Merge pull request #29 from dienerlab/fix/superkingdom

make compatible with new NCBI ranks

Author: cdiener

bin/fixk2report.R

@@ -3,21 +3,28 @@
 # Fixes missing ranks in Kraken2 reports
 
 library(data.table)
+library(futile.logger)
+
+domains <- c("Archaea", "Bacteria", "Eukaryota")
 
 args <- commandArgs(trailingOnly = TRUE)
 report <- fread(args[1], sep="\t", header=FALSE, strip.white=FALSE)
 
 fix <- function(name) {
     name <- trimws(name)
+
+    out <- "D1"
     if (name == "root") {
-        return("R")
+        out <- "R"
     } else if (name == "cellular organisms") {
-        return("R1")
-    } else {
-        return("D")
+        out <- "R1"
+    } else if (name %in% domains) {
+        out <- "D"
     }
+
+    flog.info("Mapping %s -> rank: %s", name, out)
+    return(out)
 }
 
-report[V4 == "" | V4 == "1", V4 := sapply(V6, fix)]
-report[V4 == "1", V4 := "D1"]
+report[V4 == "" | (V4 %chin% as.character(0:9)), V4 := sapply(V6, fix)]
 fwrite(report, args[2], sep="\t", quote=FALSE, col.names=FALSE)

medi.yml

@@ -23,5 +23,5 @@ dependencies:
   - r-futile.logger
   - bioconductor-biostrings
   - taxonkit
-  - architeuthis>=0.2.1
+  - architeuthis>=0.4.0
   - gxx