Merge pull request #378 from daichengxin/dev

ms2rescore for quant workflow and fixed some bugs
bigbio · Jun 17, 2024 · 274d082 · 274d082
2 parents d64652b + 4111c36
commit 274d082
Show file tree

Hide file tree

Showing 12 changed files with 254 additions and 41 deletions.
diff --git a/bin/ms2rescore_cli.py b/bin/ms2rescore_cli.py
@@ -36,6 +36,7 @@ def parse_cli_arguments_to_config(**kwargs):
                 config["ms2rescore"]["feature_generators"]["basic"] = {}
             if "ms2pip" in feature_generators:
                 config["ms2rescore"]["feature_generators"]["ms2pip"] = {
+                    "model_dir": kwargs["ms2pip_model_dir"],
                     "model": kwargs["ms2pip_model"],
                     "ms2_tolerance": kwargs["ms2_tolerance"],
                 }
@@ -147,6 +148,7 @@ def filter_out_artifact_psms(
     default="",
 )
 @click.option("-pipm", "--ms2pip_model", help="MS²PIP model (default: `Immuno-HCD`)", type=str, default="Immuno-HCD")
+@click.option("-md", "--ms2pip_model_dir", help="The path of MS²PIP model (default: `./`)", type=str, default="./")
 @click.option(
     "-ms2tol", "--ms2_tolerance", help="Fragment mass tolerance [Da](default: `0.02`)", type=float, default=0.02
 )

diff --git a/conf/modules.config b/conf/modules.config
@@ -317,7 +317,7 @@ process {
         ]
     }
 
-    withName: '.*:DDA_ID:SAGEFEATURE' {
+    withName: '.*:SAGEFEATURE' {
         publishDir  = [
             path: { "${params.outdir}/addsagefeature" },
             pattern: "*.log",

diff --git a/modules/local/extract_psm/main.nf b/modules/local/extract_psm/main.nf
@@ -2,11 +2,11 @@ process PSMCONVERSION {
     tag "$meta.mzml_id"
     label 'process_medium'
 
-    conda "bioconda::pyopenms=2.8.0"
+    conda "bioconda::pyopenms=3.1.0"
     if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
-        container "https://depot.galaxyproject.org/singularity/pyopenms:2.8.0--py38hd8d5640_1"
+        container "https://depot.galaxyproject.org/singularity/pyopenms:3.1.0--py39h9b8898c_0"
     } else {
-        container "biocontainers/pyopenms:2.8.0--py38hd8d5640_1"
+        container "biocontainers/pyopenms:3.1.0--py39h9b8898c_0"
     }
 
     input:

diff --git a/modules/local/ms2rescore/main.nf b/modules/local/ms2rescore/main.nf
@@ -2,10 +2,10 @@ process MS2RESCORE {
     tag "$meta.mzml_id"
     label 'process_high'
 
-    conda "bioconda::ms2rescore=3.0.2 bioconda::psm-utils=0.8.0 conda-forge::pydantic=1.10"
+    conda "bioconda::ms2rescore=3.0.3 bioconda::psm-utils=0.8.0 conda-forge::pydantic=1.10"
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-        'https://depot.galaxyproject.org/singularity/ms2rescore:3.0.2--pyhdfd78af_0':
-        'biocontainers/ms2rescore:3.0.2--pyhdfd78af_0' }"
+        'https://depot.galaxyproject.org/singularity/ms2rescore:3.0.3--pyhdfd78af_0':
+        'biocontainers/ms2rescore:3.0.3--pyhdfd78af_0' }"
 
     // userEmulation settings when docker is specified
     containerOptions = (workflow.containerEngine == 'docker') ? '-u $(id -u) -e "HOME=${HOME}" -v /etc/passwd:/etc/passwd:ro -v /etc/shadow:/etc/shadow:ro -v /etc/group:/etc/group:ro -v $HOME:$HOME' : ''
@@ -48,6 +48,7 @@ process MS2RESCORE {
         --spectrum_path . \\
         --ms2_tolerance $ms2_tolerence \\
         --output_path ${idxml.baseName}_ms2rescore.idXML \\
+        --ms2pip_model_dir ${params.ms2pip_model_dir} \\
         --processes $task.cpus \\
         --id_decoy_pattern $decoy_pattern \\
         $args \\

diff --git a/modules/local/mzmlstatistics/main.nf b/modules/local/mzmlstatistics/main.nf
@@ -3,11 +3,11 @@ process MZMLSTATISTICS {
     label 'process_medium'
     label 'process_single'
 
-    conda "bioconda::pyopenms=2.8.0"
+    conda "bioconda::pyopenms=3.1.0"
     if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
-        container "https://depot.galaxyproject.org/singularity/pyopenms:2.8.0--py38hd8d5640_1"
+        container "https://depot.galaxyproject.org/singularity/pyopenms:3.1.0--py39h9b8898c_0"
     } else {
-        container "biocontainers/pyopenms:2.8.0--py38hd8d5640_1"
+        container "biocontainers/pyopenms:3.1.0--py39h9b8898c_0"
     }
 
     input:

diff --git a/nextflow.config b/nextflow.config
@@ -106,6 +106,7 @@ params {
 
     // MSRESCORE flags
     ms2rescore                      = false
+    ms2pip_model_dir                = null
     rescore_range                   = 'independent_run'
     ms2pip_model                    = 'HCD2021'
     feature_generators              = 'deeplc,ms2pip'
@@ -124,7 +125,7 @@ params {
     // Percolator flags
     train_FDR                    = 0.05
     test_FDR                     = 0.05
-    fdr_level                    = 'peptide_level_fdrs'
+    fdr_level                    = 'psm_level_fdrs'
     klammer                      = false
     description_correct_features = 0
     subset_max_train             = 300000

diff --git a/nextflow_schema.json b/nextflow_schema.json
@@ -460,6 +460,11 @@
                     "fa_icon": "fas fa-font",
                     "default": "HCD2021"
                 },
+                "ms2pip_model_dir": {
+                    "type": "string",
+                    "description": "The path of ms2pip model files. Providing model file to avoid repeated download and slow internet connection",
+                    "fa_icon": "fas fa-font"
+                },
                 "feature_generators": {
                     "type": "string",
                     "description": "Which feature generator to generate feature.",
@@ -525,7 +530,7 @@
                 "fdr_level": {
                     "type": "string",
                     "description": "Calculate FDR on PSM ('psm_level_fdrs') or peptide level ('peptide_level_fdrs')?",
-                    "default": "peptide_level_fdrs",
+                    "default": "psm_level_fdrs",
                     "fa_icon": "fas fa-list-ol",
                     "enum": ["peptide_level_fdrs", "psm_level_fdrs"]
                 },

diff --git a/subworkflows/local/dda_id.nf b/subworkflows/local/dda_id.nf
@@ -86,7 +86,7 @@ workflow DDA_ID {
             } else if (params.rescore_range == "by_sample") {
                 // Sample map
                 GETSAMPLE(ch_expdesign)
-                ch_expdesign_sample = EXTRACT_SAMPLE.out.ch_expdesign_sample
+                ch_expdesign_sample = GETSAMPLE.out.ch_expdesign_sample
                 ch_expdesign_sample.splitCsv(header: true, sep: '\t')
                     .map { get_sample_map(it) }.set{ sample_map_idv }
 
@@ -155,7 +155,7 @@ workflow DDA_ID {
             IDSCORESWITCHER(MS2RESCORE.out.idxml.combine(Channel.value("PEP")))
             ch_software_versions = ch_software_versions.mix(IDSCORESWITCHER.out.version)
             ch_consensus_input = IDSCORESWITCHER.out.id_score_switcher.combine(Channel.value("MS:1001491"))
-            ch_rescoring_results = IDSCORESWITCHER.out.id_files_ForIDPEP
+            ch_rescoring_results = IDSCORESWITCHER.out.ch_consensus_input
         } else {
             ch_fdridpep = Channel.empty()
             if (params.search_engines.split(",").size() == 1) {
@@ -224,13 +224,25 @@ def convert_exp_meta(Map meta, value, file_name, sample_map) {
     } else if (value == "sample_id") {
         tag = file(file_name).name.lastIndexOf('_perc.idXML')
         if (tag == -1) {
-            position = file(file_name).name.lastIndexOf('_sage.idXML')
-            if (position == -1) {
-                position = file(file_name).name.lastIndexOf('_comet_feat.idXML')
+            ifms2rescore = file(file_name).name.lastIndexOf('_ms2rescore_')
+            if (ifms2rescore == -1) {
+                position = file(file_name).name.lastIndexOf('_sage.idXML')
                 if (position == -1) {
-                    position = file(file_name).name.lastIndexOf('_msgf_feat.idXML')
+                    position = file(file_name).name.lastIndexOf('_comet_feat.idXML')
+                    if (position == -1) {
+                        position = file(file_name).name.lastIndexOf('_msgf_feat.idXML')
+                    }
+                }
+            } else {
+                position = file(file_name).name.lastIndexOf('_sage_ms2rescore.idXML')
+                if (position == -1) {
+                    position = file(file_name).name.lastIndexOf('_comet_ms2rescore_feat.idXML')
+                    if (position == -1) {
+                        position = file(file_name).name.lastIndexOf('_msgf_ms2rescore_feat.idXML')
+                    }
                 }
             }
+
         } else {
             position = file(file_name).name.lastIndexOf('_sage_perc.idXML')
             if (position == -1) {

diff --git a/subworkflows/local/id.nf b/subworkflows/local/id.nf
@@ -16,6 +16,7 @@ workflow ID {
     take:
     ch_file_preparation_results
     ch_database_wdecoy
+    ch_expdesign
 
     main:
 
@@ -33,7 +34,7 @@ workflow ID {
     //
     // SUBWORKFLOW: PSMReScoring
     //
-    PSMRESCORING (DATABASESEARCHENGINES.out.ch_id_files_idx)
+    PSMRESCORING (ch_file_preparation_results, DATABASESEARCHENGINES.out.ch_id_files_idx, ch_expdesign)
     ch_software_versions = ch_software_versions.mix(PSMRESCORING.out.versions.ifEmpty(null))
 
     //