correctly generating JSON data for annotation result charts

Author: aw-watson

workflows/Nextflow/configs/container.config

@@ -47,7 +47,7 @@ process {
         container = 'ghcr.io/lanl-bioinformatics/edge_contigs_taxonomy:1.12.0'
     }
     withLabel: 'annotation' {
-        container = 'ghcr.io/lanl-bioinformatics/edge_annotation:0.6.0'
+        container = 'apwat/annotation:0.6.1'
     }
     withLabel: 'phageFinder' {
         container = 'ghcr.io/lanl-bioinformatics/edge_phage_finder:1.2.0'

workflows/Nextflow/modules/runAnnotation/Dockerfile

@@ -1,6 +1,6 @@
 # syntax=docker/dockerfile:1
 FROM continuumio/miniconda3:24.5.0-0 AS build
-ENV version=0.6.0
+ENV version=0.6.1
 
 ENV container=docker
 
@@ -14,7 +14,7 @@ RUN conda init bash \
     && conda activate annotation 
 
 RUN conda install -n annotation -c bioconda perl-lwp-protocol-https
-RUN conda install -n annotation -c conda-forge r-base
+RUN conda install -n annotation -c conda-forge r-base r-jsonlite=2.0.0
 RUN conda install -y -n annotation -c bioconda prokka
 RUN conda install -n annotation -c bioconda mummer
 RUN conda install -n annotation -c bioconda blast=2.16

workflows/Nextflow/modules/runAnnotation/resources/usr/bin/plot_gff3_stats.pl

@@ -213,6 +213,7 @@ sub Rplot {
     $outcmd = "png('${prefix}_plots.png',height=800, width=1000, type='cairo');" if $outfmt =~ /png/i;
 
     my $Rscript = "
+library(jsonlite)
 setwd('$cwd');
 $outcmd
 #page setup
@@ -240,6 +241,11 @@ sub Rplot {
 barc = 'grey'
 #feature plot
 table.feat<-table(gff\$feature)
+features <- as.data.frame(table.feat)
+colnames(features)[1] <- \"Feature\"
+exportJSON <- toJSON(features, pretty=TRUE)
+write(exportJSON, \"feature_count.json\")
+
 barplot(table.feat,ylab=expression('Count  log'[10]), ylim=c(1,10000), col=barc, log='y', las=3, main='Feature count')
 #protein size distribution
 dist<-hist(gff\$length/3,
@@ -249,6 +255,14 @@ sub Rplot {
          breaks=seq(0, max(gff\$length)/3+100, 100), 
          xlim=c(0,max(gff\$length)/3+400),
          probability=FALSE, col=barc, border='black', cex=0.8 )
+size_dist <- as.data.frame(dist\$breaks[1:length(dist\$breaks)-1])
+size_dist[2] <- dist\$breaks[2:length(dist\$breaks)]
+size_dist[3] <- dist\$counts
+colnames(size_dist)[1] <- \"LowerBinThreshold\"
+colnames(size_dist)[2] <- \"UpperBinThreshold\"
+colnames(size_dist)[3] <- \"Count\"
+exportJSON <- toJSON(size_dist, pretty=TRUE)
+write(exportJSON, \"size_distribution.json\")
 #detail info
 text(max(gff\$length)/3/2,max(dist\$counts),adj=c(0,1),cex=0.8,'$text')
 title('ANNOTATION STATS - $title ($ctg_num contigs, ${length}bp)', outer=TRUE)

workflows/Nextflow/modules/runAnnotation/runAnnotation.nf

@@ -140,6 +140,7 @@ process annPlot {
     output:
     path "plot_gff3.log"
     path "annotation_stats_plots.pdf", emit: annStats
+    path "{feature_count,size_distribution}.json"
     path "${settings["projName"]}.txt", optional:true
 
     script:

workflows/Nextflow/nextflow.config

@@ -301,7 +301,7 @@ process {
         container = 'ghcr.io/lanl-bioinformatics/edge_contigs_taxonomy:1.12.0'
     }
     withLabel: 'annotation' {
-        container = 'ghcr.io/lanl-bioinformatics/edge_annotation:0.6.0'
+        container = 'apwat/annotation:0.6.1'
     }
     withLabel: 'phageFinder' {
         container = 'ghcr.io/lanl-bioinformatics/edge_phage_finder:1.2.0'