mamba
+    if (workflow.profile.tokenize(',').intersect(['conda', 'mamba']).size() >= 1) {
+        error "CTATSPLICING_PREPGENOMELIB module does not support Conda. Please use Docker / Singularity / Podman instead."
+    }
+    def args = task.ext.args ?: ''
+    def prefix = task.ext.prefix ?: "${meta.id}"
+    def VERSION = '0.0.2' // WARN: Version information not provided by tool on CLI. Please update this string when bumping container versions.
+    """
+    /usr/local/src/CTAT-SPLICING/prep_genome_lib/ctat-splicing-lib-integration.py \\
+        --cancer_introns_tsv $cancer_intron_tsv \\
+        --genome_lib_dir $genome_lib
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        ctatsplicing: $VERSION
+    END_VERSIONS
+    """
+
+    stub:
+    def args = task.ext.args ?: ''
+    def prefix = task.ext.prefix ?: "${meta.id}"
+    def VERSION = '0.0.2' // WARN: Version information not provided by tool on CLI. Please update this string when bumping container versions.
+    """
+    mkdir -p $genome_lib/
+    mkdir -p $genome_lib/ref_genome.fa.star.idx
+    mkdir -p $genome_lib/__chkpts
+    mkdir -p $genome_lib/cancer_splicing_lib
+    touch $genome_lib/ref_genome.fa.star.idx/chrName.txt
+    touch $genome_lib/ref_genome.fa.star.idx/exonInfo.tab
+    touch $genome_lib/ref_genome.fa.star.idx/SAindex
+    touch $genome_lib/ref_genome.fa.star.idx/Genome
+    touch $genome_lib/ref_genome.fa.star.idx/geneInfo.tab
+    touch $genome_lib/ref_genome.fa.star.idx/sjdbList.out.tab
+    touch $genome_lib/ref_genome.fa.star.idx/sjdbInfo.txt
+    touch $genome_lib/ref_genome.fa.star.idx/transcriptInfo.tab
+    touch $genome_lib/ref_genome.fa.star.idx/sjdbList.fromGTF.out.tab
+    touch $genome_lib/ref_genome.fa.star.idx/build.ok
+    touch $genome_lib/ref_genome.fa.star.idx/chrLength.txt
+    touch $genome_lib/ref_genome.fa.star.idx/Log.out
+    touch $genome_lib/ref_genome.fa.star.idx/genomeParameters.txt
+    touch $genome_lib/ref_genome.fa.star.idx/chrStart.txt
+    touch $genome_lib/ref_genome.fa.star.idx/SA
+    touch $genome_lib/ref_genome.fa.star.idx/chrNameLength.txt
+    touch $genome_lib/ref_genome.fa.star.idx/exonGeTrInfo.tab
+    touch $genome_lib/ref_annot.cds
+    touch $genome_lib/ref_annot.gtf.gene_spans
+    touch $genome_lib/pfam_domains.dbm
+    touch $genome_lib/ref_annot.cdsplus.fa.idx
+    touch $genome_lib/ref_genome.fa.ndb
+    touch $genome_lib/ref_genome.fa.nin
+    touch $genome_lib/ref_genome.fa
+    touch $genome_lib/ref_annot.prot_info.dbm
+    echo | gzip > $genome_lib/PFAM.domtblout.dat.gz
+    touch $genome_lib/trans.blast.align_coords.align_coords.dat
+    touch $genome_lib/fusion_annot_lib.idx
+    touch $genome_lib/ref_annot.pep
+    touch $genome_lib/__chkpts/trans.blast.dat.cp.ok
+    touch $genome_lib/__chkpts/cp_ref_annot_cdna.ok
+    touch $genome_lib/__chkpts/ref_annot.cdsplus.dfam_masked.fa.cp.ok
+    touch $genome_lib/__chkpts/annotfiltrule_cp.ok
+    touch $genome_lib/__chkpts/validate_ctat_genome_lib.ok
+    touch $genome_lib/__chkpts/cp_pfam_dat.ok
+    touch $genome_lib/__chkpts/fusion_annot_lib.cp.ok
+    touch $genome_lib/__chkpts/ref_genome.fa.ok
+    touch $genome_lib/__chkpts/cp_gene_blast_pairs.ok
+    touch $genome_lib/__chkpts/mm2_genome_idx.ok
+    touch $genome_lib/__chkpts/trans.blast.dat.index.ok
+    touch $genome_lib/__chkpts/ref_genome_fai.ok
+    touch $genome_lib/__chkpts/index_ref_annot_cdna.ok
+    touch $genome_lib/__chkpts/ref_annot.gtf.ok
+    touch $genome_lib/__chkpts/blast_pairs.idx.ok
+    touch $genome_lib/__chkpts/ref_annot.cdsplus.dfam_masked.fa.idx.ok
+    touch $genome_lib/__chkpts/_prot_info_db.ok
+    touch $genome_lib/__chkpts/_fusion_annot_lib.idx.ok
+    touch $genome_lib/__chkpts/index_pfam_hits.ok
+    touch $genome_lib/__chkpts/makeblastdb.ok
+    touch $genome_lib/__chkpts/ref_annot.gtf.gene_spans.ok
+    touch $genome_lib/__chkpts/mm2.splice_bed.ok
+    touch $genome_lib/__chkpts/ref_annot.gtf.mini.sortu.ok
+    touch $genome_lib/ref_genome.fa.ntf
+    touch $genome_lib/ref_genome.fa.nto
+    touch $genome_lib/ref_annot.gtf
+    touch $genome_lib/ref_annot.gtf.mm2.splice.bed
+    touch $genome_lib/ref_genome.fa.mm2
+    echo | gzip > $genome_lib/fusion_annot_lib.gz
+    touch $genome_lib/ref_annot.cdsplus.fa
+    touch $genome_lib/AnnotFilterRule.pm
+    echo | gzip > $genome_lib/trans.blast.dat.gz
+    touch $genome_lib/ref_genome.fa.nhr
+    touch $genome_lib/blast_pairs.idx
+    touch $genome_lib/ref_genome.fa.nsq
+    echo | gzip > $genome_lib/blast_pairs.dat.gz
+    touch $genome_lib/ref_annot.cdna.fa.idx
+    touch $genome_lib/ref_genome.fa.not
+    touch $genome_lib/ref_annot.cdna.fa
+    touch $genome_lib/ref_annot.gtf.mini.sortu
+    touch $genome_lib/trans.blast.align_coords.align_coords.dbm
+    touch $genome_lib/ref_genome.fa.njs
+    touch $genome_lib/ref_genome.fa.fai
+    touch $genome_lib/refGene.bed
+    echo | gzip > $genome_lib/refGene.sort.bed.gz
+    echo | gzip > $genome_lib/refGene.sort.bed.gz.tbi
+    touch $genome_lib/cancer_splicing_lib/cancer_splicing.idx
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        ctatsplicing: $VERSION
+    END_VERSIONS
+    """
+}

modules/nf-core/ctatsplicing/prepgenomelib/meta.yml

@@ -0,0 +1,57 @@
+---
+# yaml-language-server: $schema=https://raw.githubusercontent.com/nf-core/modules/master/modules/meta-schema.json
+name: "ctatsplicing_prepgenomelib"
+description: Reference preparation for CTAT-splicing
+long_description: |
+  This module prepares the CTAT genome library reference for splicing analysis by integrating cancer intron annotations.
+  It takes a CTAT genome library and a cancer intron TSV file as input, and outputs a modified genome library reference.
+  The output can be used in subsequent CTAT-splicing analyses to detect and annotate cancer splicing aberrations.
+  Make sure to provide the correct genome library and cancer intron data resource.
+keywords:
+  - splicing
+  - cancer
+  - rna
+  - rnaseq
+tools:
+  - "ctatsplicing":
+      description: "Detection and annotation of cancer splicing aberrations"
+      homepage: "https://github.com/TrinityCTAT/CTAT-SPLICING"
+      documentation: "https://github.com/TrinityCTAT/CTAT-SPLICING"
+      tool_dev_url: "https://github.com/TrinityCTAT/CTAT-SPLICING"
+
+input:
+  - - meta:
+        type: map
+        description: |
+          Groovy Map containing sample information
+          e.g. `[ id:'sample1', single_end:false ]`
+    - genome_lib:
+        type: file
+        description: CTAT genome library reference
+        pattern: "*"
+  - cancer_intron_tsv:
+      type: file
+      description: CTAT-splicing data resource supplement
+
+output:
+  reference:
+    - - meta:
+          type: map
+          description: |
+            Groovy Map containing sample information
+            e.g. `[ id:'sample1', single_end:false ]`
+      - genome_lib:
+          type: file
+          description: Modified CTAT genome library reference
+          pattern: "*"
+  versions:
+    - "versions.yml":
+        type: file
+        description: File containing software versions
+        pattern: "versions.yml"
+
+authors:
+  - "@anoronh4"
+  - "@nvnieuwk"
+maintainers:
+  - "@anoronh4"

modules/nf-core/ctatsplicing/prepgenomelib/tests/main.nf.test

@@ -0,0 +1,109 @@
+nextflow_process {
+
+    name "Test Process CTATSPLICING_PREPGENOMELIB"
+    script "../main.nf"
+    process "CTATSPLICING_PREPGENOMELIB"
+
+    tag "modules"
+    tag "modules_nfcore"
+    tag "ctatsplicing"
+    tag "ctatsplicing/prepgenomelib"
+    tag "starfusion/build"
+
+    test("homo_sapiens - reference") {
+        setup {
+            run("STARFUSION_BUILD") {
+                script "../../../starfusion/build/main.nf"
+                process {
+                    """
+                    input[0] = [
+                        [ id:'minigenome_fasta' ],
+                        file(params.modules_testdata_base_path + 'genomics/homo_sapiens/genome/minigenome.fa')
+                    ]
+                    input[1] = [
+                        [ id:'minigenome_gtf' ],
+                        file(params.modules_testdata_base_path + 'genomics/homo_sapiens/genome/minigenome.gtf')
+                    ]
+                    input[2] = file(params.modules_testdata_base_path + 'genomics/homo_sapiens/genome/CTAT_HumanFusionLib.mini.dat.gz')
+                    input[3] = "homo_sapiens"
+                    input[4] = "http://ftp.ebi.ac.uk/pub/databases/Pfam/releases/Pfam37.4/Pfam-A.hmm.gz"
+                    input[5] = [
+                        "https://www.dfam.org/releases/Dfam_3.8/infrastructure/dfamscan/homo_sapiens_dfam.hmm",
+                        "https://www.dfam.org/releases/Dfam_3.8/infrastructure/dfamscan/homo_sapiens_dfam.hmm.h3f",
+                        "https://www.dfam.org/releases/Dfam_3.8/infrastructure/dfamscan/homo_sapiens_dfam.hmm.h3i",
+                        "https://www.dfam.org/releases/Dfam_3.8/infrastructure/dfamscan/homo_sapiens_dfam.hmm.h3m",
+                        "https://www.dfam.org/releases/Dfam_3.8/infrastructure/dfamscan/homo_sapiens_dfam.hmm.h3p"
+                    ]
+                    input[6] = "https://data.broadinstitute.org/Trinity/CTAT_RESOURCE_LIB/AnnotFilterRule.pm"
+                    """
+                }
+            }
+        }
+
+        when {
+            process {
+                """
+                input[0] = STARFUSION_BUILD.out.reference
+                input[1] = [
+                    file("https://data.broadinstitute.org/Trinity/CTAT_RESOURCE_LIB/CANCER_SPLICING_LIB_SUPPLEMENT/cancer_introns.GRCh38.Jun232020.tsv.gz", checkIfExists:true)
+                ]
+                """
+            }
+        }
+
+        then {
+            def stable_name    = getAllFilesFromDir(process.out.reference.get(0).get(1), includeDir: true)
+            def stable_files = getAllFilesFromDir(process.out.reference.get(0).get(1), includeDir: false, include: ['*','**/*'],
+                ignore: ['refGene.sort.bed.gz.tbi','**/refGene.sort.bed.gz.tbi', 'refGene.bed', '**/cancer_splicing.idx', 'ref_annot.c*',
+                    'PFAM.domtblout.dat.gz', 'blast_pairs*', 'fusion_annot_lib.idx', 'pfam_domains.dbm', 'ref_annot.gtf.gene_spans',
+                    'ref_annot.prot_info.dbm', 'ref_annot.pep', 'ref_genome.fa.nin', 'ref_genome.fa.njs', '**/Log.out', '**/genomeParameters.txt', 'trans.blast.align_coords.align_coords.dbm'])
+            def stable_content = stable_files.findAll { it.size() > 0 }
+            assertAll(
+                { assert process.success },
+                {
+                    assert snapshot(
+                        stable_name*.name,
+                        stable_content,
+                        process.out.versions
+                    ).match()
+                }
+            )
+
+        }
+    }
+
+    test("homo_sapiens - reference - stub") {
+
+        options "-stub"
+
+        when {
+            process {
+                """
+                new File('ctat_genome_lib_build_dir').mkdirs()
+                input[0] = [[id:"ctat_genome_lib_build_dir"],file("ctat_genome_lib_build_dir/")]
+                input[1] = [
+                    file("https://data.broadinstitute.org/Trinity/CTAT_RESOURCE_LIB/CANCER_SPLICING_LIB_SUPPLEMENT/cancer_introns.GRCh38.Jun232020.tsv.gz", checkIfExists:true)
+                ]
+                """
+            }
+        }
+
+        then {
+            def stable_name    = getAllFilesFromDir(process.out.reference.get(0).get(1), includeDir: true)
+            def stable_content = getAllFilesFromDir(process.out.reference.get(0).get(1), includeDir: false, include: ['*','**/*'], ignore: ['refGene.sort.bed.gz.tbi','**/refGene.sort.bed.gz.tbi','**/cancer_splicing.idx'])
+            assertAll(
+                { assert process.success },
+                {
+                    assert snapshot(
+                        stable_name*.name,
+                        stable_content,
+                        process.out.versions
+                    ).match()
+                }
+            )
+
+        }
+
+    }
+
+}