Add nf-test to STARFUSION_BUILD and refactor module

conf/modules.config

@@ -329,6 +329,7 @@ process {
             mode: params.publish_dir_mode,
             saveAs: { filename -> filename.equals('versions.yml') ? null : filename },
         ]
+        ext.args = "--max_readlength ${params.read_length}"
     }
 
     withName: 'STARFUSION_DOWNLOAD' {

conf/test_build.config

@@ -21,4 +21,7 @@ params {
     all                        = true
 
     skip_salmon_index          = true
+    starfusion_build           = true
+    fusion_annot_lib           = 'https://github.com/STAR-Fusion/STAR-Fusion-Tutorial/raw/master/CTAT_HumanFusionLib.mini.dat.gz'
+
 }

modules/local/starfusion/build/main.nf

@@ -1,5 +1,6 @@
 process STARFUSION_BUILD {
-    tag 'star-fusion'
+    tag "$meta.id"
+    label 'process_high'
 
     conda "${moduleDir}/environment.yml"
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
@@ -9,30 +10,22 @@ process STARFUSION_BUILD {
     input:
     tuple val(meta), path(fasta)
     tuple val(meta2), path(gtf)
+    path fusion_annot_lib
+    val dfam_species
 
     output:
-    path "*"  , emit: reference
+    tuple val(meta), path("ctat_genome_lib_build_dir"), emit: reference
 
     script:
+    def args = task.ext.args ?: ''
     """
-    export TMPDIR=/tmp
-    wget http://ftp.ebi.ac.uk/pub/databases/Pfam/releases/Pfam34.0/Pfam-A.hmm.gz --no-check-certificate
-    wget https://github.com/FusionAnnotator/CTAT_HumanFusionLib/releases/download/v0.3.0/fusion_lib.Mar2021.dat.gz -O CTAT_HumanFusionLib_Mar2021.dat.gz --no-check-certificate
-    wget https://data.broadinstitute.org/Trinity/CTAT_RESOURCE_LIB/AnnotFilterRule.pm -O AnnotFilterRule.pm --no-check-certificate
-    wget https://www.dfam.org/releases/Dfam_3.4/infrastructure/dfamscan/homo_sapiens_dfam.hmm --no-check-certificate
-    wget https://www.dfam.org/releases/Dfam_3.4/infrastructure/dfamscan/homo_sapiens_dfam.hmm.h3f --no-check-certificate
-    wget https://www.dfam.org/releases/Dfam_3.4/infrastructure/dfamscan/homo_sapiens_dfam.hmm.h3i --no-check-certificate
-    wget https://www.dfam.org/releases/Dfam_3.4/infrastructure/dfamscan/homo_sapiens_dfam.hmm.h3m --no-check-certificate
-    wget https://www.dfam.org/releases/Dfam_3.4/infrastructure/dfamscan/homo_sapiens_dfam.hmm.h3p --no-check-certificate
-    gunzip Pfam-A.hmm.gz && hmmpress Pfam-A.hmm
     prep_genome_lib.pl \\
         --genome_fa $fasta \\
         --gtf $gtf \\
-        --annot_filter_rule AnnotFilterRule.pm \\
-        --fusion_annot_lib CTAT_HumanFusionLib_Mar2021.dat.gz \\
-        --pfam_db Pfam-A.hmm \\
-        --dfam_db homo_sapiens_dfam.hmm \\
-        --max_readlength $params.read_length \\
+        --dfam_db ${dfam_species} \\
+        --pfam_db current \\
+        --fusion_annot_lib $fusion_annot_lib \\
+        ${args} \\
         --CPU $task.cpus
 
     cat <<-END_VERSIONS > versions.yml
@@ -43,8 +36,88 @@ process STARFUSION_BUILD {
 
     stub:
     """
-    mkdir ctat_genome_lib_build_dir
-    touch ref_annot.cdna.fa
+    mkdir -p ctat_genome_lib_build_dir
+
+    touch ctat_genome_lib_build_dir/AnnotFilterRule.pm
+    gzip -c /dev/null > ctat_genome_lib_build_dir/blast_pairs.dat.gz
+    touch ctat_genome_lib_build_dir/blast_pairs.idx
+
+    mkdir -p ctat_genome_lib_build_dir/__chkpts
+    touch ctat_genome_lib_build_dir/__chkpts/annotfiltrule_cp.ok
+    touch ctat_genome_lib_build_dir/__chkpts/blast_pairs.idx.ok
+    touch ctat_genome_lib_build_dir/__chkpts/cp_gene_blast_pairs.ok
+    touch ctat_genome_lib_build_dir/__chkpts/cp_pfam_dat.ok
+    touch ctat_genome_lib_build_dir/__chkpts/cp_ref_annot_cdna.ok
+    touch ctat_genome_lib_build_dir/__chkpts/fusion_annot_lib.cp.ok
+    touch ctat_genome_lib_build_dir/__chkpts/_fusion_annot_lib.idx.ok
+    touch ctat_genome_lib_build_dir/__chkpts/index_pfam_hits.ok
+    touch ctat_genome_lib_build_dir/__chkpts/index_ref_annot_cdna.ok
+    touch ctat_genome_lib_build_dir/__chkpts/makeblastdb.ok
+    touch ctat_genome_lib_build_dir/__chkpts/mm2_genome_idx.ok
+    touch ctat_genome_lib_build_dir/__chkpts/mm2.splice_bed.ok
+    touch ctat_genome_lib_build_dir/__chkpts/_prot_info_db.ok
+    touch ctat_genome_lib_build_dir/__chkpts/ref_annot.cdsplus.dfam_masked.fa.cp.ok
+    touch ctat_genome_lib_build_dir/__chkpts/ref_annot.cdsplus.dfam_masked.fa.idx.ok
+    touch ctat_genome_lib_build_dir/__chkpts/ref_annot.gtf.gene_spans.ok
+    touch ctat_genome_lib_build_dir/__chkpts/ref_annot.gtf.mini.sortu.ok
+    touch ctat_genome_lib_build_dir/__chkpts/ref_annot.gtf.ok
+    touch ctat_genome_lib_build_dir/__chkpts/ref_genome_fai.ok
+    touch ctat_genome_lib_build_dir/__chkpts/ref_genome.fa.ok
+    touch ctat_genome_lib_build_dir/__chkpts/trans.blast.dat.cp.ok
+    touch ctat_genome_lib_build_dir/__chkpts/trans.blast.dat.index.ok
+    touch ctat_genome_lib_build_dir/__chkpts/validate_ctat_genome_lib.ok
+
+    gzip -c /dev/null > ctat_genome_lib_build_dir/fusion_annot_lib.gz
+    touch ctat_genome_lib_build_dir/fusion_annot_lib.idx
+    touch ctat_genome_lib_build_dir/pfam_domains.dbm
+    gzip -c /dev/null > ctat_genome_lib_build_dir/PFAM.domtblout.dat.gz
+
+    touch ctat_genome_lib_build_dir/ref_annot.cdna.fa
+    touch ctat_genome_lib_build_dir/ref_annot.cdna.fa.idx
+    touch ctat_genome_lib_build_dir/ref_annot.cds
+    touch ctat_genome_lib_build_dir/ref_annot.cdsplus.fa
+    touch ctat_genome_lib_build_dir/ref_annot.cdsplus.fa.idx
+    touch ctat_genome_lib_build_dir/ref_annot.gtf
+    touch ctat_genome_lib_build_dir/ref_annot.gtf.gene_spans
+    touch ctat_genome_lib_build_dir/ref_annot.gtf.mini.sortu
+    touch ctat_genome_lib_build_dir/ref_annot.gtf.mm2.splice.bed
+    touch ctat_genome_lib_build_dir/ref_annot.pep
+    touch ctat_genome_lib_build_dir/ref_annot.prot_info.dbm
+
+    touch ctat_genome_lib_build_dir/ref_genome.fa
+    touch ctat_genome_lib_build_dir/ref_genome.fa.fai
+    touch ctat_genome_lib_build_dir/ref_genome.fa.mm2
+    touch ctat_genome_lib_build_dir/ref_genome.fa.ndb
+    touch ctat_genome_lib_build_dir/ref_genome.fa.nhr
+    touch ctat_genome_lib_build_dir/ref_genome.fa.nin
+    touch ctat_genome_lib_build_dir/ref_genome.fa.njs
+    touch ctat_genome_lib_build_dir/ref_genome.fa.not
+    touch ctat_genome_lib_build_dir/ref_genome.fa.nsq
+    touch ctat_genome_lib_build_dir/ref_genome.fa.ntf
+    touch ctat_genome_lib_build_dir/ref_genome.fa.nto
+
+    mkdir -p ctat_genome_lib_build_dir/ref_genome.fa.star.idx
+    touch ctat_genome_lib_build_dir/ref_genome.fa.star.idx/build.ok
+    touch ctat_genome_lib_build_dir/ref_genome.fa.star.idx/chrLength.txt
+    touch ctat_genome_lib_build_dir/ref_genome.fa.star.idx/chrNameLength.txt
+    touch ctat_genome_lib_build_dir/ref_genome.fa.star.idx/chrName.txt
+    touch ctat_genome_lib_build_dir/ref_genome.fa.star.idx/chrStart.txt
+    touch ctat_genome_lib_build_dir/ref_genome.fa.star.idx/exonGeTrInfo.tab
+    touch ctat_genome_lib_build_dir/ref_genome.fa.star.idx/exonInfo.tab
+    touch ctat_genome_lib_build_dir/ref_genome.fa.star.idx/geneInfo.tab
+    touch ctat_genome_lib_build_dir/ref_genome.fa.star.idx/Genome
+    touch ctat_genome_lib_build_dir/ref_genome.fa.star.idx/genomeParameters.txt
+    touch ctat_genome_lib_build_dir/ref_genome.fa.star.idx/Log.out
+    touch ctat_genome_lib_build_dir/ref_genome.fa.star.idx/SA
+    touch ctat_genome_lib_build_dir/ref_genome.fa.star.idx/SAindex
+    touch ctat_genome_lib_build_dir/ref_genome.fa.star.idx/sjdbInfo.txt
+    touch ctat_genome_lib_build_dir/ref_genome.fa.star.idx/sjdbList.fromGTF.out.tab
+    touch ctat_genome_lib_build_dir/ref_genome.fa.star.idx/sjdbList.out.tab
+    touch ctat_genome_lib_build_dir/ref_genome.fa.star.idx/transcriptInfo.tab
+
+    touch ctat_genome_lib_build_dir/trans.blast.align_coords.align_coords.dat
+    touch ctat_genome_lib_build_dir/trans.blast.align_coords.align_coords.dbm
+    gzip -c /dev/null > ctat_genome_lib_build_dir/trans.blast.dat.gz
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":

modules/local/starfusion/build/meta.yml

@@ -1,7 +1,7 @@
-name: starfusion_downloadgenome
+name: starfusion_build
 description: Download STAR-fusion genome resource required to run STAR-Fusion caller
 keywords:
-  - downoad
+  - download
 tools:
   - star-fusion:
       description: Fusion calling algorithm for RNAseq data
@@ -20,6 +20,13 @@ input:
       type: file
       description: genome gtf file
       pattern: "*.{gtf}"
+  - fusion_annot_lib:
+      type: file
+      description: Fusion annotation library (key/val pairs, tab-delimited).
+      pattern: "*.dat.gz"
+  - dfam_species:
+      type: string
+      description: DNA transposable element database (Dfam.hmm), required for repeat masking. Only 'human' or 'mouse' are accepted (will automatically pull the resources from dfam).
 
 output:
   - reference:

modules/local/starfusion/build/tests/main.nf.test

@@ -0,0 +1,138 @@
+nextflow_process {
+
+    name "Test Process STARFUSION_BUILD"
+    script "../main.nf"
+    process "STARFUSION_BUILD"
+
+    test("STARFUSION_BUILD - human - minigenome") {
+
+        when {
+            process {
+                """
+                input[0] = [
+                    [ id:'minigenome fasta' ],
+                    file("https://github.com/STAR-Fusion/STAR-Fusion-Tutorial/raw/master/minigenome.fa")
+                ]
+                input[1] = [
+                    [ id:'minigenome gtf' ],
+                    file("https://github.com/STAR-Fusion/STAR-Fusion-Tutorial/raw/master/minigenome.gtf")
+                ]
+
+                input [2] = file("https://github.com/STAR-Fusion/STAR-Fusion-Tutorial/raw/master/CTAT_HumanFusionLib.mini.dat.gz")
+                input [3] = "human"
+
+                """
+            }
+        }
+
+        then {
+            assert snapshot(
+                    path(process.out.reference[0][1]).resolve("AnnotFilterRule.pm"),
+                    path(process.out.reference[0][1]).resolve("blast_pairs.dat.gz").exists(),
+                    path(process.out.reference[0][1]).resolve("blast_pairs.idx").exists(),
+                    path(process.out.reference[0][1]).resolve("__chkpts/annotfiltrule_cp.ok"),
+                    path(process.out.reference[0][1]).resolve("__chkpts/blast_pairs.idx.ok"),
+                    path(process.out.reference[0][1]).resolve("__chkpts/cp_gene_blast_pairs.ok"),
+                    path(process.out.reference[0][1]).resolve("__chkpts/cp_pfam_dat.ok"),
+                    path(process.out.reference[0][1]).resolve("__chkpts/cp_ref_annot_cdna.ok"),
+                    path(process.out.reference[0][1]).resolve("__chkpts/fusion_annot_lib.cp.ok"),
+                    path(process.out.reference[0][1]).resolve("__chkpts/_fusion_annot_lib.idx.ok"),
+                    path(process.out.reference[0][1]).resolve("__chkpts/index_pfam_hits.ok"),
+                    path(process.out.reference[0][1]).resolve("__chkpts/index_ref_annot_cdna.ok"),
+                    path(process.out.reference[0][1]).resolve("__chkpts/makeblastdb.ok"),
+                    path(process.out.reference[0][1]).resolve("__chkpts/mm2_genome_idx.ok"),
+                    path(process.out.reference[0][1]).resolve("__chkpts/mm2.splice_bed.ok"),
+                    path(process.out.reference[0][1]).resolve("__chkpts/_prot_info_db.ok"),
+                    path(process.out.reference[0][1]).resolve("__chkpts/ref_annot.cdsplus.dfam_masked.fa.cp.ok"),
+                    path(process.out.reference[0][1]).resolve("__chkpts/ref_annot.cdsplus.dfam_masked.fa.idx.ok"),
+                    path(process.out.reference[0][1]).resolve("__chkpts/ref_annot.gtf.gene_spans.ok"),
+                    path(process.out.reference[0][1]).resolve("__chkpts/ref_annot.gtf.mini.sortu.ok"),
+                    path(process.out.reference[0][1]).resolve("__chkpts/ref_annot.gtf.ok"),
+                    path(process.out.reference[0][1]).resolve("__chkpts/ref_genome_fai.ok"),
+                    path(process.out.reference[0][1]).resolve("__chkpts/ref_genome.fa.ok"),
+                    path(process.out.reference[0][1]).resolve("__chkpts/trans.blast.dat.cp.ok"),
+                    path(process.out.reference[0][1]).resolve("__chkpts/trans.blast.dat.index.ok"),
+                    path(process.out.reference[0][1]).resolve("__chkpts/validate_ctat_genome_lib.ok"),
+                    path(process.out.reference[0][1]).resolve("fusion_annot_lib.gz"),
+                    path(process.out.reference[0][1]).resolve("fusion_annot_lib.idx").exists(),
+                    path(process.out.reference[0][1]).resolve("pfam_domains.dbm").exists(),
+                    path(process.out.reference[0][1]).resolve("PFAM.domtblout.dat.gz").exists(),
+                    path(process.out.reference[0][1]).resolve("ref_annot.cdna.fa").exists(),
+                    path(process.out.reference[0][1]).resolve("ref_annot.cdna.fa.idx").exists(),
+                    path(process.out.reference[0][1]).resolve("ref_annot.cds").exists(),
+                    path(process.out.reference[0][1]).resolve("ref_annot.cdsplus.fa").exists(),
+                    path(process.out.reference[0][1]).resolve("ref_annot.cdsplus.fa.idx").exists(),
+                    path(process.out.reference[0][1]).resolve("ref_annot.gtf"),
+                    path(process.out.reference[0][1]).resolve("ref_annot.gtf.gene_spans").exists(),
+                    path(process.out.reference[0][1]).resolve("ref_annot.gtf.mini.sortu"),
+                    path(process.out.reference[0][1]).resolve("ref_annot.gtf.mm2.splice.bed"),
+                    path(process.out.reference[0][1]).resolve("ref_annot.pep").exists(),
+                    path(process.out.reference[0][1]).resolve("ref_annot.prot_info.dbm").exists(),
+                    path(process.out.reference[0][1]).resolve("ref_genome.fa"),
+                    path(process.out.reference[0][1]).resolve("ref_genome.fa.fai"),
+                    path(process.out.reference[0][1]).resolve("ref_genome.fa.mm2"),
+                    path(process.out.reference[0][1]).resolve("ref_genome.fa.ndb"),
+                    path(process.out.reference[0][1]).resolve("ref_genome.fa.nhr"),
+                    path(process.out.reference[0][1]).resolve("ref_genome.fa.nin").exists(),
+                    path(process.out.reference[0][1]).resolve("ref_genome.fa.njs").exists(),
+                    path(process.out.reference[0][1]).resolve("ref_genome.fa.not"),
+                    path(process.out.reference[0][1]).resolve("ref_genome.fa.nsq"),
+                    path(process.out.reference[0][1]).resolve("ref_genome.fa.ntf"),
+                    path(process.out.reference[0][1]).resolve("ref_genome.fa.nto"),
+                    path(process.out.reference[0][1]).resolve("ref_genome.fa.star.idx").exists(),
+                    path(process.out.reference[0][1]).resolve("ref_genome.fa.star.idx/build.ok"),
+                    path(process.out.reference[0][1]).resolve("ref_genome.fa.star.idx/chrLength.txt"),
+                    path(process.out.reference[0][1]).resolve("ref_genome.fa.star.idx/chrNameLength.txt"),
+                    path(process.out.reference[0][1]).resolve("ref_genome.fa.star.idx/chrName.txt"),
+                    path(process.out.reference[0][1]).resolve("ref_genome.fa.star.idx/chrStart.txt"),
+                    path(process.out.reference[0][1]).resolve("ref_genome.fa.star.idx/exonGeTrInfo.tab"),
+                    path(process.out.reference[0][1]).resolve("ref_genome.fa.star.idx/exonInfo.tab"),
+                    path(process.out.reference[0][1]).resolve("ref_genome.fa.star.idx/geneInfo.tab"),
+                    path(process.out.reference[0][1]).resolve("ref_genome.fa.star.idx/Genome"),
+                    path(process.out.reference[0][1]).resolve("ref_genome.fa.star.idx/genomeParameters.txt").exists(),
+                    path(process.out.reference[0][1]).resolve("ref_genome.fa.star.idx/Log.out").exists(),
+                    path(process.out.reference[0][1]).resolve("ref_genome.fa.star.idx/SA"),
+                    path(process.out.reference[0][1]).resolve("ref_genome.fa.star.idx/SAindex"),
+                    path(process.out.reference[0][1]).resolve("ref_genome.fa.star.idx/sjdbInfo.txt"),
+                    path(process.out.reference[0][1]).resolve("ref_genome.fa.star.idx/sjdbList.fromGTF.out.tab"),
+                    path(process.out.reference[0][1]).resolve("ref_genome.fa.star.idx/sjdbList.out.tab"),
+                    path(process.out.reference[0][1]).resolve("ref_genome.fa.star.idx/transcriptInfo.tab"),
+                    path(process.out.reference[0][1]).resolve("trans.blast.align_coords.align_coords.dat"),
+                    path(process.out.reference[0][1]).resolve("trans.blast.align_coords.align_coords.dbm").exists(),
+                    path(process.out.reference[0][1]).resolve("trans.blast.dat.gz"),
+                    process.out.versions
+                ).match()
+        }
+
+    }
+
+    test("STARFUSION_BUILD - human - minigenome - stub") {
+
+        options "-stub"
+
+        when {
+            process {
+                """
+                input[0] = [
+                    [ id:'minigenome fasta' ],
+                    file("https://github.com/STAR-Fusion/STAR-Fusion-Tutorial/raw/master/minigenome.fa")
+                ]
+                input[1] = [
+                    [ id:'minigenome gtf' ],
+                    file("https://github.com/STAR-Fusion/STAR-Fusion-Tutorial/raw/master/minigenome.gtf")
+                ]
+
+                input [2] = file("https://github.com/STAR-Fusion/STAR-Fusion-Tutorial/raw/master/CTAT_HumanFusionLib.mini.dat.gz")
+                input [3] = "human"
+
+                """
+            }
+        }
+
+        then {
+            assert snapshot(process.out).match()
+        }
+
+    }
+
+}