Merge pull request #229 from ypriverol/dev

testing with dev version of OpenMS 2.8.0

bin/mzml_statistics.py

@@ -0,0 +1,59 @@
+#!/usr/bin/env python
+
+from pyopenms import MzMLFile, MSExperiment
+import os
+import pandas as pd
+import click
+
+CONTEXT_SETTINGS = dict(help_option_names=["-h", "--help"])
+
+
+@click.group(context_settings=CONTEXT_SETTINGS)
+def cli():
+    pass
+
+
+@click.command("mzml_dataframe")
+@click.option("--mzml_folder", "-d")
+@click.pass_context
+def mzml_dataframe(ctx, mzml_folder):
+
+    file_columns = ["File_Name", "SpectrumID", "MSLevel", "Charge", "MS2_peaks", "Base_Peak_Intensity"]
+    mzml_paths = list(i for i in os.listdir(mzml_folder) if i.endswith(".mzML"))
+    mzml_count = 1
+
+    def parse_mzml(file_name, file_columns):
+        info = []
+        exp = MSExperiment()
+        MzMLFile().load(file_name, exp)
+        for i in exp:
+            name = os.path.split(file_name)[1]
+            id = i.getNativeID()
+            MSLevel = i.getMSLevel()
+            if MSLevel == 2:
+                charge_state = i.getPrecursors()[0].getCharge()
+                peaks_tuple = i.get_peaks()
+                peak_per_ms2 = len(peaks_tuple[0])
+                if i.getMetaValue("base peak intensity"):
+                    base_peak_intensity = i.getMetaValue("base peak intensity")
+                else:
+                    base_peak_intensity = max(peaks_tuple[1]) if len(peaks_tuple[1]) > 0 else "null"
+                info_list = [name, id, 2, charge_state, peak_per_ms2, base_peak_intensity]
+            else:
+                info_list = [name, id, MSLevel, "null", "null", "null"]
+
+            info.append(info_list)
+
+        return pd.DataFrame(info, columns=file_columns)
+
+    for i in mzml_paths:
+        mzml_df = parse_mzml(mzml_folder + i, file_columns)
+        tsv_header = True if mzml_count == 1 else False
+        mzml_df.to_csv("mzml_info.tsv", mode="a", sep="\t", index=False, header=tsv_header)
+        mzml_count += 1
+
+
+cli.add_command(mzml_dataframe)
+
+if __name__ == "__main__":
+    cli()

modules/local/diannconvert/main.nf

@@ -2,11 +2,11 @@ process DIANNCONVERT {
     tag "$exp_design.Name"
     label 'process_low'
 
-    conda (params.enable_conda ? "conda-forge::pandas_schema conda-forge::lzstring bioconda::pmultiqc=0.0.13" : null)
+    conda (params.enable_conda ? "conda-forge::pandas_schema conda-forge::lzstring bioconda::pmultiqc=0.0.16" : null)
     if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
-        container "https://depot.galaxyproject.org/singularity/pmultiqc:0.0.13--pyhdfd78af_0"
+        container "https://depot.galaxyproject.org/singularity/pmultiqc:0.0.16--pyhdfd78af_0"
     } else {
-        container "quay.io/biocontainers/pmultiqc:0.0.13--pyhdfd78af_0"
+        container "quay.io/biocontainers/pmultiqc:0.0.16--pyhdfd78af_0"
     }
 
     input:

modules/local/mzmlstatistics/main.nf

@@ -0,0 +1,34 @@
+process MZMLSTATISTICS {
+    label 'process_medium'
+    // TODO could be easily parallelized
+    label 'process_single_thread'
+
+    conda (params.enable_conda ? "conda-forge::pandas_schema conda-forge::lzstring bioconda::pmultiqc=0.0.16" : null)
+    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+        container "https://depot.galaxyproject.org/singularity/pmultiqc:0.0.16--pyhdfd78af_0"
+    } else {
+        container "quay.io/biocontainers/pmultiqc:0.0.16--pyhdfd78af_0"
+    }
+
+    input:
+    path("out/*")
+
+    output:
+    path "mzml_info.tsv", emit: mzml_statistics
+    path "versions.yml", emit: version
+    path "*.log", emit: log
+
+    script:
+    def args = task.ext.args ?: ''
+
+    """
+    mzml_statistics.py mzml_dataframe \\
+        --mzml_folder "./out/" \\
+        |& tee mzml_statistics.log
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        pyopenms: \$(echo "2.8.0")
+    END_VERSIONS
+    """
+}

modules/local/mzmlstatistics/meta.yml

@@ -0,0 +1,26 @@
+name: MZMLSTATISTICS
+description: A module for mzMLs statistics
+keywords:
+  - mzML
+  - statistics
+tools:
+  - custom:
+      description: |
+        A custom module for mzMLs statistics.
+      homepage: https://github.com/bigbio/quantms
+      documentation: https://github.com/bigbio/quantms/tree/readthedocs
+input:
+  - mzmls:
+      type: dir
+      description: mzML files directory
+output:
+  - mzml_statistics:
+      type: file
+      description: mzMLs statistics file
+      pattern: "mzml_info.tsv"
+  - version:
+      type: file
+      description: File containing software version
+      pattern: "versions.yml"
+authors:
+  - "@wanghong"

modules/local/openms/consensusid/main.nf

@@ -7,8 +7,8 @@ process CONSENSUSID {
 
     conda (params.enable_conda ? "bioconda::openms=2.8.0" : null)
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-        'https://depot.galaxyproject.org/singularity/openms:2.8.0--h7ca0330_1' :
-        'quay.io/biocontainers/openms:2.8.0--h7ca0330_1' }"
+        'https://ftp.pride.ebi.ac.uk/pub/databases/pride/resources/tools/ghcr.io-openms-openms-executables-latest.img' :
+        'ghcr.io/openms/openms-executables:latest' }"
 
     input:
     tuple val(meta), path(id_file), val(qval_score)
@@ -37,7 +37,7 @@ process CONSENSUSID {
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
-        ConsensusID: \$(ConsensusID 2>&1  | grep -E '^Version(.*)' | sed 's/Version: //g')
+        ConsensusID: \$(ConsensusID 2>&1  | grep -E '^Version(.*)' | sed 's/Version: //g' | cut -d ' ' -f 1)
     END_VERSIONS
     """
 }

modules/local/openms/decoydatabase/main.nf

@@ -4,8 +4,8 @@ process DECOYDATABASE {
 
     conda (params.enable_conda ? "bioconda::openms=2.8.0" : null)
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-        'https://depot.galaxyproject.org/singularity/openms:2.8.0--h7ca0330_1' :
-        'quay.io/biocontainers/openms:2.8.0--h7ca0330_1' }"
+        'https://ftp.pride.ebi.ac.uk/pub/databases/pride/resources/tools/ghcr.io-openms-openms-executables-latest.img' :
+        'ghcr.io/openms/openms-executables:latest' }"
 
     input:
     path(db_for_decoy)
@@ -33,7 +33,7 @@ process DECOYDATABASE {
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
-        DecoyDatabase: \$(DecoyDatabase 2>&1  | grep -E '^Version(.*)' | sed 's/Version: //g')
+        DecoyDatabase: \$(DecoyDatabase 2>&1  | grep -E '^Version(.*)' | sed 's/Version: //g' | cut -d ' ' -f 1)
     END_VERSIONS
     """
 }

modules/local/openms/epifany/main.nf

@@ -6,8 +6,8 @@ process EPIFANY {
 
     conda (params.enable_conda ? "bioconda::openms=2.8.0" : null)
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-        'https://depot.galaxyproject.org/singularity/openms:2.8.0--h7ca0330_1' :
-        'quay.io/biocontainers/openms:2.8.0--h7ca0330_1' }"
+        'https://ftp.pride.ebi.ac.uk/pub/databases/pride/resources/tools/ghcr.io-openms-openms-executables-latest.img' :
+        'ghcr.io/openms/openms-executables:latest' }"
 
     input:
     tuple val(meta), path(consus_file)
@@ -36,7 +36,7 @@ process EPIFANY {
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
-        Epifany: \$(Epifany 2>&1 | grep -E '^Version(.*)' | sed 's/Version: //g')
+        Epifany: \$(Epifany 2>&1 | grep -E '^Version(.*)' | sed 's/Version: //g' | cut -d ' ' -f 1)
     END_VERSIONS
     """
 }

modules/local/openms/extractpsmfeatures/main.nf

@@ -6,8 +6,8 @@ process EXTRACTPSMFEATURES {
 
     conda (params.enable_conda ? "bioconda::openms=2.8.0" : null)
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-        'https://depot.galaxyproject.org/singularity/openms:2.8.0--h7ca0330_1' :
-        'quay.io/biocontainers/openms:2.8.0--h7ca0330_1' }"
+        'https://ftp.pride.ebi.ac.uk/pub/databases/pride/resources/tools/ghcr.io-openms-openms-executables-latest.img' :
+        'ghcr.io/openms/openms-executables:latest' }"
 
     input:
     tuple val(meta), path(id_file)
@@ -31,7 +31,7 @@ process EXTRACTPSMFEATURES {
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
-        PSMFeatureExtractor: \$(PSMFeatureExtractor 2>&1 | grep -E '^Version(.*)' | sed 's/Version: //g')
+        PSMFeatureExtractor: \$(PSMFeatureExtractor 2>&1 | grep -E '^Version(.*)' | sed 's/Version: //g' | cut -d ' ' -f 1)
     END_VERSIONS
     """
 }

modules/local/openms/falsediscoveryrate/main.nf

@@ -6,8 +6,8 @@ process FALSEDISCOVERYRATE {
 
     conda (params.enable_conda ? "bioconda::openms=2.8.0" : null)
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-        'https://depot.galaxyproject.org/singularity/openms:2.8.0--h7ca0330_1' :
-        'quay.io/biocontainers/openms:2.8.0--h7ca0330_1' }"
+        'https://ftp.pride.ebi.ac.uk/pub/databases/pride/resources/tools/ghcr.io-openms-openms-executables-latest.img' :
+        'ghcr.io/openms/openms-executables:latest' }"
 
     input:
     tuple val(meta), path(id_file)
@@ -33,7 +33,7 @@ process FALSEDISCOVERYRATE {
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
-        FalseDiscoveryRate: \$(FalseDiscoveryRate 2>&1 | grep -E '^Version(.*)' | sed 's/Version: //g')
+        FalseDiscoveryRate: \$(FalseDiscoveryRate 2>&1 | grep -E '^Version(.*)' | sed 's/Version: //g' | cut -d ' ' -f 1)
     END_VERSIONS
     """
 }

modules/local/openms/filemerge/main.nf

@@ -5,8 +5,8 @@ process FILEMERGE {
 
     conda (params.enable_conda ? "bioconda::openms=2.8.0" : null)
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-        'https://depot.galaxyproject.org/singularity/openms:2.8.0--h7ca0330_1' :
-        'quay.io/biocontainers/openms:2.8.0--h7ca0330_1' }"
+        'https://ftp.pride.ebi.ac.uk/pub/databases/pride/resources/tools/ghcr.io-openms-openms-executables-latest.img' :
+        'ghcr.io/openms/openms-executables:latest' }"
 
     input:
     file(id_map)
@@ -32,7 +32,7 @@ process FILEMERGE {
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
-        FileMerger: \$(FileMerger 2>&1 | grep -E '^Version(.*)' | sed 's/Version: //g')
+        FileMerger: \$(FileMerger 2>&1 | grep -E '^Version(.*)' | sed 's/Version: //g' | cut -d ' ' -f 1)
     END_VERSIONS
     """
 }

modules/local/openms/idconflictresolver/main.nf

@@ -4,8 +4,8 @@ process IDCONFLICTRESOLVER {
 
     conda (params.enable_conda ? "bioconda::openms=2.8.0" : null)
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-        'https://depot.galaxyproject.org/singularity/openms:2.8.0--h7ca0330_1' :
-        'quay.io/biocontainers/openms:2.8.0--h7ca0330_1' }"
+        'https://ftp.pride.ebi.ac.uk/pub/databases/pride/resources/tools/ghcr.io-openms-openms-executables-latest.img' :
+        'ghcr.io/openms/openms-executables:latest' }"
 
     input:
     path consus_file
@@ -28,7 +28,7 @@ process IDCONFLICTRESOLVER {
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
-        IDConflictResolver: \$(IDConflictResolver 2>&1 | grep -E '^Version(.*)' | sed 's/Version: //g')
+        IDConflictResolver: \$(IDConflictResolver 2>&1 | grep -E '^Version(.*)' | sed 's/Version: //g' | cut -d ' ' -f 1)
     END_VERSIONS
     """
 }

modules/local/openms/idfilter/main.nf

@@ -6,8 +6,8 @@ process IDFILTER {
 
     conda (params.enable_conda ? "bioconda::openms=2.8.0" : null)
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-        'https://depot.galaxyproject.org/singularity/openms:2.8.0--h7ca0330_1' :
-        'quay.io/biocontainers/openms:2.8.0--h7ca0330_1' }"
+        'https://ftp.pride.ebi.ac.uk/pub/databases/pride/resources/tools/ghcr.io-openms-openms-executables-latest.img' :
+        'ghcr.io/openms/openms-executables:latest' }"
 
     input:
     tuple val(meta), path(id_file)
@@ -31,7 +31,7 @@ process IDFILTER {
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
-        IDFilter: \$(IDFilter 2>&1 | grep -E '^Version(.*)' | sed 's/Version: //g')
+        IDFilter: \$(IDFilter 2>&1 | grep -E '^Version(.*)' | sed 's/Version: //g' | cut -d ' ' -f 1)
     END_VERSIONS
     """
 }

modules/local/openms/idmapper/main.nf

@@ -6,8 +6,8 @@ process IDMAPPER {
 
     conda (params.enable_conda ? "bioconda::openms=2.8.0" : null)
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-        'https://depot.galaxyproject.org/singularity/openms:2.8.0--h7ca0330_1' :
-        'quay.io/biocontainers/openms:2.8.0--h7ca0330_1' }"
+        'https://ftp.pride.ebi.ac.uk/pub/databases/pride/resources/tools/ghcr.io-openms-openms-executables-latest.img' :
+        'ghcr.io/openms/openms-executables:latest' }"
 
     input:
     tuple val(meta), path(id_file), path(map_file)
@@ -32,7 +32,7 @@ process IDMAPPER {
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
-        IDMapper: \$(IDMapper 2>&1 | grep -E '^Version(.*)' | sed 's/Version: //g')
+        IDMapper: \$(IDMapper 2>&1 | grep -E '^Version(.*)' | sed 's/Version: //g' | cut -d ' ' -f 1)
     END_VERSIONS
     """
 }

modules/local/openms/idpep/main.nf

@@ -4,8 +4,8 @@ process IDPEP {
 
     conda (params.enable_conda ? "bioconda::openms=2.8.0" : null)
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-        'https://depot.galaxyproject.org/singularity/openms:2.8.0--h7ca0330_1' :
-        'quay.io/biocontainers/openms:2.8.0--h7ca0330_1' }"
+        'https://ftp.pride.ebi.ac.uk/pub/databases/pride/resources/tools/ghcr.io-openms-openms-executables-latest.img' :
+        'ghcr.io/openms/openms-executables:latest' }"
 
     input:
     tuple val(meta), path(id_file)
@@ -30,7 +30,7 @@ process IDPEP {
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
-        IDPosteriorErrorProbability: \$(IDPosteriorErrorProbability 2>&1 | grep -E '^Version(.*)' | sed 's/Version: //g')
+        IDPosteriorErrorProbability: \$(IDPosteriorErrorProbability 2>&1 | grep -E '^Version(.*)' | sed 's/Version: //g' | cut -d ' ' -f 1)
     END_VERSIONS
     """
 }

modules/local/openms/idscoreswitcher/main.nf

@@ -5,8 +5,8 @@ process IDSCORESWITCHER {
 
     conda (params.enable_conda ? "bioconda::openms=2.8.0" : null)
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-        'https://depot.galaxyproject.org/singularity/openms:2.8.0--h7ca0330_1' :
-        'quay.io/biocontainers/openms:2.8.0--h7ca0330_1' }"
+        'https://ftp.pride.ebi.ac.uk/pub/databases/pride/resources/tools/ghcr.io-openms-openms-executables-latest.img' :
+        'ghcr.io/openms/openms-executables:latest' }"
 
     input:
     tuple val(meta), path(id_file), val(new_score)
@@ -31,7 +31,7 @@ process IDSCORESWITCHER {
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
-        IDScoreSwitcher: \$(IDScoreSwitcher 2>&1 | grep -E '^Version(.*)' | sed 's/Version: //g')
+        IDScoreSwitcher: \$(IDScoreSwitcher 2>&1 | grep -E '^Version(.*)' | sed 's/Version: //g' | cut -d ' ' -f 1)
     END_VERSIONS
     """
 }

modules/local/openms/indexpeptides/main.nf

@@ -4,8 +4,8 @@ process INDEXPEPTIDES {
 
     conda (params.enable_conda ? "bioconda::openms=2.8.0" : null)
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-        'https://depot.galaxyproject.org/singularity/openms:2.8.0--h7ca0330_1' :
-        'quay.io/biocontainers/openms:2.8.0--h7ca0330_1' }"
+        'https://ftp.pride.ebi.ac.uk/pub/databases/pride/resources/tools/ghcr.io-openms-openms-executables-latest.img' :
+        'ghcr.io/openms/openms-executables:latest' }"
 
     input:
     tuple val(meta), path(id_file), path(database)
@@ -57,7 +57,7 @@ process INDEXPEPTIDES {
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
-        PeptideIndexer: \$(PeptideIndexer 2>&1 | grep -E '^Version(.*)' | sed 's/Version: //g')
+        PeptideIndexer: \$(PeptideIndexer 2>&1 | grep -E '^Version(.*)' | sed 's/Version: //g' | cut -d ' ' -f 1)
     END_VERSIONS
     """
 }

modules/local/openms/isobaricanalyzer/main.nf

@@ -4,8 +4,8 @@ process ISOBARICANALYZER {
 
     conda (params.enable_conda ? "bioconda::openms=2.8.0" : null)
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-        'https://depot.galaxyproject.org/singularity/openms:2.8.0--h7ca0330_1' :
-        'quay.io/biocontainers/openms:2.8.0--h7ca0330_1' }"
+        'https://ftp.pride.ebi.ac.uk/pub/databases/pride/resources/tools/ghcr.io-openms-openms-executables-latest.img' :
+        'ghcr.io/openms/openms-executables:latest' }"
 
     input:
     tuple val(meta), path(mzml_file)
@@ -44,7 +44,7 @@ process ISOBARICANALYZER {
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
-        IsobaricAnalyzer: \$(IsobaricAnalyzer --version 2>&1 | grep -E '^Version(.*)' | sed 's/Version: //g')
+        IsobaricAnalyzer: \$(IsobaricAnalyzer --version 2>&1 | grep -E '^Version(.*)' | sed 's/Version: //g' | cut -d ' ' -f 1)
     END_VERSIONS
     """
 }