version 1 - plotting test

mdibl · Jul 17, 2024 · 3a3739c · 3a3739c
1 parent 72f2a1e
commit 3a3739c
Show file tree

Hide file tree

Showing 13 changed files with 385 additions and 107 deletions.
diff --git a/Samples.csv b/Samples.csv
@@ -1,2 +1,3 @@
-id,data_directory,genes_2_rm,mito_genes
-test,
+id,data_directory,features_2_rm,mito_genes
+samp_1,/Users/rgrindle/Desktop/mdibl/wd/scRNA-seq/test_pipe/Samp1,,test_mito.csv
+samp_2,/Users/rgrindle/Desktop/mdibl/wd/scRNA-seq/test_pipe/Samp2,,test_mito.csv
diff --git a/conf/modules.config b/conf/modules.config
@@ -12,11 +12,65 @@
 
 process {
 
-    publishDir = [
-        path: { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" },
-        mode: params.publish_dir_mode,
-        saveAs: { filename -> filename.equals('versions.yml') ? null : filename }
-    ]
+    withName: 'DISPLAY_REDUCTION' {
+        publishDir = [
+            path: { "${params.outdir}/Group_Analysis/${meta}/Dimensional_Reduction" },
+            mode: params.publish_dir_mode
+        ]
+        ext.args   = params.integration_method
+    }
+
+    withName: 'FIND_NN_CLUSTER' {
+        publishDir = [
+            path: { "${params.outdir}/Group_Analysis/${meta.group}/Nearest_Neighbor-Clusters" },
+            mode: params.publish_dir_mode
+        ]
+    }
+
+    withName: 'INTEGRATION' {
+        publishDir = [
+            path: { "${params.outdir}/Group_Analysis/${meta}/Integration" },
+            mode: params.publish_dir_mode
+        ]
+    }
+
+    withName: 'RUN_PCA' {
+        publishDir = [
+            path: { "${params.outdir}/Group_Analysis/${meta}/PCA" },
+            mode: params.publish_dir_mode
+        ]
+        ext.args   = params.pcMax
+    }
+
+    withName: 'MERGE_SO' {
+        publishDir = [
+            path: { "${params.outdir}/Group_Analysis/${meta}/Merge" },
+            mode: params.publish_dir_mode
+        ]
+        ext.args2 = params.n_features_2_scale
+        ext.args3 = params.scale_method
+    }
+
+    withName: 'FIND_DOUBLETS' {
+        publishDir = [
+            path: { "${params.outdir}/Sample_Processing/${meta.id}/Doublet_Detection" },
+            mode: params.publish_dir_mode
+        ]
+    }
+
+    withName: 'NORMALIZE_QC' {
+        publishDir = [
+            path: { "${params.outdir}/Sample_Processing/${meta.id}/Normalize_QC" },
+            mode: params.publish_dir_mode
+        ]
+    }
+
+    withName: 'MAKE_SEURAT' {
+        publishDir = [
+            path: { "${params.outdir}/Sample_Processing/${meta.id}/Inital_SO" },
+            mode: params.publish_dir_mode
+        ]
+    }
 
     withName: SAMPLESHEET_CHECK {
         publishDir = [
@@ -26,9 +80,6 @@ process {
         ]
     }
 
-    withName: FASTQC {
-        ext.args = '--quiet'
-    }
 
     withName: CUSTOM_DUMPSOFTWAREVERSIONS {
         publishDir = [

diff --git a/modules/local/doubletfinder.nf b/modules/local/doubletfinder.nf
@@ -1,7 +1,7 @@
 process FIND_DOUBLETS {
 
     tag "${meta.id}"
-    label 'process_medium'
+    label 'process_small'
 
     conda "conda-forge::python=3.9.5"
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
@@ -12,11 +12,12 @@ process FIND_DOUBLETS {
     tuple val(meta), path (rds)
     tuple val(meta), path (data_directory)
     val vars_2_regress
-    
+
 
     output:
-    tuple val(meta), path ("*_DoubletsRemoved.rds"), emit: rds
-    path("*.validation.log"),           emit: log
+    tuple val(meta), path ("*_DoubletsRemoved.rds"),     emit: rds
+    tuple val(meta), path("*.validation.log"),           emit: log
+    path("*.pdf")
     //path ("versions.yml"),            emit: versions
 
     when:
@@ -30,12 +31,8 @@ process FIND_DOUBLETS {
         $vars_2_regress \\
         $data_directory \\
         ${meta.id} \\
-        ${args}      
+        ${args}
 
-    //cat <<-END_VERSIONS > versions.yml
-    //"${task.process}":
-        //Seurat: \$(echo \$( version) | sed "s/, version //g" )
-    //END_VERSIONS
     """
 
     stub:
@@ -46,4 +43,4 @@ process FIND_DOUBLETS {
     END_VERSIONS
     """
 
-} 
+}
diff --git a/modules/local/find_NN_clusters.nf b/modules/local/find_NN_clusters.nf
@@ -1,7 +1,7 @@
 process FIND_NN_CLUSTER {
 
-    tag "${meta.id}"
-    label 'process_medium'
+    tag "${meta.group}"
+    label 'process_small'
 
     conda "conda-forge::python=3.9.5"
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
@@ -10,35 +10,37 @@ process FIND_NN_CLUSTER {
 
     input:
     tuple val(meta), path (rds)
-    path validation_log
+    tuple val(meta), path (validation_log)
     val resolutions
-    val integration_method
+    val integration_tool
 
     output:
     tuple val(meta), path ("*_Clustered.rds"), emit: rds
-    path("*validation.log"),                   emit: log
+    tuple val(meta), path("*validation.log"),                   emit: log
+    path("markers")
+    path("*.pdf")
     //path ("versions.yml"),                   emit: versions
 
     when:
     task.ext.when == null || task.ext.when
 
     script:
-    def args = task.ext.args  ?: ''
+    if (meta.integrated) {
+        integration_method = integration_tool
+    } else {
+        integration_method = "NULL"
+    }
+    def args = task.ext.args ?: ''
     """
-    pcMax=\$(paste -s  <(grep PC $validation_log| grep -E -o "[0-9]") | sed 's|\t||')    
+    pcMax=\$(paste -s  <(grep PC $validation_log| grep -E -o "[0-9]") | sed 's|\\t||')
 
     FindNeighborsClustersMarkers.R \\
         $rds\\
         $resolutions \\
         \$pcMax \\
-        $integration \\
+        ${integration_method} \\
         ${meta.group} \\
-        ${args}      
-
-    //cat <<-END_VERSIONS > versions.yml
-    //"${task.process}":
-        //Seurat: \$(echo \$( version) | sed "s/, version //g" )
-    //END_VERSIONS
+        ${args}
     """
 
     stub:
@@ -49,4 +51,4 @@ process FIND_NN_CLUSTER {
     END_VERSIONS
     """
 
-} 
+}
diff --git a/modules/local/integration.nf b/modules/local/integration.nf
@@ -1,7 +1,7 @@
 process INTEGRATION {
 
-    tag "${meta.group}"
-    label 'process_medium'
+    tag "${meta}"
+    label 'process_small'
 
     conda "conda-forge::python=3.9.5"
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
@@ -11,10 +11,10 @@ process INTEGRATION {
     input:
     tuple val(meta), path (rds)
     val integration_method
-    
+
     output:
     tuple val(meta), path ("*_Integrated.rds"), emit: rds
-    path("*validation.log"),           emit: log
+    tuple val(meta), path("*validation.log"),           emit: log
     //path ("versions.yml"),            emit: versions
 
     when:
@@ -26,13 +26,9 @@ process INTEGRATION {
     Integration.R \\
         $rds \\
         $integration_method \\
-        ${meta.group} \\
-        ${args}      
+        ${meta} \\
+        ${args}
 
-    //cat <<-END_VERSIONS > versions.yml
-    //"${task.process}":
-        //Seurat: \$(echo \$( version) | sed "s/, version //g" )
-    //END_VERSIONS
     """
 
     stub:
@@ -43,4 +39,4 @@ process INTEGRATION {
     END_VERSIONS
     """
 
-} 
+}
diff --git a/modules/local/makeseurat.nf b/modules/local/makeseurat.nf
@@ -16,9 +16,9 @@ process MAKE_SEURAT {
     val gene_identifier
 
     output:
-    tuple val(meta), path ("*_SO.rds"), emit: rds
-    path("*.validation.log"),           emit: log
-    path ("versions.yml"),            emit: versions
+    tuple val(meta), path ("*_SO.rds"),        emit: rds
+    tuple val(meta), path("*_Validation.log"), emit: log
+    path ("versions.yml"),                     emit: versions
 
     when:
     task.ext.when == null || task.ext.when
@@ -34,11 +34,13 @@ process MAKE_SEURAT {
         $gene_identifier \\
         $genes_2_rm \\
         ${meta.id} \\
-        ${meta.group} \\
+        "${meta.groups}" \\
         $min_cells \\
         $min_features \\
         ${args}
 
+    perl -i -pe 's/"//g;s/\\[\\d\\d?\\d?\\] //g' fileName.log *_Validation.log
+
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
         Seurat: \$(echo \$( version) | sed "s/, version //g" )

diff --git a/modules/local/merge.nf b/modules/local/merge.nf
@@ -1,7 +1,7 @@
 process MERGE_SO {
 
-    tag "${meta.group}"
-    label 'process_medium'
+    tag "${meta}"
+    label 'process_small'
 
     conda "conda-forge::python=3.9.5"
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
@@ -15,23 +15,24 @@ process MERGE_SO {
     output:
     tuple val(meta), path ("*Merged_SO.rds"), emit: rds
     path("*validation.log"),           emit: log
+    path("*.pdf")
     //path ("versions.yml"),            emit: versions
 
     when:
     task.ext.when == null || task.ext.when
 
     script:
+    def n_features = task.ext.args2 ?: 'ALL'
+    def scale_method = task.ext.args3 ?: 'SCT'
     def args = task.ext.args  ?: ''
     """
     Merge.R \\
-        ${Meta.group} \\
+        ${meta} \\
         $vars_2_regress \\
-        ${args}      
+        $n_features \\
+        $scale_method \\
+        ${args}
 
-    //cat <<-END_VERSIONS > versions.yml
-    //"${task.process}":
-        //Seurat: \$(echo \$( version) | sed "s/, version //g" )
-    //END_VERSIONS
     """
 
     stub:
@@ -42,4 +43,4 @@ process MERGE_SO {
     END_VERSIONS
     """
 
-} 
+}
diff --git a/modules/local/normalize_qc.nf b/modules/local/normalize_qc.nf
@@ -1,7 +1,7 @@
 process NORMALIZE_QC {
 
     tag "${meta.id}"
-    label 'process_medium'
+    label 'process_small'
 
     conda "conda-forge::python=3.9.5"
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
@@ -16,18 +16,21 @@ process NORMALIZE_QC {
     val ncount_lower
     val ncount_upper
     val max_mito_pct
-    
+
 
     output:
-    tuple val(meta), path ("*_QC.rds"), emit: rds
-    path("*.validation.log"),           emit: log
+    tuple val(meta), path ("*_QC.rds"),        emit: rds
+    tuple val(meta), path("*.validation.log"), emit: log
     //path ("versions.yml"),            emit: versions
 
     when:
     task.ext.when == null || task.ext.when
 
     script:
     def args = task.ext.args  ?: ''
+    if (mito_genes.getClass().name == "nextflow.util.BlankSeparatedList") {
+        mito_genes = 'NULL'
+    }
     """
     Normalize_QC.R \\
         $mito_genes \\
@@ -38,12 +41,12 @@ process NORMALIZE_QC {
         $max_mito_pct \\
         $rds \\
         ${meta.id} \\
-        ${args}      
+        ${args}
 
-    //cat <<-END_VERSIONS > versions.yml
-    //"${task.process}":
-        //Seurat: \$(echo \$( version) | sed "s/, version //g" )
-    //END_VERSIONS
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        Seurat: \$(echo \$( version) | sed "s/, version //g" )
+    END_VERSIONS
     """
 
     stub:
@@ -54,4 +57,4 @@ process NORMALIZE_QC {
     END_VERSIONS
     """
 
-} 
+}