improved genome guessing

dge_workflow/dge_utils.sh

@@ -384,12 +384,23 @@ mysub "${project}__cuffdiff" "$cdCmd"  -q long -n 4 -R span[hosts=1] | blockScri
 tmpDbDir=$(mktemp -d)
 cp -r . $tmpDbDir
 
-## todo remove this hack
-genome=$(echo $gtfFile | cut -f8 -d'/' | tr '[:upper:]' '[:lower:]'); echo "genome is $genome"
+
+#genome=$(echo $gtfFile | cut -f8 -d'/' | tr '[:upper:]' '[:lower:]'); echo "genome is $genome"
 
 ## make sure to use temp-r to avoid file locking problems
 #export R_LIBS=/tmp/r_index
 
+
+genome=$(scala -e '
+val gtfFile = args(0); //val gtfFile="mm10_igenomes_pc.gtf"
+val pattern = "mm10|mm9|hg19|zv9".r
+println(pattern.findFirstIn(gtfFile).getOrElse(""))
+' $(readlink -f $gtfFile)
+)
+echo $genome
+
+
+
 echo '
 require(cummeRbund)
 dbDir=commandArgs(T)[1]

misc/guess_species_from_gtf.scalah

+#!/bin/sh
+exec scalas "$0" "$@"
+!#
+
+/** Work in progress: Guess the species from a gtf file. By name first and then by file content. A more simplistic approach is already implementd in dge_workflow/dge_utils.sh
+  */
+
+import java.io.File
+
+import scala.io.Source
+
+// http://alvinalexander.com/scala/scala-shell-script-command-line-arguments-args
+val gtfFile = args(1)
+//val gtfFile="mm10_igenomes_pc.gtf"
+
+val pattern = "mm10|mm9|h19|zv9".r
+val genomeByName = pattern.findFirstIn(gtfFile)
+
+if (genomeByName.isEmpty) {
+  System.exit(1)
+}
+
+def guessFromContent(gtfFile: File): Option[String] = {
+  //  Source.fromString(s"grep ENSMUSG $gtfFile | "!!).getLines().hasNext
+  //  Bash.evalCapture(s"grep ENSMUSG $gtfFile | wc -l")
+
+  if (Source.fromFile(gtfFile).getLines().exists(_.contains("ENSMUSG"))) return Some("mouse")
+  if (Source.fromFile(gtfFile).getLines().exists(_.contains("ENSCAFG"))) return Some("dog")
+
+  None
+}
+
+
+genomeByName.get match {
+  case "mm9" =>
+}
+println(genomeByName)
\ No newline at end of file