Merge pull request #374 from daichengxin/dev

add msgfdb_indexing
bigbio · May 16, 2024 · d64652b · d64652b
2 parents bb0b114 + 55c8d3b
commit d64652b
Show file tree

Hide file tree

Showing 4 changed files with 72 additions and 2 deletions.
diff --git a/modules/local/openms/thirdparty/msgfdb_indexing/main.nf b/modules/local/openms/thirdparty/msgfdb_indexing/main.nf
@@ -0,0 +1,36 @@
+process MSGFDBINDEXING {
+    tag "$database.baseName"
+    label 'process_low'
+    label 'openms'
+
+    conda "bioconda::openms-thirdparty=3.1.0"
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/openms-thirdparty:3.1.0--h9ee0642_1' :
+        'biocontainers/openms-thirdparty:3.1.0--h9ee0642_1' }"
+
+    input:
+    path(database)
+
+    output:
+    tuple path("${database.baseName}.cnlcp"), path("${database.baseName}.canno"), path("${database.baseName}.csarr"), path("${database.baseName}.cseq"), emit: msgfdb_idx
+    path "versions.yml",   emit: version
+    path "*.log",   emit: log
+
+    script:
+    def args = task.ext.args ?: ''
+
+    """
+    msgf_plus edu.ucsd.msjava.msdbsearch.BuildSA \\
+        -d ${database} \\
+        -o ./ \\
+        -tda 0 \\
+        -debug $params.db_debug \\
+        $args \\
+        2>&1 | tee ${database.baseName}_msgfdb_idx.log
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        msgf_plus: \$(msgf_plus 2>&1 | grep -E '^MS-GF\\+ Release.*')
+    END_VERSIONS
+    """
+}
diff --git a/modules/local/openms/thirdparty/msgfdb_indexing/meta.yml b/modules/local/openms/thirdparty/msgfdb_indexing/meta.yml
@@ -0,0 +1,32 @@
+name: msgfdb_indexing
+description: Indexing the database prior to the database search for MS-GF+.
+keywords:
+  - Indexing
+  - database
+  - msgf
+tools:
+  - msgf_plus:
+      description: |
+        The MS-GF+ protein identification (database search) engine.
+      homepage: https://abibuilder.cs.uni-tuebingen.de/archive/openms/Documentation/nightly/html/TOPP_MSGFPlusAdapter.html
+      documentation: https://abibuilder.cs.uni-tuebingen.de/archive/openms/Documentation/nightly/html/TOPP_MSGFPlusAdapter.html
+input:
+  - database:
+      type: file
+      description: fasta file
+      pattern: "*.{fasta,fa}"
+output:
+  - msgfdb_idx:
+      type: file
+      description: Database indexing files
+      pattern: "*.{cnlcp,canno,csarr,cseq}"
+  - log:
+      type: file
+      description: log file
+      pattern: "*.log"
+  - version:
+      type: file
+      description: File containing software version
+      pattern: "versions.yml"
+authors:
+  - "@daichengxin"
diff --git a/modules/local/openms/thirdparty/searchenginemsgf/main.nf b/modules/local/openms/thirdparty/searchenginemsgf/main.nf
@@ -9,7 +9,7 @@ process SEARCHENGINEMSGF {
         'biocontainers/openms-thirdparty:3.1.0--h9ee0642_1' }"
 
     input:
-    tuple val(meta),  path(mzml_file), path(database)
+    tuple val(meta),  path(mzml_file), path(database), path(cnlcp), path(canno), path(csarr), path(cseq)
 
     output:
     tuple val(meta), path("${mzml_file.baseName}_msgf.idXML"),  emit: id_files_msgf

diff --git a/subworkflows/local/databasesearchengines.nf b/subworkflows/local/databasesearchengines.nf
@@ -5,6 +5,7 @@
 include { SEARCHENGINEMSGF  } from '../../modules/local/openms/thirdparty/searchenginemsgf/main'
 include { SEARCHENGINECOMET } from '../../modules/local/openms/thirdparty/searchenginecomet/main'
 include { SEARCHENGINESAGE  } from '../../modules/local/openms/thirdparty/searchenginesage/main'
+include { MSGFDBINDEXING    } from '../../modules/local/openms/thirdparty/msgfdb_indexing/main'
 
 workflow DATABASESEARCHENGINES {
     take:
@@ -15,7 +16,8 @@ workflow DATABASESEARCHENGINES {
     (ch_id_msgf, ch_id_comet, ch_id_sage, ch_versions) = [ Channel.empty(), Channel.empty(), Channel.empty(), Channel.empty() ]
 
     if (params.search_engines.contains("msgf")) {
-        SEARCHENGINEMSGF(ch_mzmls_search.combine(ch_searchengine_in_db))
+        MSGFDBINDEXING(ch_searchengine_in_db)
+        SEARCHENGINEMSGF(ch_mzmls_search.combine(ch_searchengine_in_db).combine(MSGFDBINDEXING.out.msgfdb_idx))
         ch_versions = ch_versions.mix(SEARCHENGINEMSGF.out.version)
         ch_id_msgf = ch_id_msgf.mix(SEARCHENGINEMSGF.out.id_files_msgf)
     }