Patch a bug in obitag when some reference sequences have taxid absent from the taxonomy

pkg/obioptions/version.go

@@ -8,7 +8,7 @@ import (
 // corresponds to the last commit, and not the one when the file will be
 // commited
 
-var _Commit = "fd80249"
+var _Commit = "2d52322"
 var _Version = "Release 4.4.0"
 
 // Version returns the version of the obitools package.

pkg/obitax/taxon.go

@@ -27,7 +27,7 @@ type Taxon struct {
 // Returns:
 //   - A formatted string representing the Taxon in the form "taxonomy_code:taxon_id [scientific_name]".
 func (taxon *Taxon) String() string {
-	if taxon == nil {
+	if taxon == nil || taxon.Node == nil {
 		return "NA"
 	}
 	return taxon.Node.String(taxon.Taxonomy.code)

pkg/obitax/taxonslice.go

@@ -145,3 +145,18 @@ func (slice *TaxonSlice) Set(index int, taxon *Taxon) *TaxonSlice {
 
 	return slice
 }
+
+func (slice *TaxonSlice) Push(taxon *Taxon) *TaxonSlice {
+	if slice.taxonomy != taxon.Taxonomy {
+		log.Panic("Cannot add taxon from a different taxonomy")
+	}
+
+	slice.slice = append(slice.slice, taxon.Node)
+
+	return slice
+}
+
+func (slice *TaxonSlice) ReduceToSize(size int) *TaxonSlice {
+	slice.slice = slice.slice[:size]
+	return slice
+}

pkg/obitools/obitag/obitag.go

@@ -249,16 +249,16 @@ func CLIAssignTaxonomy(iterator obiiter.IBioSequence,
 		[]*obikmer.Table4mer,
 		len(references))
 
-	taxa := taxo.NewTaxonSlice(references.Len(), references.Len())
+	taxa := taxo.NewTaxonSlice(0, references.Len())
 	buffer := make([]byte, 0, 1000)
 
 	j := 0
 	for _, seq := range references {
-		references[j] = seq
-		refcounts[j] = obikmer.Count4Mer(seq, &buffer, nil)
 		taxon := seq.Taxon(taxo)
-		if taxon != nil {
-			taxa.Set(j, taxon)
+		if taxon != nil && taxon.Node != nil {
+			references[j] = seq
+			refcounts[j] = obikmer.Count4Mer(seq, &buffer, nil)
+			taxa.Push(taxon)
 			j++
 		} else {
 			obilog.Warnf("Taxid %s is not described in the taxonomy %s."+
@@ -267,6 +267,7 @@ func CLIAssignTaxonomy(iterator obiiter.IBioSequence,
 		}
 	}
 
+	log.Infof("%d reference sequences conserved on %d", j, len(references))
 	references = references[:j]
 	refcounts = refcounts[:j]