Change obiclean algorithm for a better evaluation of ratio

pkg/obitax/path.go

@@ -2,6 +2,8 @@ package obitax
 
 import (
 	"fmt"
+
+	log "github.com/sirupsen/logrus"
 )
 
 func (taxon *TaxNode) Path() (*TaxonSlice, error) {
@@ -22,6 +24,34 @@ func (taxon *TaxNode) Path() (*TaxonSlice, error) {
 	return &path, nil
 }
 
+func (taxon *TaxNode) TaxonAtRank(rank string) *TaxNode {
+	for taxon.rank != rank && taxon != taxon.pparent {
+		taxon = taxon.pparent
+
+		if taxon == nil {
+			log.Panicln("Taxonomy must be reindexed")
+		}
+	}
+
+	if taxon == taxon.pparent {
+		taxon = nil
+	}
+
+	return taxon
+}
+
+func (taxon *TaxNode) Species() *TaxNode {
+	return taxon.TaxonAtRank("species")
+}
+
+func (taxon *TaxNode) Genus() *TaxNode {
+	return taxon.TaxonAtRank("genus")
+}
+
+func (taxon *TaxNode) Family() *TaxNode {
+	return taxon.TaxonAtRank("family")
+}
+
 // Returns a TaxonSet listing the requested taxon and all
 // its ancestors in the taxonomy down to the root.
 func (taxonomy *Taxonomy) Path(taxid int) (*TaxonSlice, error) {

pkg/obitax/sequence_methods.go

@@ -0,0 +1,37 @@
+package obitax
+
+import (
+	"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiseq"
+)
+
+// Setting the taxon at a given rank for a given sequence.
+//
+// Two attributes are added to the sequence. One named by the rank name stores
+// the taxid, a second named by the rank name suffixed with '_name' contains the
+// Scientific name of the genus.
+// If the taxon at the given rank doesn't exist for the taxonomy annotation
+// of the sequence, nothing happens.
+func (taxonomy *Taxonomy) SetTaxonAtRank(sequence *obiseq.BioSequence, rank string) *TaxNode {
+	taxid := sequence.Taxid()
+	taxon, err := taxonomy.Taxon(taxid)
+	taxonAtRank := taxon.TaxonAtRank(rank)
+
+	if err == nil && taxonAtRank != nil {
+		sequence.SetAttribute(rank, taxonAtRank.taxid)
+		sequence.SetAttribute(rank+"_name", taxonAtRank.scientificname)
+	}
+
+	return taxonAtRank
+}
+
+func (taxonomy *Taxonomy) SetSpecies(sequence *obiseq.BioSequence) *TaxNode {
+	return taxonomy.SetTaxonAtRank(sequence, "species")
+}
+
+func (taxonomy *Taxonomy) SetGenus(sequence *obiseq.BioSequence) *TaxNode {
+	return taxonomy.SetTaxonAtRank(sequence, "genus")
+}
+
+func (taxonomy *Taxonomy) SetFamily(sequence *obiseq.BioSequence) *TaxNode {
+	return taxonomy.SetTaxonAtRank(sequence, "family")
+}

pkg/obitax/sequence_predicate.go

@@ -22,7 +22,10 @@ func (taxonomy *Taxonomy) IsAValidTaxon(withAutoCorrection ...bool) obiseq.Seque
 		if err == nil && taxon.taxid != taxid {
 			if autocorrection {
 				sequence.SetTaxid(taxon.taxid)
-				log.Printf("Sequence %s : Taxid %d updated with %d", taxid, taxon.taxid)
+				log.Printf("Sequence %s : Taxid %d updated with %d",
+					sequence.Id(),
+					taxid,
+					taxon.taxid)
 			} else {
 				if _, ok := deprecatedTaxidsWarning[taxid]; !ok {
 					deprecatedTaxidsWarning[taxid] = true

pkg/obitax/sequence_workers.go

@@ -0,0 +1,56 @@
+package obitax
+
+import (
+	"git.metabarcoding.org/lecasofts/go/obitools/pkg/goutils"
+	"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiiter"
+	"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiseq"
+	log "github.com/sirupsen/logrus"
+)
+
+func (taxonomy *Taxonomy) MakeSetTaxonAtRankWorker(rank string) obiiter.SeqWorker {
+
+	if !goutils.Contains(taxonomy.RankList(), rank) {
+		log.Fatalf("%s is not a valid rank (allowed ranks are %v)",
+			rank,
+			taxonomy.RankList())
+	}
+
+	w := func(sequence *obiseq.BioSequence) *obiseq.BioSequence {
+		taxonomy.SetTaxonAtRank(sequence, rank)
+		return sequence
+	}
+
+	return w
+}
+
+func (taxonomy *Taxonomy) MakeSetSpeciesWorker() obiiter.SeqWorker {
+
+	w := func(sequence *obiseq.BioSequence) *obiseq.BioSequence {
+		taxonomy.SetSpecies(sequence)
+		return sequence
+	}
+
+	return w
+}
+
+func (taxonomy *Taxonomy) MakeSetGenusWorker() obiiter.SeqWorker {
+
+	w := func(sequence *obiseq.BioSequence) *obiseq.BioSequence {
+		taxonomy.SetGenus(sequence)
+		return sequence
+	}
+
+	return w
+}
+
+func (taxonomy *Taxonomy) MakeSetFamilyWorker() obiiter.SeqWorker {
+
+	w := func(sequence *obiseq.BioSequence) *obiseq.BioSequence {
+		taxonomy.SetFamily(sequence)
+		return sequence
+	}
+
+	return w
+}
+
+