compute MAF and ref panel with W chr

compute_GW_reference_panel.nf

+
+/*
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+                           JASS suite pipeline: 
+This subworklow allows to compute genome wide reference panel
+authors : Hanna Julienne, Hervé Ménager & Lucie Troubat
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+*/
+nextflow.enable.dsl=2
+
+params.output_folder = "${launchDir}"
+
+params.ancestry="EUR"
+params.prefix="ALL_ensemble_1000G_hg38_EUR_chr"
+params.suffix=""
+
+chr_channel = Channel.from(1..22).mix(Channel.of("X"))
+params.ref_panel = '/pasteur/zeus/projets/p02/GGS_JASS/jass_analysis_pipeline/Ref_panel_by_chr/'
+
+ref_chr_channel=Channel.fromPath("/pasteur/zeus/projets/p02/GGS_WKD/DATA_1000G/Panels/EUR/ALL_ensemble_1000G_hg38_EUR_chr*.bim")
+ref_chr_channel3=Channel.fromPath("/pasteur/zeus/projets/p02/GGS_WKD/DATA_1000G/Panels/EUR/ALL_ensemble_1000G_hg38_EUR_chr*.*")
+
+/*****************************/
+/*   process inclusion   */
+/*****************************/
+include {Compute_MAF; Create_WG_reference_panel} from "./modules/GW_reference_panels"
+
+workflow{
+    /****** PREPROCESSING ******/
+    Compute_MAF(ref_chr_channel3.collect(), chr_channel)
+    Create_WG_reference_panel(Compute_MAF.out.collect(), ref_chr_channel.collect())
+}
\ No newline at end of file

modules/GW_reference_panels.nf

 
-/*
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-                           JASS suite pipeline
-authors : Hanna Julienne, Hervé Ménager & Lucie Troubat
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-*/
-
-/* Parameter to set if optional pipeline steps are performed */
-params.compute_project=false
-
-
-/* path of input data */
-params.ref_panel = '/pasteur/zeus/projets/p02/GGS_WKD/DATA_1000G/Panels/EAS/'
-
-params.region = '/pasteur/zeus/projets/p02/GGS_WKD/DATA_1000G/Panels/Regions_LD/EAS/All_Regions_ALL_ensemble_1000G_hg38_EAS.bed'
-params.output_folder = "${launchDir}"
-
-params.ancestry="EAS"
-params.prefix="ALL_ensemble_1000G_hg38_EAS_chr"
-params.prefix_Impute_GWAS="ALL_ensemble_1000G_hg38_EAS_"
-params.suffix=""
-
-
-chr_channel = Channel.from(1..22)
-
-ref_chr_channel=Channel.fromPath(params.ref_panel+"/ALL_ensemble_1000G_hg38_EAS_chr*.bim")
-ref_chr_channel2=Channel.fromPath(params.ref_panel+"/ALL_ensemble_1000G_hg38_EAS_chr*.bim")
-ref_chr_channel3=Channel.fromPath(params.ref_panel+"/ALL_ensemble_1000G_hg38_EAS_chr*.*")
-
-
 process Compute_MAF{
 
     input:
-        file ref_panel
+        path ref_panel
         val chr
     output:
-        file "*.frq"
+        path "*.frq", emit: freq
     """
     echo "Compute_MAF"
 
@@ -48,13 +18,13 @@ process Compute_MAF{
 }
 
 process Create_WG_reference_panel{
-    publishDir "${launchDir}/Ref_Panel", pattern: "*.csv", mode: 'copy'
+    publishDir "${params.output_folder}/Ref_Panel", pattern: "*.csv", mode: 'copy'
 
     input:
-        file maf_files
-        file chr_files
+        path maf_files
+        path chr_files
     output:
-        file "1000G_${params.ancestry}_0_01.csv"
+        path "1000G_${params.ancestry}_0_01.csv", emit: ref_panel
     """
     #!/usr/bin/env python3
     import subprocess as sub
@@ -65,7 +35,8 @@ process Create_WG_reference_panel{
     pref="${params.prefix}"
     suf="${params.suffix}"
     refchr_list = []
-    for chrom in range(1,23):
+    for chrom in [*range(1,23), "X"]:
+        print(chrom)
         fi = "{0}{1}{2}.bim".format(pref,chrom,suf)
         print(fi)
         print(type(fi))
@@ -80,11 +51,3 @@ process Create_WG_reference_panel{
     ref.loc[~(ref.A1+ref.A2).isin(["AT", 'TA','CG','GC'])][["CHR","SNP", "MAF", "pos", "A1", "A2"]].to_csv("1000G_${params.ancestry}_0_01.csv",  index=False, header=False, sep="\t")
     """
 }
-
-
-workflow {
-
-    Compute_MAF(ref_chr_channel3.collect(), chr_channel)
-    create_WG_reference_panel(Compute_MAF.out.collect(), ref_chr_channel.collect())
-
-}
\ No newline at end of file