Merge branch 'feature/make_cff' of github.com:mskcc/forte into featur…

…e/make_cff
mskcc · Aug 14, 2023 · f81c807 · f81c807
2 parents 694f369 + 301b395
commit f81c807
Show file tree

Hide file tree

Showing 4 changed files with 30 additions and 4 deletions.
diff --git a/bin/add_annotations_cff.R b/bin/add_annotations_cff.R
@@ -17,7 +17,7 @@ usage <- function() {
 args = commandArgs(TRUE)
 
 if (is.null(args) | length(args)<1) {
-        usage()
+    usage()
     quit()
 }
 
@@ -58,9 +58,12 @@ cff_file = args_opt$cff
 out_prefix = args_opt$out_prefix
 
 cff = fread(cff_file)
+final_cff_cols <- c(names(cff))
 agfusion_tab = fread(agfusion_file) %>% select(c(`5'_transcript`,`3'_transcript`,`5'_breakpoint`,`3'_breakpoint`,Fusion_effect))
+final_cff_cols <- c(final_cff_cols,"Fusion_effect")
 if (!is.null(oncokb_file)){
     oncokb_tab = fread(oncokb_file) %>% select(-Fusion)
+    final_cff_cols = c(final_cff_cols,names(oncokb_tab %>% select(-Tumor_Sample_Barcode)))
     cff <- merge(
         cff,
         oncokb_tab,
@@ -71,7 +74,6 @@ if (!is.null(oncokb_file)){
     )
 }
 
-
 cff <- merge(
     cff,
     agfusion_tab,
@@ -81,11 +83,15 @@ cff <- merge(
     all.y = T
 )
 
+cff <- as.data.frame(cff)[,c(final_cff_cols)]
+#cff <- cff %>% mutate(!!final_cff_cols[34] := Fusion_effect) %>% select(-c(Fusion_effect))
+
 write.table(
     cff,
     paste0(out_prefix, ".cff"),
     row.names = F,
     quote = F,
-    sep = "\t"
+    sep = "\t",
+    col.names = ! "V1" %in% final_cff_cols
 )
 
diff --git a/conf/igenomes.config b/conf/igenomes.config
@@ -56,7 +56,7 @@ params {
             arriba_blacklist       = "/usr/local/var/lib/arriba/blacklist_hg19_hs37d5_GRCh37_v2.3.0.tsv.gz"
             arriba_known_fusions   = "/usr/local/var/lib/arriba/known_fusions_hg19_hs37d5_GRCh37_v2.3.0.tsv.gz"
             arriba_protein_domains = "/usr/local/var/lib/arriba/protein_domains_hg19_hs37d5_GRCh37_v2.3.0.gff3"
-            cdna           = "https://ftp.ensembl.org/pub/release-75/fasta/homo_sapiens/cdna/Homo_sapiens.GRCh37.75.cdna.all.fa.gz"
+            cdna           = "http://ftp.ensemblgenomes.org/pub/viruses/fasta/sars_cov_2/cdna/Sars_cov_2.ASM985889v3.cdna.all.fa.gz"
             metafusion_blocklist = "https://raw.githubusercontent.com/anoronh4/forte-references/main/GRCh37_test/blocklist_breakpoints.bedpe"
             metafusion_gene_bed  = "https://raw.githubusercontent.com/anoronh4/forte-references/main/GRCh37_test/v75_gene.bed"
             metafusion_gene_info = "https://raw.githubusercontent.com/anoronh4/forte-references/main/GRCh37_test/gene_info_20230714.txt"

diff --git a/conf/modules.config b/conf/modules.config
@@ -211,6 +211,22 @@ process {
         ext.when = params.run_oncokb_fusionannotator
         secret = ["ONCOKB_TOKEN"]
         ext.args = "-b \$ONCOKB_TOKEN"
+        publishDir = [
+            path: { "${params.outdir}/analysis/${meta.id}/oncokb" },
+            mode: params.publish_dir_mode,
+            pattern: '*.oncokb.tsv',
+            enabled: false
+        ]
+    }
+
+    withName: CFF_FINALIZE {
+        ext.prefix = { "${meta.id}.final" }
+        publishDir = [
+            path: { "${params.outdir}/analysis/${meta.id}/metafusion" },
+            mode: params.publish_dir_mode,
+            pattern: '*.final.cff',
+            enabled: true
+        ]
     }
 
     withName: STAR_FOR_ARRIBA {

diff --git a/tests/nextflow.config b/tests/nextflow.config
@@ -11,6 +11,10 @@ process {
     cpus = 2
     memory = 3.GB
     time = 2.h
+
+    withName: '.*_TO_CFF|GUNZIP_METAFUSION.*|AGFUSION.*' {
+        ext.when = { ! (workflow.profile.toString().split(",").contains(["singularity"]) && workflow.profile.toString().split(",").contains(["test"])) }
+    }
 }
 
 if ("$PROFILE" == "singularity") {