split plot from the analysis

src/bin/kmerclust.R

@@ -8,15 +8,8 @@ data <- readr::read_tsv(args[2], show_col_types = FALSE)
 format(object.size(data), units = "Mb")
 annotation <- parse_annotation(args[1])
 count <- annotate_counts(annotation, data, args[3])
-save(count, file = paste0(args[3], ".Rdata"))
 
 res <- compare_models(count %>% dplyr::ungroup(), nboot = 100, bootsize = 0.1, core = as.numeric(args[4]))
+
+save(count, file = paste0(args[3], ".Rdata"))
 save(res, file = paste0(args[3], "_boot.Rdata"))
-res %>%
-    ggplot(aes(x = name, y = BIC)) +
-    geom_violin()
-ggsave(paste0(args[3], "_BIC.pdf"))
-res %>%
-    ggplot(aes(x = name, y = WSS_f / BSS)) +
-    geom_violin()
-ggsave(paste0(args[3], "_WSS_BSS.pdf"))

src/main.nf

@@ -8,8 +8,9 @@ params.kmer_size = 12
 include { SPLIT } from "./modules/split"
 include { FASTKMERS } from "./modules/fastkmers"
 include { MERGEKMER } from "./modules/mergekmer"
-include { COLLATEKMER} from "./modules/mergekmer"
-include { KMERCLUST} from "./modules/kmerclust"
+include { COLLATEKMER } from "./modules/mergekmer"
+include { KMERCLUST } from "./modules/kmerclust"
+include { KMERCLUST_PLOT } from "./modules/kmerclust"
 
 
 Channel.fromPath( file(params.csv) )
@@ -30,4 +31,5 @@ workflow {
   MERGEKMER(FASTKMERS.out.csv.groupTuple())
   COLLATEKMER(MERGEKMER.out.csv.map{it -> [it[0].specie, it[1]] }.groupTuple())
   KMERCLUST(COLLATEKMER.out.csv, params_csv.collect())
+  KMERCLUST_PLOT(KMERCLUST.out.rdata)
 }

src/modules/kmerclust.nf

@@ -10,7 +10,6 @@ process KMERCLUST {
     path params_csv
 
     output:
-    tuple val(specie), path("*.pdf"), emit: pdf 
     tuple val(specie), path("*.Rdata"), emit: rdata 
     path "versions.yml"           , emit: versions
 
@@ -25,3 +24,29 @@ process KMERCLUST {
     END_VERSIONS
     """
 }
+
+process KMERCLUST_PLOT {
+    tag "$specie"
+    label 'big_mem_mono_cpus'
+
+    container "lbmc/kmerclust:0.0.2"
+    publishDir "results/${params.kmer_size}/${specie}/", mode: 'copy'
+
+    input:
+    tuple val(specie), path(rdata)
+
+    output:
+    tuple val(specie), path("*.pdf"), emit: pdf 
+    path "versions.yml"           , emit: versions
+
+    script:
+    def args = task.ext.args ?: ''
+    """
+    Rscript ${projectDir}/bin/kmerclust_plot.R ${specie}
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        Rkmerclust: 0.0.1 
+    END_VERSIONS
+    """
+}