Create RSQ plot from results

environment.yml

@@ -8,9 +8,9 @@ channels:
 dependencies:
   - plink2
   - r=4.1.0
-  - r-rmarkdown=2.11
-  - r-ggplot2=3.3.5
-  - r-dplyr=1.0.7
+  - r-rmarkdown=2.16
+  - r-ggplot2=3.3.6
+  - r-dplyr=1.1.0
   - r-dt=0.19
   - r-qqman=0.1.8
   - r-htmltools=0.5.2
@@ -20,11 +20,12 @@ dependencies:
   - r-data.table=1.14.0
   - r-r.utils=2.10.1
   - r-ggrepel=0.9.1
-  - pandoc=2.14.2
+  - pandoc=2.19.2
   - cairo=1.16.0
   - xorg-libxt=1.2.1
   - fonts-anaconda=1
   - openjdk=11.0.9
   - unzip=6.0
   - bioconductor-ramwas=1.16.0
   - bedtools=2.30.0
+  - csvtk=0.26.0

files/rsq-report.Rmd

@@ -0,0 +1,52 @@
+---
+title: "Imputation Quality Report"
+output:
+  rmdformats::robobook:
+    self_contained: true
+    thumbnails: false
+    lightbox: true
+    gallery: false
+    highlight: tango
+params:
+  input: ""
+  name: ""
+  population: ""
+  version: ""
+  date: ""
+  service: ""
+---
+
+```{r setup, echo=FALSE, include=FALSE}
+
+library(ggplot2)
+library(dplyr)
+```
+
+## Parameters
+
+| Parameter        | Value                       |
+|------------------|-----------------------------|
+| Job             | `r params$name`            |
+| Pipeline Version | `r params$version`          |
+| Date             | `r params$date`             |
+
+## RSQ Plot
+```{r echo=FALSE}
+dataset = read.csv(params$input, row.names=NULL, sep = '\t')
+dataset_weighted = dataset %>%
+    group_by ( 
+        X.Bin.Aggregated.by.MAF 
+    ) %>%
+    summarise ( 
+        mean_MAF = mean(Average.MAF),
+        variants_sum = sum(X.Variants), 
+        weighted_R2 = weighted.mean(Imputation.R2,X.Variants)
+    )
+
+ggplot() + geom_line(data = dataset_weighted, mapping = aes(x = mean_MAF, y= weighted_R2))  +  scale_x_log10() +  expand_limits(x = 0, y = 0) 
+
+```
+
+<small>
+This report has been created with **`r params$service`**.
+</small>

modules/local/calculate_rsq.nf

@@ -1,6 +1,6 @@
 process CALCULATE_RSQ {
 
-    publishDir "${params.pubDir}/aggRSquare", mode: 'copy', pattern: '*aggRSquare'
+    publishDir "${params.pubDir}/rsq", mode: 'copy', pattern: '*aggRSquare'
 
     input:
     tuple val(chr), path(dosage_data), path(sequence_data)
@@ -11,5 +11,6 @@ process CALCULATE_RSQ {
 
     """
     aggRSquare -v ${sequence_data} -i ${dosage_data} -o ${sequence_data.baseName} --d
+    sed -i '/^##/d' ${sequence_data.baseName}.aggRSquare
     """
 }

modules/local/plot_rsq.nf

@@ -0,0 +1,29 @@
+process PLOT_RSQ {
+
+    publishDir "${params.pubDir}/rsq", mode: 'copy', pattern: '*html'
+
+    input:
+    path(agg_rsq)
+    path(rsq_report)
+
+    output:
+    path("*.html")
+
+    script:
+    """
+    csvtk concat -C '\$' -t -T $agg_rsq -o combined.txt
+
+    Rscript -e "require( 'rmarkdown' ); render('${rsq_report}',
+        params = list(
+            input = 'combined.txt',
+            name = '${params.project}',
+            version = paste('${workflow.manifest.name}', '(v${workflow.manifest.version})'),
+            date = '${params.project_date}'
+        ),
+        intermediates_dir='\$PWD',
+        knit_root_dir='\$PWD',
+        output_file='\$PWD/rsq-report.html'
+    )"
+    """
+
+}

nextflow.config

@@ -1,18 +1,20 @@
 manifest {
     name                                  = 'microarray-eval-nf'
-    version                               = '0.0.2'
-    description                           = 'Nextflow pipeline to eval microarray chips'
+    version                               = '0.1.0'
+    description                           = 'Nextflow pipeline to evaluate imputation quality'
     author                                = 'Sebastian Schönherr, Lukas Forer, Martin Eberle'
-    homePage                              = 'https://github.com/seppinho/exome-cnv-nf'
+    homePage                              = 'https://github.com/genepi/microarray-eval-nf'
     mainScript                            = 'main.nf'
     nextflowVersion                       = '!>=21.04.0'
 }
 
 // Global default params, used in configs 
 params {
 
+    project_date             = "`date`"
     //Required inputs
-    wf                       = null
+
+    workflow_name            = null
     project                  = null
     sequence_data            = null
     dosage_data              = null

workflows/rsq.nf

@@ -1,5 +1,6 @@
 include { LIFT_OVER } from '../modules/local/lift_over'
 include { CALCULATE_RSQ } from '../modules/local/calculate_rsq'
+include { PLOT_RSQ } from '../modules/local/plot_rsq'
 
 workflow RSQ {
 
@@ -25,7 +26,14 @@ workflow RSQ {
         r2_input_data_lifted = r2_input_data_combined
     }
 
-    CALCULATE_RSQ ( r2_input_data_lifted )
+    CALCULATE_RSQ ( 
+        r2_input_data_lifted 
+    )
+
+    PLOT_RSQ (
+        CALCULATE_RSQ.out.agg_rsquare.collect(),
+        file("$baseDir/files/rsq-report.Rmd", checkIfExists: true)
+    )
 
 }