Add a first version of obitag the successor of ecotag

pkg/obikmer/counting.go

@@ -0,0 +1,76 @@
+package obikmer
+
+import (
+	"math"
+
+	"git.metabarcoding.org/lecasofts/go/obitools/pkg/goutils"
+	"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiseq"
+)
+
+type Table4mer [256]uint16
+
+func Count4Mer(seq *obiseq.BioSequence, buffer *[]byte, counts *Table4mer) *Table4mer {
+
+	iternal_buffer := Encode4mer(seq, buffer)
+
+	if counts == nil {
+		var w Table4mer
+		counts = &w
+	}
+
+	// Every cells of the counter is set to zero
+	for i := 0; i < 256; i++ {
+		(*counts)[i] = 0
+	}
+
+	for _, code := range iternal_buffer {
+		(*counts)[code]++
+	}
+	return counts
+}
+
+func Common4Mer(count1, count2 *Table4mer) int {
+	sum := 0
+	for i := 0; i < 256; i++ {
+		sum += int(goutils.MinUInt16((*count1)[i], (*count2)[i]))
+	}
+	return sum
+}
+
+func Sum4Mer(count *Table4mer) int {
+	sum := 0
+	for i := 0; i < 256; i++ {
+		sum += int((*count)[i])
+	}
+	return sum
+}
+
+func LCS4MerBounds(count1, count2 *Table4mer) (int, int) {
+	s1 := Sum4Mer(count1)
+	s2 := Sum4Mer(count2)
+	smin := goutils.MinInt(s1, s2)
+
+	cw := Common4Mer(count1, count2)
+
+	lcsMax := smin + 3 - int(math.Ceil(float64(smin-cw)/4.0))
+	lcsMin := cw
+
+	if cw > 0 {
+		lcsMin += 3
+	}
+
+	return lcsMin, lcsMax
+}
+
+func Error4MerBounds(count1, count2 *Table4mer) (int, int) {
+	s1 := Sum4Mer(count1)
+	s2 := Sum4Mer(count2)
+	smax := goutils.MaxInt(s1, s2)
+
+	cw := Common4Mer(count1, count2)
+
+	errorMax := smax - cw + 2* int(math.Floor(float64(cw+5)/8.0))
+	errorMin := int(math.Ceil(float64(errorMax) / 4.0))
+
+	return errorMin, errorMax
+}

pkg/obikmer/encodefourmer.go

@@ -21,7 +21,7 @@ var __single_base_code__ = []byte{0,
 }
 
 // Encode4mer transforms an obiseq.BioSequence into a sequence
-// of kmer of length 4. Each letter of the sequence noot belonging
+// of kmer of length 4. Each letter of the sequence not belonging
 // A, C, G, T, U are considered as a A. The kmer is encoded as a byte
 // value ranging from 0 to 255. Each nucleotite is represented by
 // two bits. The values 0, 1, 2, 3 correspond respectively to A, C, G,
@@ -65,15 +65,24 @@ func Encode4mer(seq *obiseq.BioSequence, buffer *[]byte) []byte {
 	return *buffer
 }
 
+// Index4mer returns an index where the occurrence position of every fourmer is
+// stored. The index is returned as an array of slices of integer.  The first
+// dimention corresponds to the code of the 4mer, the second
 func Index4mer(seq *obiseq.BioSequence, index *[][]int, buffer *[]byte) [][]int {
 
 	iternal_buffer := Encode4mer(seq, buffer)
 
 	if index == nil || cap(*index) < 256 {
+		// A new index is created
 		i := make([][]int, 256)
-		index = &i
+		if index == nil {
+			index = &i
+		} else {
+			*index = i
+		}
 	}
 
+	// Every cells of the index is emptied
 	for i := 0; i < 256; i++ {
 		(*index)[i] = (*index)[i][:0]
 	}
@@ -85,6 +94,9 @@ func Index4mer(seq *obiseq.BioSequence, index *[][]int, buffer *[]byte) [][]int
 	return *index
 }
 
+// FastShiftFourMer runs a Fast algorithm (similar to the one used in FASTA) to compare two sequences.
+// The returned values are two integer values. The shift between both the sequences and the count of
+// matching 4mer when this shift is applied between both the sequences.
 func FastShiftFourMer(index [][]int, seq *obiseq.BioSequence, buffer *[]byte) (int, int) {
 
 	iternal_buffer := Encode4mer(seq, buffer)
@@ -115,3 +127,4 @@ func FastShiftFourMer(index [][]int, seq *obiseq.BioSequence, buffer *[]byte) (i
 
 	return maxshift, maxcount
 }
+