add yml and modify config

modules/local/CNAqc/main.nf

@@ -3,18 +3,17 @@ process CNAQC {
     container = 'docker://lvaleriani/cnaqc:version1.0'
 
     input:
-
     tuple val(meta), path(cna_rds), path(snv_rds)
 
     output:
+    tuple val(meta), path("*_qc.rds"),                                  emit: qc_rds
+    tuple val(meta), path("*_data_plot.rds"), path("*_qc_plot.rds"),    emit: plot_rds
+    tuple val(meta), path("*_data.pdf"),                                emit: plot_pdf_data
+    tuple val(meta), path("*_qc.pdf"),                                  emit: plot_pdf_qc
+    path "versions.yml",                                                emit: versions
 
-    tuple val(meta), path("*_qc.rds"), emit: qc_rds
-    tuple val(meta), path("*_data_plot.rds"), path("*_qc_plot.rds"), emit: plot_rds
-    tuple val(meta), path("*_data.pdf"), emit: plot_pdf_data
-    tuple val(meta), path("*_qc.pdf"), emit: plot_pdf_qc
 
     script:
-
     def args                                = task.ext.args
     def prefix                              = task.ext.prefix                                       ?: "${meta.id}"
     def matching_strategy                   = args!='' && args.matching_strategy                    ?  "$args.matching_strategy" : ""
@@ -39,8 +38,7 @@ process CNAQC {
 
     """
     #!/usr/bin/env Rscript
-
-    library(tidyverse)
+    library(dplyr)
     library(CNAqc)
 
     SNV = readRDS("$snv_rds")
@@ -92,7 +90,6 @@ process CNAQC {
         nrow = 2
     )
 
-
     pl_exp = patchwork::wrap_plots(
         CNAqc::plot_gw_counts(tmp_x),
         CNAqc::plot_gw_vaf(tmp_x, N = 10000),
@@ -119,5 +116,15 @@ process CNAQC {
 
     ggplot2::ggsave(plot = pl_exp, filename = paste0("$prefix", "_data.pdf"), width = 210, height = 297, units="mm", dpi = 200)
     ggplot2::ggsave(plot = pl_qc, filename = paste0("$prefix", "_qc.pdf"), width = 210, height = 297, units="mm", dpi = 200)
+
+    # version export
+    f <- file("versions.yml","w")
+    dplyr_version <- sessionInfo()\$otherPkgs\$dplyr\$Version
+    cnaqc_version <- sessionInfo()\$otherPkgs\$CNAqc\$Version
+    writeLines(paste0('"', "$task.process", '"', ":"), f)
+    writeLines(paste("    CNAqc:", cnaqc_version), f)
+    writeLines(paste("    dplyr:", dplyr_version), f)
+    close(f)
+
     """
 }

modules/local/CNAqc2tsv/main.nf

@@ -7,25 +7,24 @@ process RDS_PROCESSING {
     container = 'docker://lvaleriani/cnaqc:version1.0'
 
     input:
-        tuple val(meta), path(rds_join), val(tumour_samples)
+    tuple val(meta), path(rds_join), val(tumour_samples)
 
     output:
-        tuple val(meta), path("*_joint_table.tsv"), val(tumour_samples), emit: tsv
+    tuple val(meta), path("*_joint_table.tsv"), val(tumour_samples),    emit: tsv
+    path "versions.yml",                                                emit: versions
+
 
     script:
-    def args                              = task.ext.args                                 ?: ''
-    def prefix                              = task.ext.prefix                                       ?: "${meta.id}"
+    def args    = task.ext.args     ?:  ''
+    def prefix  = task.ext.prefix   ?:  "${meta.id}"
 
     """
     #!/usr/bin/env Rscript
-
     library(dplyr)
     library(CNAqc)
-    library(tidyr)
 
     source("$moduleDir/utils.R")
 
-
     multi_cnaqc = readRDS(file = "$rds_join")
     mutations_multisample <- get_sample(m_cnaqc_obj = multi_cnaqc,sample = get_sample_name(multi_cnaqc),
                                     which_obj = "original")
@@ -41,5 +40,14 @@ process RDS_PROCESSING {
 
     write.table(joint_table, file = paste0("$prefix","_joint_table.tsv"), append = F, quote = F, sep = "\t", row.names = FALSE)
 
+        # version export
+    f <- file("versions.yml","w")
+    dplyr_version <- sessionInfo()\$otherPkgs\$dplyr\$Version
+    cnaqc_version <- sessionInfo()\$otherPkgs\$CNAqc\$Version
+    writeLines(paste0('"', "$task.process", '"', ":"), f)
+    writeLines(paste("    dplyr:", dplyr_version), f)
+    writeLines(paste("    CNAqc:", cnaqc_version), f)
+    close(f)
+
     """
 }

modules/local/annotate_driver/main.nf

@@ -4,23 +4,22 @@ process ANNOTATE_DRIVER {
     container = 'docker://lvaleriani/cnaqc:version1.0'
 
     input:
-
     tuple val(meta), path(rds), path(driver_list)
 
     output:
-    tuple val(meta), path("*.rds"), emit: rds
+    tuple val(meta), path("*.rds"),     emit: rds
+    path "versions.yml",                emit: versions
 
 
     script:
-
-    def args = task.ext.args ?: ''
-    def prefix = task.ext.prefix ?: "${meta.id}"
+    def args    =   task.ext.args   ?: ''
+    def prefix  =   task.ext.prefix ?: "${meta.id}"
 
     """
     #!/usr/bin/env Rscript
-
     library(dplyr)
     library(readr)
+    library(tidyr)
 
     data = readRDS("$rds")
     SNV = data[["$meta.tumour_sample"]]
@@ -66,8 +65,19 @@ process ANNOTATE_DRIVER {
         select(-tmp_s2) %>%
         mutate(is_driver = ifelse(is.na(is_driver), FALSE, is_driver))
 
-
     data[["$meta.tumour_sample"]]\$mutations = x
     saveRDS(object = data, file = paste0("$prefix", "_driver.rds"))
+
+    # version export
+    f <- file("versions.yml","w")
+    readr_version <- sessionInfo()\$otherPkgs\$readr\$Version
+    dplyr_version <- sessionInfo()\$otherPkgs\$dplyr\$Version
+    tidyr_version <- sessionInfo()\$otherPkgs\$tidyr\$Version
+    writeLines(paste0('"', "$task.process", '"', ":"), f)
+    writeLines(paste("    readr:", readr_version), f)
+    writeLines(paste("    dplyr:", dplyr_version), f)
+    writeLines(paste("    tidyr:", tidyr_version), f)
+    close(f)
+
     """
 }

modules/local/cna2CNAqc/main.nf

@@ -4,17 +4,18 @@ process CNA_PROCESSING {
     container "docker://lvaleriani/cnaqc:version1.0"
 
     input:
-        tuple val(meta), path(cna_segs), path(cna_extra)
+    tuple val(meta), path(cna_segs), path(cna_extra)
 
     output:
-        tuple val(meta), path("*.rds"), emit: rds
+    tuple val(meta), path("*.rds"),                            emit: rds
+    path "versions.yml",                                       emit: versions
 
     when:
-        task.ext.when == null || task.ext.when
+    task.ext.when == null || task.ext.when
 
     script:
-    def args = task.ext.args ?: ''
-    def prefix = task.ext.prefix ?: "${meta.id}"
+    def args    = task.ext.args     ?:  ''
+    def prefix  = task.ext.prefix   ?:  "${meta.id}"
 
     """
     #!/usr/bin/env Rscript
@@ -31,5 +32,15 @@ process CNA_PROCESSING {
     }
 
     saveRDS(object = CNA, file = paste0("$prefix", "_cna.rds"))
+
+    # version export
+    f <- file("versions.yml","w")
+    readr_version <- sessionInfo()\$otherPkgs\$readr\$Version
+    dplyr_version <- sessionInfo()\$otherPkgs\$dplyr\$Version
+    writeLines(paste0('"', "$task.process", '"', ":"), f)
+    writeLines(paste("    readr:", readr_version), f)
+    writeLines(paste("    dplyr:", dplyr_version), f)
+    close(f)
+
     """
 }

modules/local/cna2CNAqc/parser_CNA.R

@@ -1,18 +1,16 @@
-library(tidyverse)
+library(dplyr)
+library(readr)
 
 parse_Sequenza = function(segments_file, extra_file){
-    #segments_file = paste0(run, '/', sample, '.smoothedSegs.txt') #_segments.txt')
-    #purity_file = paste0(run, '/', sample, '_confints_CP.txt')
-
     # Extract the segments information
     segments = readr::read_tsv(segments_file, col_types = readr::cols()) %>%
-                dplyr::rename(
-                        chr = chromosome,
-                        from = start.pos,
-                        to = end.pos,
-                        Major = A,
-                        minor = B) %>%
-                        dplyr::select(chr, from, to, Major, minor, dplyr::everything())
+        dplyr::rename(
+            chr = chromosome,
+            from = start.pos,
+            to = end.pos,
+            Major = A,
+            minor = B) %>%
+        dplyr::select(chr, from, to, Major, minor, dplyr::everything())
 
     solutions = readr::read_tsv(extra_file, col_types = readr::cols())
     purity = solutions$cellularity[2]
@@ -22,21 +20,18 @@ parse_Sequenza = function(segments_file, extra_file){
 
 
 parse_ASCAT = function(segments_file, extra_file){
-    #segments_file = paste0(run, '/', sample, '.segments.txt')
-    #purity_file = paste0(run, '/', sample, '.purityploidy.txt')
-
     # Extract the segments information
     segments = readr::read_tsv(segments_file, col_types = readr::cols()) %>%
         dplyr::mutate(chr = paste0("chr",chr)) %>%
         dplyr::rename(
-        from = startpos,
-        to = endpos,
-        Major = nMajor,
-        minor = nMinor) %>%
+            from = startpos,
+            to = endpos,
+            Major = nMajor,
+            minor = nMinor) %>%
         dplyr::select(chr, from, to, Major, minor)
+
     solutions = readr::read_tsv(extra_file, col_types = readr::cols())
     purity = solutions$AberrantCellFraction
     ploidy = solutions$Ploidy
-
     return(list(segments = segments, purity = purity, ploidy = ploidy))
 }

modules/local/get_positions/main.nf

@@ -6,23 +6,31 @@ process GET_POSITIONS_ALL {
     tuple val(meta), path(rds_list, stageAs: '*.rds')
 
     output:
+    tuple val(meta), path("*_all_positions.rds"),   emit: all_pos
+    path "versions.yml",                            emit: versions
 
-    tuple val(meta), path("*_all_positions.rds"), emit: all_pos
 
     script:
-    def args = task.ext.args ?: ''
-    def prefix = task.ext.prefix ?: "${meta.id}"
+    def args    = task.ext.args     ?:  ''
+    def prefix  = task.ext.prefix   ?:  "${meta.id}"
 
     """
     #!/usr/bin/env Rscript
+    library(dplyr)
 
-    library(tidyverse)
     positions = lapply(strsplit("$rds_list", " ")[[1]], FUN = function(rds){
         df = readRDS(rds)
         df = df[[1]]\$mutations %>% dplyr::mutate(id = paste(chr, from, to, sep = ":")) %>% dplyr::select(chr, from, to, ref, alt, id)
     })
     all = positions %>% dplyr::bind_rows() %>% dplyr::distinct() %>% dplyr::select(-id)
     saveRDS(object = all, file = paste0("$prefix", "_all_positions.rds"))
-    #write.table(file = paste0("$prefix", "_all_positions.bed"), all, quote = F, sep = "\t", row.names = F, col.names = T)
+
+    # version export
+    f <- file("versions.yml","w")
+    dplyr_version <- sessionInfo()\$otherPkgs\$dplyr\$Version
+    writeLines(paste0('"', "$task.process", '"', ":"), f)
+    writeLines(paste("    dplyr:", dplyr_version), f)
+    close(f)
+
     """
 }

modules/local/get_positions/main_rel.nf

@@ -3,19 +3,20 @@ process GET_POSITIONS_REL {
     container = 'docker://lvaleriani/cnaqc:dev1'
 
     input:
-        tuple val(meta), path(rds), path(all_pos)
+    tuple val(meta), path(rds), path(all_pos)
 
     output:
-        tuple val(meta), path("*_positions_missing"), emit: bed
+    tuple val(meta), path("*_positions_missing"),   emit: bed
+    path "versions.yml",                            emit: versions
+
 
     script:
-    def args = task.ext.args ?: ''
-    def prefix = task.ext.prefix ?: "${meta.id}"
+    def args    = task.ext.args     ?:  ''
+    def prefix  = task.ext.prefix   ?:  "${meta.id}"
 
     """
     #!/usr/bin/env Rscript
-
-    library(tidyverse)
+    library(dplyr)
 
     all_positions = readRDS("$all_pos") %>%
                 dplyr::mutate(id = paste(chr, from, to, sep = ":"))
@@ -27,8 +28,16 @@ process GET_POSITIONS_REL {
 
     missed = all_positions %>% filter(!(id %in%  positions\$id)) %>%
             dplyr::filter(chr %in% c(paste0('chr', seq(1,22)), 'chrX', 'chrY')) %>%
-            select(chr, from)
+            dplyr::select(chr, from)
 
     write.table(file = paste0("$meta.tumour_sample", "_positions_missing"), missed, quote = F, sep = "\t", row.names = F, col.names = F)
+
+    # version export
+    f <- file("versions.yml","w")
+    dplyr_version <- sessionInfo()\$otherPkgs\$dplyr\$Version
+    writeLines(paste0('"', "$task.process", '"', ":"), f)
+    writeLines(paste("    dplyr:", dplyr_version), f)
+    close(f)
+
     """
 }

modules/local/join_CNAqc/main.nf

@@ -3,15 +3,14 @@ process JOIN_CNAQC {
     container = 'docker://lvaleriani/cnaqc:version1.0'
 
     input:
-        tuple val(meta), path(rds_list, stageAs: '*.rds'), val(tumour_samples)
+    tuple val(meta), path(rds_list, stageAs: '*.rds'), val(tumour_samples)
 
     output:
-
-        tuple val(meta), path("*ALL.rds"), val(tumour_samples), emit: rds_all, optional: true
-        tuple val(meta), path("*PASS.rds"), val(tumour_samples), emit: rds_pass, optional: true
+    tuple val(meta), path("*ALL.rds"), val(tumour_samples),     emit: rds_all,  optional: true
+    tuple val(meta), path("*PASS.rds"), val(tumour_samples),    emit: rds_pass, optional: true
+    path "versions.yml",                                        emit: versions
 
     script:
-
     def args                                = task.ext.args
     def prefix                              = task.ext.prefix                                       ?: "${meta.id}"                                      ?: ''
     def qc_filter                           = args!='' && args.qc_filter                            ?  "$args.qc_filter" : ""
@@ -20,7 +19,7 @@ process JOIN_CNAQC {
     """
     #!/usr/bin/env Rscript
 
-    library(tidyverse)
+    library(dplyr)
     library(CNAqc)
 
     samples = substr("$tumour_samples", 2, nchar("$tumour_samples")-1)
@@ -47,5 +46,15 @@ process JOIN_CNAQC {
 
     saveRDS(object = out_all, file = paste0("$prefix", "_multi_cnaqc_ALL.rds"))
     saveRDS(object = out_PASS, file = paste0("$prefix", "_multi_cnaqc_PASS.rds"))
+
+    # version export
+    f <- file("versions.yml","w")
+    dplyr_version <- sessionInfo()\$otherPkgs\$dplyr\$Version
+    cnaqc_version <- sessionInfo()\$otherPkgs\$CNAqc\$Version
+    writeLines(paste0('"', "$task.process", '"', ":"), f)
+    writeLines(paste("    CNAqc:", cnaqc_version), f)
+    writeLines(paste("    dplyr:", dplyr_version), f)
+    close(f)
+
     """
 }