Add the --taxonomic-path option to obiannotate

Former-commit-id: 385309a1c4bc5ed33aeaafc63eedb9fc552f78a6

pkg/obitax/sequence_methods.go

@@ -2,6 +2,7 @@ package obitax
 
 import (
 	"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obiseq"
+	log "github.com/sirupsen/logrus"
 )
 
 // Setting the taxon at a given rank for a given sequence.
@@ -46,3 +47,22 @@ func (taxonomy *Taxonomy) SetGenus(sequence *obiseq.BioSequence) *TaxNode {
 func (taxonomy *Taxonomy) SetFamily(sequence *obiseq.BioSequence) *TaxNode {
 	return taxonomy.SetTaxonAtRank(sequence, "family")
 }
+
+func (taxonomy *Taxonomy) SetPath(sequence *obiseq.BioSequence) string {
+	taxid, err := taxonomy.Taxon(sequence.Taxid())
+
+	if err != nil {
+		log.Fatalf("Taxid %d not defined in the current taxonomy", sequence.Taxid())
+	}
+
+	path, err := taxid.Path()
+
+	if err != nil {
+		log.Fatalf("Taxonomy index error: %v", err)
+	}
+
+	tpath := path.String()
+	sequence.SetAttribute("taxonomic_path", tpath)
+
+	return tpath
+}

pkg/obitax/sequence_workers.go

@@ -51,3 +51,14 @@ func (taxonomy *Taxonomy) MakeSetFamilyWorker() obiseq.SeqWorker {
 
 	return w
 }
+
+func (taxonomy *Taxonomy) MakeSetPathWorker() obiseq.SeqWorker {
+
+	w := func(s *obiseq.BioSequence) *obiseq.BioSequence {
+		taxonomy.SetPath(s)
+		return s
+	}
+
+	return w
+
+}

pkg/obitax/taxonslice.go

@@ -1,5 +1,10 @@
 package obitax
 
+import (
+	"bytes"
+	"fmt"
+)
+
 type TaxonSlice []*TaxNode
 
 func (set *TaxonSlice) Get(i int) *TaxNode {
@@ -9,3 +14,25 @@ func (set *TaxonSlice) Get(i int) *TaxNode {
 func (set *TaxonSlice) Len() int {
 	return len(*set)
 }
+
+func (path *TaxonSlice) String() string {
+	var buffer bytes.Buffer
+
+	if len(*path) > 0 {
+		taxon := (*path)[len(*path)-1]
+		fmt.Fprintf(&buffer, "%d@%s@%s",
+			taxon.Taxid(),
+			taxon.ScientificName(),
+			taxon.Rank())
+
+		for i := len(*path) - 2; i >= 0; i-- {
+			taxon := (*path)[i]
+			fmt.Fprintf(&buffer, "|%d@%s@%s",
+				taxon.Taxid(),
+				taxon.ScientificName(),
+				taxon.Rank())
+		}
+	}
+
+	return buffer.String()
+}

pkg/obitools/obiannotate/obiannotate.go

@@ -256,6 +256,12 @@ func CLIAnnotationWorker() obiseq.SeqWorker {
 		annotator = annotator.ChainWorkers(w)
 	}
 
+	if CLISetTaxonomicPath() {
+		taxo := obigrep.CLILoadSelectedTaxonomy()
+		w := taxo.MakeSetPathWorker()
+		annotator = annotator.ChainWorkers(w)
+	}
+
 	if CLIHasAddLCA() {
 		taxo := obigrep.CLILoadSelectedTaxonomy()
 		w := obitax.AddLCAWorker(taxo, CLILCASlotName(), CLILCAThreshold())

pkg/obitools/obiannotate/options.go

@@ -31,6 +31,7 @@ var _lcaSlot = ""
 var _lcaError = 0.0
 var _setId = ""
 var _cut = ""
+var _taxonomicPath = false
 
 func SequenceAnnotationOptionSet(options *getoptions.GetOpt) {
 	// options.BoolVar(&_addRank, "seq-rank", _addRank,
@@ -113,6 +114,9 @@ func SequenceAnnotationOptionSet(options *getoptions.GetOpt) {
 		options.ArgName("RANK_NAME"),
 		options.Description("Adds taxonomic annotation at taxonomic rank <RANK_NAME>."))
 
+	options.BoolVar(&_taxonomicPath, "taxonomic-path", _taxonomicPath,
+		options.Description("Annotate the sequence with its taxonomic path"))
+
 	// options.StringVar(&_tagList, "tag-list", _tagList,
 	// 	options.ArgName("FILENAME"),
 	// 	options.Description("<FILENAME> points to a file containing attribute names"+
@@ -299,3 +303,7 @@ func CLIPatternError() int {
 func CLIPatternInDels() bool {
 	return _pattern_indel
 }
+
+func CLISetTaxonomicPath() bool {
+	return _taxonomicPath
+}