Handle GAF lines missing trailing tabs. Fixes #11. (#12)

geneontology · Jul 9, 2019 · 088b860 · 088b860
1 parent 813430f
commit 088b860
Show file tree

Hide file tree

Showing 2 changed files with 19 additions and 10 deletions.
diff --git a/src/main/scala/org/geneontology/gaferencer/Gaferencer.scala b/src/main/scala/org/geneontology/gaferencer/Gaferencer.scala
@@ -102,16 +102,19 @@ object Gaferencer extends LazyLogging {
 
   def processLine(line: String, propertyByName: Map[String, OWLObjectProperty], curieUtil: MultiCurieUtil): Set[(TermWithTaxon, ExtendedAnnotation)] = {
     val items = line.split("\t", -1)
-    val maybeTaxon = items(12).split(raw"\|", -1).headOption.map(_.trim.replaceAllLiterally("taxon:", TaxonPrefix)).map(Class(_))
-    if (maybeTaxon.isEmpty) logger.warn(s"Skipping row with badly formatted taxon: ${items(12)}")
-    val aspect = items(8).trim
-    val relation = AspectToGAFRelation(aspect)
-    val term = Class(items(4).trim.replaceAllLiterally("GO:", GOPrefix))
-    (for {
-      taxon <- maybeTaxon.toIterable
-      conjunction <- items(15).split("\\|", -1)
-      links = conjunction.split(",", -1).toSet[String].map(_.trim).flatMap(parseLink(_, propertyByName, curieUtil).toSet)
-    } yield (TermWithTaxon(term, taxon), ExtendedAnnotation(Link(relation, term), taxon, links))).toSet
+    if (items.size < 13) Set.empty
+    else {
+      val maybeTaxon = items(12).split(raw"\|", -1).headOption.map(_.trim.replaceAllLiterally("taxon:", TaxonPrefix)).map(Class(_))
+      if (maybeTaxon.isEmpty) logger.warn(s"Skipping row with badly formatted taxon: ${items(12)}")
+      val aspect = items(8).trim
+      val relation = AspectToGAFRelation(aspect)
+      val term = Class(items(4).trim.replaceAllLiterally("GO:", GOPrefix))
+      (for {
+        taxon <- maybeTaxon.toIterable
+        conjunction <- if (items.size > 15) items(15).split("\\|", -1) else Array("")
+        links = conjunction.split(",", -1).toSet[String].map(_.trim).flatMap(parseLink(_, propertyByName, curieUtil).toSet)
+      } yield (TermWithTaxon(term, taxon), ExtendedAnnotation(Link(relation, term), taxon, links))).toSet
+    }
   }
 
   def parseLink(text: String, propertyByName: Map[String, OWLObjectProperty], curieUtil: MultiCurieUtil): Option[Link] = text match {

diff --git a/src/test/scala/org/geneontology/gaferencer/TestParsing.scala b/src/test/scala/org/geneontology/gaferencer/TestParsing.scala
@@ -22,6 +22,8 @@ object TestParsing extends TestSuite {
       val line2 = "FOO\tFOO:2\tfoo2\t\tGO:0006412\tTEST:1\tIDA\t\tP\t\t\tgene\ttaxon:10090\t20100209\tFOO\tregulates(GO:0051594)\t"
       val line3 = "FOO\tFOO:2\tfoo2\t\tGO:0006412\tTEST:1\tIDA\t\tP\t\t\tgene\ttaxon:10090\t20100209\tFOO\tregulates(GO:0051594),occurs_in(GO:0005739)\t"
       val line4 = "FOO\tFOO:2\tfoo2\t\tGO:0006412\tTEST:1\tIDA\t\tP\t\t\tgene\ttaxon:10090\t20100209\tFOO\tregulates(GO:0051594),occurs_in(GO:0005739)|occurs_in(GO:0005739),negatively_regulates(GO:0051594)\t"
+      val line5 = "FOO\tFOO:1\tfoo1\t\tGO:0006412\tTEST:1\tIDA\t\tP\t\t\tgene\ttaxon:10090\t20100209\tFOO"
+      val line6 = "FOO\tFOO:1\tfoo1\t\tGO:0006412"
       val properties = Map(
         "regulates" -> Regulates,
         "negatively_regulates" -> NegativelyRegulates,
@@ -41,6 +43,10 @@ object TestParsing extends TestSuite {
       assert(res4.size == 2)
       assert(res4(TermWithTaxon(GOTerm, Mouse) -> ExtendedAnnotation(Link(relation, GOTerm), Mouse, Set(Link(Regulates, Class("http://purl.obolibrary.org/obo/GO_0051594")), Link(OccursIn, Class("http://purl.obolibrary.org/obo/GO_0005739"))))))
       assert(res4(TermWithTaxon(GOTerm, Mouse) -> ExtendedAnnotation(Link(relation, GOTerm), Mouse, Set(Link(NegativelyRegulates, Class("http://purl.obolibrary.org/obo/GO_0051594")), Link(OccursIn, Class("http://purl.obolibrary.org/obo/GO_0005739"))))))
+      val res5 = Gaferencer.processLine(line5, properties, cu)
+      assert(res5.size == 1)
+      val res6 = Gaferencer.processLine(line6, properties, cu)
+      assert(res6.size == 0)
     }
 
   }