Add genmod (#1950)

* add annotate

* add models

* add compound

* add score

* update annotate recipe

* update meta and main

* main updates

* add test for annotate

* update all tests

* refactor

* update tests

* fix conda issue

* review suggestions

modules/genmod/annotate/main.nf

@@ -0,0 +1,46 @@
+process GENMOD_ANNOTATE {
+    tag "$meta.id"
+    label 'process_medium'
+
+    conda (params.enable_conda ? "bioconda::genmod=3.7.4" : null)
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/genmod:3.7.4--pyh5e36f6f_0':
+        'quay.io/biocontainers/genmod:3.7.4--pyh5e36f6f_0' }"
+
+    input:
+    tuple val(meta), path(input_vcf)
+
+    output:
+    tuple val(meta), path("*_annotate.vcf"), emit: vcf
+    path "versions.yml"           , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args = task.ext.args ?: ''
+    def prefix = task.ext.prefix ?: "${meta.id}"
+    """
+    genmod \\
+        annotate \\
+        $args \\
+        --outfile ${prefix}_annotate.vcf \\
+        $input_vcf
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        genmod: \$(echo \$(genmod --version 2>&1) | sed 's/^.*genmod version: //' ))
+    END_VERSIONS
+    """
+
+    stub:
+    def prefix = task.ext.prefix ?: "${meta.id}"
+    """
+    touch ${prefix}_annotate.vcf
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        genmod: \$(echo \$(genmod --version 2>&1) | sed 's/^.*genmod version: //' ))
+    END_VERSIONS
+    """
+}

modules/genmod/annotate/meta.yml

@@ -0,0 +1,40 @@
+name: "genmod_annotate"
+description: for annotating regions, frequencies, cadd scores
+keywords:
+  - annotate
+  - genmod
+tools:
+  - "genmod":
+      description: "Annotate genetic inheritance models in variant files"
+      homepage: "https://github.com/Clinical-Genomics/genmod"
+      documentation: "https://github.com/Clinical-Genomics/genmod"
+      tool_dev_url: "https://github.com/moonso"
+      doi: ""
+      licence: "['MIT']"
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - input_vcf:
+      type: file
+      description: VCF file
+      pattern: "*.{vcf}"
+output:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - versions:
+      type: file
+      description: File containing software versions
+      pattern: "versions.yml"
+  - vcf:
+      type: vcf
+      description: Annotated VCF file
+      pattern: "*.{vcf}"
+
+authors:
+  - "@ramprasadn"

modules/genmod/compound/main.nf

@@ -0,0 +1,46 @@
+process GENMOD_COMPOUND {
+    tag "$meta.id"
+    label 'process_medium'
+
+    conda (params.enable_conda ? "bioconda::genmod=3.7.4" : null)
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/genmod:3.7.4--pyh5e36f6f_0':
+        'quay.io/biocontainers/genmod:3.7.4--pyh5e36f6f_0' }"
+
+    input:
+    tuple val(meta), path(input_vcf)
+
+    output:
+    tuple val(meta), path("*_compound.vcf"), emit: vcf
+    path "versions.yml"           , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args = task.ext.args ?: ''
+    def prefix = task.ext.prefix ?: "${meta.id}"
+    """
+    genmod \\
+        compound \\
+        $args \\
+        --outfile ${prefix}_compound.vcf \\
+        $input_vcf
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        genmod: \$(echo \$(genmod --version 2>&1) | sed 's/^.*genmod version: //' ))
+    END_VERSIONS
+    """
+
+    stub:
+    def prefix = task.ext.prefix ?: "${meta.id}"
+    """
+    touch ${prefix}_compound.vcf
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        genmod: \$(echo \$(genmod --version 2>&1) | sed 's/^.*genmod version: //' ))
+    END_VERSIONS
+    """
+}

modules/genmod/compound/meta.yml

@@ -0,0 +1,40 @@
+name: "genmod_compound"
+description: Score compounds
+keywords:
+  - compound
+  - genmod
+tools:
+  - "genmod":
+      description: "Annotate genetic inheritance models in variant files"
+      homepage: "https://github.com/Clinical-Genomics/genmod"
+      documentation: "https://github.com/Clinical-Genomics/genmod"
+      tool_dev_url: "https://github.com/moonso"
+      doi: ""
+      licence: "['MIT']"
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - input_vcf:
+      type: file
+      description: VCF file
+      pattern: "*.{vcf}"
+output:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]  #
+  - versions:
+      type: file
+      description: File containing software versions
+      pattern: "versions.yml"
+  - vcf:
+      type: file
+      description: Output VCF file
+      pattern: "*.{vcf}"
+
+authors:
+  - "@ramprasadn"

modules/genmod/models/main.nf

@@ -0,0 +1,52 @@
+process GENMOD_MODELS {
+    tag "$meta.id"
+    label 'process_medium'
+
+    conda (params.enable_conda ? "bioconda::genmod=3.7.4 conda-forge::python=3.4.5" : null)
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/genmod:3.7.4--pyh5e36f6f_0':
+        'quay.io/biocontainers/genmod:3.7.4--pyh5e36f6f_0' }"
+
+    input:
+    tuple val(meta), path(input_vcf)
+    path (fam)
+    path (reduced_penetrance)
+
+    output:
+    tuple val(meta), path("*_models.vcf"), emit: vcf
+    path "versions.yml"           , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args        = task.ext.args ?: ''
+    def prefix      = task.ext.prefix ?: "${meta.id}"
+    def family_file =  fam ? "--family_file ${fam}" : ""
+    def pen_file    = reduced_penetrance ? "--reduced_penetrance ${reduced_penetrance}" : ""
+    """
+    genmod \\
+        models \\
+        $args \\
+        $pen_file \\
+        $family_file \\
+        --outfile ${prefix}_models.vcf \\
+        $input_vcf
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        genmod: \$(echo \$(genmod --version 2>&1) | sed 's/^.*genmod version: //' ))
+    END_VERSIONS
+    """
+
+    stub:
+    def prefix = task.ext.prefix ?: "${meta.id}"
+    """
+    touch ${prefix}_models.vcf
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        genmod: \$(echo \$(genmod --version 2>&1) | sed 's/^.*genmod version: //' ))
+    END_VERSIONS
+    """
+}

modules/genmod/models/meta.yml

@@ -0,0 +1,49 @@
+name: "genmod_models"
+description: annotate models of inheritance
+keywords:
+  - models
+  - genmod
+tools:
+  - "genmod":
+      description: "Annotate genetic inheritance models in variant files"
+      homepage: "https://github.com/Clinical-Genomics/genmod"
+      documentation: "https://github.com/Clinical-Genomics/genmod"
+      tool_dev_url: "https://github.com/moonso"
+      doi: ""
+      licence: "['MIT']"
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - input_vcf:
+      type: file
+      description: vcf file
+      pattern: "*.{vcf}"
+  - reduced_penetrance:
+      type: file
+      description: file with gene ids that have reduced penetrance
+      pattern: "*.{tsv}"
+  - family_file:
+      type: file
+      description: ped file
+      pattern: "*.{ped}"
+
+output:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - versions:
+      type: file
+      description: File containing software versions
+      pattern: "versions.yml"
+  - vcf:
+      type: file
+      description: Output VCF file
+      pattern: "*.{vcf}"
+
+authors:
+  - "@ramprasadn"

modules/genmod/score/main.nf

@@ -0,0 +1,55 @@
+process GENMOD_SCORE {
+    tag "$meta.id"
+    label 'process_medium'
+
+    conda (params.enable_conda ? "bioconda::genmod=3.7.4" : null)
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/genmod:3.7.4--pyh5e36f6f_0':
+        'quay.io/biocontainers/genmod:3.7.4--pyh5e36f6f_0' }"
+
+    input:
+    tuple val(meta), path(input_vcf)
+    path (fam)
+    path (reduced_penetrance)
+    path (score_config)
+
+    output:
+    tuple val(meta), path("*_score.vcf"), emit: vcf
+    path "versions.yml"           , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args        = task.ext.args ?: ''
+    def prefix      = task.ext.prefix ?: "${meta.id}"
+    def family_file = fam ? "--family_file ${fam}" : ""
+    def pen_file    = reduced_penetrance ? "--reduced_penetrance ${reduced_penetrance}" : ""
+    def config_file = score_config ? "--score_config ${score_config}" : ""
+    """
+    genmod \\
+        score \\
+        $args \\
+        $pen_file \\
+        $family_file \\
+        $config_file \\
+        --outfile ${prefix}_score.vcf \\
+        $input_vcf
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        genmod: \$(echo \$(genmod --version 2>&1) | sed 's/^.*genmod version: //' ))
+    END_VERSIONS
+    """
+
+    stub:
+    def prefix = task.ext.prefix ?: "${meta.id}"
+    """
+    touch ${prefix}_score.vcf
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        genmod: \$(echo \$(genmod --version 2>&1) | sed 's/^.*genmod version: //' ))
+    END_VERSIONS
+    """
+}

modules/genmod/score/meta.yml

@@ -0,0 +1,53 @@
+name: "genmod_score"
+description: Score the variants of a vcf based on their annotation
+keywords:
+  - score
+  - genmod
+tools:
+  - "genmod":
+      description: "Annotate genetic inheritance models in variant files"
+      homepage: "https://github.com/Clinical-Genomics/genmod"
+      documentation: "https://github.com/Clinical-Genomics/genmod"
+      tool_dev_url: "https://github.com/moonso"
+      doi: ""
+      licence: "['MIT']"
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - input_vcf:
+      type: file
+      description: vcf file
+      pattern: "*.{vcf}"
+  - reduced_penetrance:
+      type: file
+      description: file with gene ids that have reduced penetrance
+      pattern: "*.{tsv}"
+  - family_file:
+      type: file
+      description: ped file
+      pattern: "*.{ped}"
+  - score_config:
+      type: file
+      description: rank model config file
+      pattern: "*.{ini}"
+
+output:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - versions:
+      type: file
+      description: File containing software versions
+      pattern: "versions.yml"
+  - vcf:
+      type: file
+      description: Output VCF file
+      pattern: "*.{vcf}"
+
+authors:
+  - "@ramprasadn"

tests/config/pytest_modules.yml

@@ -983,6 +983,22 @@ genmap/mappability:
   - modules/genmap/mappability/**
   - tests/modules/genmap/mappability/**
 
+genmod/annotate:
+  - modules/genmod/annotate/**
+  - tests/modules/genmod/annotate/**
+
+genmod/compound:
+  - modules/genmod/compound/**
+  - tests/modules/genmod/compound/**
+
+genmod/models:
+  - modules/genmod/models/**
+  - tests/modules/genmod/models/**
+
+genmod/score:
+  - modules/genmod/score/**
+  - tests/modules/genmod/score/**
+
 genomescope2:
   - modules/genomescope2/**
   - tests/modules/genomescope2/**

tests/config/test_data.config

@@ -333,10 +333,15 @@ params {
 
                 test_rnaseq_vcf                               = "${test_data_dir}/genomics/homo_sapiens/illumina/vcf/test.rnaseq.vcf"
                 test_sv_vcf                                   = "${test_data_dir}/genomics/homo_sapiens/illumina/vcf/sv_query.vcf.gz"
+                genmod_vcf_gz                                 = "${test_data_dir}/genomics/homo_sapiens/illumina/vcf/genmod.vcf.gz"
+                genmod_annotate_vcf_gz                        = "${test_data_dir}/genomics/homo_sapiens/illumina/vcf/test_annotate.vcf.gz"
+                genmod_models_vcf_gz                          = "${test_data_dir}/genomics/homo_sapiens/illumina/vcf/test_models.vcf.gz"
+                genmod_score_vcf_gz                           = "${test_data_dir}/genomics/homo_sapiens/illumina/vcf/test_score.vcf.gz"
 
                 test_mito_vcf                                 = "${test_data_dir}/genomics/homo_sapiens/illumina/vcf/NA12878_chrM.vcf.gz"
 
                 test_pytor                                    = "${test_data_dir}/genomics/homo_sapiens/illumina/pytor/test.pytor"
+                rank_model                                    = "${test_data_dir}/genomics/homo_sapiens/illumina/genmod/svrank_model_-v1.8-.ini"
 
                 test_flowcell                                 = "${test_data_dir}/genomics/homo_sapiens/illumina/bcl/flowcell.tar.gz"
                 test_flowcell_samplesheet                     = "${test_data_dir}/genomics/homo_sapiens/illumina/bcl/flowcell_samplesheet.csv"

tests/modules/genmod/annotate/main.nf

@@ -0,0 +1,20 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { GENMOD_ANNOTATE } from '../../../../modules/genmod/annotate/main.nf'
+
+input = [
+    [ id:'test', single_end:false ], // meta map
+    file(params.test_data['homo_sapiens']['illumina']['genmod_vcf_gz'], checkIfExists: true)
+]
+
+workflow test_genmod_annotate {
+
+    GENMOD_ANNOTATE ( input )
+}
+
+workflow test_genmod_annotate_stub {
+
+    GENMOD_ANNOTATE ( input )
+}

tests/modules/genmod/annotate/nextflow.config

@@ -0,0 +1,8 @@
+process {
+
+    publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
+
+    withName: GENMOD_ANNOTATE {
+        ext.args = " --annotate_regions "
+    }
+}

tests/modules/genmod/annotate/test.yml

@@ -0,0 +1,17 @@
+- name: "genmod annotate"
+  command: nextflow run ./tests/modules/genmod/annotate -entry test_genmod_annotate -c ./tests/config/nextflow.config -c ./tests/modules/genmod/annotate/nextflow.config
+  tags:
+    - "genmod"
+    - "genmod/annotate"
+  files:
+    - path: "output/genmod/test_annotate.vcf"
+    - path: output/genmod/versions.yml
+
+- name: "genmod annotate stub"
+  command: nextflow run ./tests/modules/genmod/annotate -entry test_genmod_annotate_stub -c ./tests/config/nextflow.config -c ./tests/modules/genmod/annotate/nextflow.config -stub-run
+  tags:
+    - "genmod"
+    - "genmod/annotate"
+  files:
+    - path: "output/genmod/test_annotate.vcf"
+    - path: output/genmod/versions.yml

tests/modules/genmod/compound/main.nf

@@ -0,0 +1,22 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { GENMOD_COMPOUND } from '../../../../modules/genmod/compound/main.nf'
+
+input = [
+    [ id:'test', single_end:false ], // meta map
+    file(params.test_data['homo_sapiens']['illumina']['genmod_score_vcf_gz'], checkIfExists: true)
+]
+fam = file(params.test_data['homo_sapiens']['genome']['justhusky_ped'], checkIfExists: true)
+config = file(params.test_data['homo_sapiens']['illumina']['rank_model'], checkIfExists: true)
+
+workflow test_genmod_compound {
+
+    GENMOD_COMPOUND ( input )
+}
+
+workflow test_genmod_compound_stub {
+
+    GENMOD_COMPOUND ( input )
+}

tests/modules/genmod/compound/nextflow.config

@@ -0,0 +1,5 @@
+process {
+
+    publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
+
+}

tests/modules/genmod/compound/test.yml

@@ -0,0 +1,17 @@
+- name: "genmod compound"
+  command: nextflow run ./tests/modules/genmod/compound -entry test_genmod_compound -c ./tests/config/nextflow.config -c ./tests/modules/genmod/compound/nextflow.config
+  tags:
+    - "genmod"
+    - "genmod/compound"
+  files:
+    - path: output/genmod/test_compound.vcf
+    - path: output/genmod/versions.yml
+
+- name: "genmod compound stub"
+  command: nextflow run ./tests/modules/genmod/compound -entry test_genmod_compound_stub -c ./tests/config/nextflow.config -c ./tests/modules/genmod/compound/nextflow.config -stub-run
+  tags:
+    - "genmod"
+    - "genmod/compound"
+  files:
+    - path: output/genmod/test_compound.vcf
+    - path: output/genmod/versions.yml

tests/modules/genmod/models/main.nf

@@ -0,0 +1,21 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { GENMOD_MODELS   } from '../../../../modules/genmod/models/main.nf'
+
+input = [
+    [ id:'test', single_end:false ], // meta map
+    file(params.test_data['homo_sapiens']['illumina']['genmod_annotate_vcf_gz'], checkIfExists: true)
+]
+fam = file(params.test_data['homo_sapiens']['genome']['justhusky_ped'], checkIfExists: true)
+
+workflow test_genmod_models {
+
+    GENMOD_MODELS ( input, fam, [] )
+}
+
+workflow test_genmod_models_stub {
+
+    GENMOD_MODELS ( input, fam, [] )
+}

tests/modules/genmod/models/nextflow.config

@@ -0,0 +1,5 @@
+process {
+
+    publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
+
+}

tests/modules/genmod/models/test.yml

@@ -0,0 +1,17 @@
+- name: "genmod models"
+  command: nextflow run ./tests/modules/genmod/models -entry test_genmod_models -c ./tests/config/nextflow.config -c ./tests/modules/genmod/models/nextflow.config
+  tags:
+    - "genmod"
+    - "genmod/models"
+  files:
+    - path: output/genmod/test_models.vcf
+    - path: output/genmod/versions.yml
+
+- name: "genmod models stub"
+  command: nextflow run ./tests/modules/genmod/models -entry test_genmod_models_stub -c ./tests/config/nextflow.config -c ./tests/modules/genmod/models/nextflow.config -stub-run
+  tags:
+    - "genmod"
+    - "genmod/models"
+  files:
+    - path: output/genmod/test_models.vcf
+    - path: output/genmod/versions.yml

tests/modules/genmod/score/main.nf

@@ -0,0 +1,22 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { GENMOD_SCORE    } from '../../../../modules/genmod/score/main.nf'
+
+input = [
+    [ id:'test', single_end:false ], // meta map
+    file(params.test_data['homo_sapiens']['illumina']['genmod_models_vcf_gz'], checkIfExists: true)
+]
+fam = file(params.test_data['homo_sapiens']['genome']['justhusky_ped'], checkIfExists: true)
+config = file(params.test_data['homo_sapiens']['illumina']['rank_model'], checkIfExists: true)
+
+workflow test_genmod_score {
+
+    GENMOD_SCORE    ( input, fam, [], config)
+}
+
+workflow test_genmod_score_stub {
+
+    GENMOD_SCORE    ( input, fam, [], config)
+}

tests/modules/genmod/score/nextflow.config

@@ -0,0 +1,5 @@
+process {
+
+    publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
+
+}

tests/modules/genmod/score/test.yml

@@ -0,0 +1,17 @@
+- name: "genmod score"
+  command: nextflow run ./tests/modules/genmod/score -entry test_genmod_score -c ./tests/config/nextflow.config -c ./tests/modules/genmod/score/nextflow.config
+  tags:
+    - "genmod"
+    - "genmod/score"
+  files:
+    - path: output/genmod/test_score.vcf
+    - path: output/genmod/versions.yml
+
+- name: "genmod score stub"
+  command: nextflow run ./tests/modules/genmod/score -entry test_genmod_score_stub -c ./tests/config/nextflow.config -c ./tests/modules/genmod/score/nextflow.config -stub-run
+  tags:
+    - "genmod"
+    - "genmod/score"
+  files:
+    - path: output/genmod/test_score.vcf
+    - path: output/genmod/versions.yml