From 7140ac8c62f8ad6c8d03496008a2df3fe43acdc9 Mon Sep 17 00:00:00 2001
From: Laurent Modolo <laurent.modolo@ens-lyon.fr>
Date: Wed, 25 Oct 2023 14:49:39 +0200
Subject: [PATCH] add kmermerclust_clust process and switch to v0.1.0 of
kmerclust
---
src/bin/kmerclust_boot.R | 9 +++++++--
src/main.nf | 1 +
src/modules/kmerclust.nf | 26 +++++++++++++-------------
3 files changed, 21 insertions(+), 15 deletions(-)
diff --git a/src/bin/kmerclust_boot.R b/src/bin/kmerclust_boot.R
index d5c9a96..f2ecc17 100644
--- a/src/bin/kmerclust_boot.R
+++ b/src/bin/kmerclust_boot.R
@@ -1,11 +1,16 @@
library(kmerclust)
+library(tidyverse)
args <- commandArgs(trailingOnly = TRUE)
print(args)
load(file = paste0(args[1], ".Rdata"))
res <- count %>%
dplyr::select(count_m, count_f) %>%
+ mutate(
+ count_m = round(count_m),
+ count_f = round(count_f)
+ ) %>%
+ dplyr::filter(count_m + count_f > 0) %>%
as.matrix() %>%
- poiss_compare_models(count, nboot = as.numeric(args[2]) * 2, frac = 1e-3, bootsize = 1, core = as.numeric(args[2]))
+ poiss_compare_models(count, nboot = as.numeric(args[2]) * 2, nbatch = 100, max_iter = 1e4, bootsize = 1, core = as.numeric(args[2]))
save(res, file = paste0(args[1], "_boot_", args[3], ".Rdata"))
-
diff --git a/src/main.nf b/src/main.nf
index dd2bba6..877692c 100644
--- a/src/main.nf
+++ b/src/main.nf
@@ -35,6 +35,7 @@ workflow {
COLLATEKMER(MERGEKMER.out.csv.map{it -> [it[0].specie, it[1]] }.groupTuple())
KMERCLUST_LOAD(COLLATEKMER.out.csv, params_csv.collect())
KMERCLUST_BOOT(Channel.of(1..params.bootstrap).combine(KMERCLUST_LOAD.out.rdata))
+ KMERCLUST_CLUST(Channel.of(["XY", "XO", "OO"]).combine(KMERCLUST_LOAD.out.rdata))
KMERCLUST_MERGE(KMERCLUST_BOOT.out.rdata.groupTuple())
KMERCLUST_PLOT(KMERCLUST_MERGE.out.rdata.mix(KMERCLUST_LOAD.out.rdata).groupTuple())
}
diff --git a/src/modules/kmerclust.nf b/src/modules/kmerclust.nf
index 61842d6..e8115b5 100644
--- a/src/modules/kmerclust.nf
+++ b/src/modules/kmerclust.nf
@@ -2,7 +2,7 @@ process KMERCLUST_LOAD {
tag "$specie"
label 'big_mem_mono_cpus'
- container "lbmc/kmerclust:0.0.7"
+ container "lbmc/kmerclust:0.1.0"
publishDir "results/${params.kmer_size}/${specie}/", mode: 'copy'
input:
@@ -20,7 +20,7 @@ process KMERCLUST_LOAD {
cat <<-END_VERSIONS > versions.yml
"${task.process}":
- Rkmerclust: 0.0.7
+ Rkmerclust: 0.1.0
END_VERSIONS
"""
}
@@ -29,7 +29,7 @@ process KMERCLUST_BOOT {
tag "$specie"
label 'big_mem_multi_cpus'
- container "lbmc/kmerclust:0.0.7"
+ container "lbmc/kmerclust:0.1.0"
publishDir "results/${params.kmer_size}/${specie}/", mode: 'copy'
input:
@@ -46,16 +46,16 @@ process KMERCLUST_BOOT {
cat <<-END_VERSIONS > versions.yml
"${task.process}":
- Rkmerclust: 0.0.7
+ Rkmerclust: 0.1.0
END_VERSIONS
"""
}
-process KMERCLUST {
+process KMERCLUST_CLUST {
tag "$specie"
- label 'big_mem_multi_cpus'
+ label 'small_mem_mono_cpus'
- container "lbmc/kmerclust:0.0.7"
+ container "lbmc/kmerclust:0.1.0"
publishDir "results/${params.kmer_size}/${specie}/", mode: 'copy'
input:
@@ -68,11 +68,11 @@ process KMERCLUST {
script:
def args = task.ext.args ?: ''
"""
- Rscript ${projectDir}/bin/kmerclust.R ${specie} ${task.cpus}
+ Rscript ${projectDir}/bin/kmerclust_clust.R ${specie} ${task.cpus} ${id}
cat <<-END_VERSIONS > versions.yml
"${task.process}":
- Rkmerclust: 0.0.7
+ Rkmerclust: 0.1.0
END_VERSIONS
"""
}
@@ -81,7 +81,7 @@ process KMERCLUST_MERGE {
tag "$specie"
label 'big_mem_mono_cpus'
- container "lbmc/kmerclust:0.0.7"
+ container "lbmc/kmerclust:0.1.0"
publishDir "results/${params.kmer_size}/${specie}/", mode: 'copy'
input:
@@ -98,7 +98,7 @@ process KMERCLUST_MERGE {
cat <<-END_VERSIONS > versions.yml
"${task.process}":
- Rkmerclust: 0.0.7
+ Rkmerclust: 0.1.0
END_VERSIONS
"""
}
@@ -107,7 +107,7 @@ process KMERCLUST_PLOT {
tag "$specie"
label 'big_mem_mono_cpus'
- container "lbmc/kmerclust:0.0.7"
+ container "lbmc/kmerclust:0.1.0"
publishDir "results/${params.kmer_size}/${specie}/", mode: 'copy'
input:
@@ -124,7 +124,7 @@ process KMERCLUST_PLOT {
cat <<-END_VERSIONS > versions.yml
"${task.process}":
- Rkmerclust: 0.0.7
+ Rkmerclust: 0.1.0
END_VERSIONS
"""
}
--
GitLab