obitools/obitools4
2
0
Fork 0
mirror of https://github.com/metabarcoding/obitools4.git synced 2025-06-29 16:20:46 +00:00
Code Issues Packages Projects Releases Wiki Activity

Add a first version of obitag the successor of ecotag

Browse Source
This commit is contained in:
Eric Coissac
2022-10-26 13:16:56 +02:00
parent e17d1fbca6
commit 8aa323dad5
17 changed files with 884 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

118
pkg/obialign/fulllcs.go Normal file
View File

@ -0,0 +1,118 @@
package obialign
import (
"log"
"git.metabarcoding.org/lecasofts/go/obitools/pkg/obiseq"
)
type FullLCSMatrix struct {
matrix []int16 // Score matrix
lenght []int16 // Alignment length matrix
ll int // Length of the longest sequence
l int // Length of the shortest sequence
}
func NewFullLCSMatrix(matrix *FullLCSMatrix, L, l int) *FullLCSMatrix {
if matrix == nil {
matrix = &FullLCSMatrix{}
}
if l > L {
log.Panicf("L (%d) must be greater or equal to l (%d)", L, l)
}
needed := (L) * (l)
if needed > matrix.Cap() {
matrix.matrix = make([]int16, needed)
matrix.lenght = make([]int16, needed)
}
matrix.matrix = matrix.matrix[:needed]
matrix.lenght = matrix.lenght[:needed]
matrix.ll = L
matrix.l = l
return matrix
}
func (matrix *FullLCSMatrix) Cap() int {
return cap(matrix.matrix)
}
func (matrix *FullLCSMatrix) Length() int {
return len(matrix.matrix)
}
func (matrix *FullLCSMatrix) Get(i, j int) (int16, int16) {
if i == 0 {
return 0, int16(j)
}
if j == 0 {
return 0, int16(i)
}
pos := (i-1)*matrix.ll + j - 1
return matrix.matrix[pos], matrix.lenght[pos]
}
func (matrix *FullLCSMatrix) Set(i, j int, score, length int16) {
if i > 0 && j > 0 {
pos := (i-1)*matrix.ll + j - 1
matrix.matrix[pos] = score
matrix.lenght[pos] = length
}
}
// Computes the LCS between two DNA sequences and the length of the
// corresponding alignment
func FullLCSScore(seqA, seqB *obiseq.BioSequence,
matrix *FullLCSMatrix) (int, int) {
if seqA.Length() == 0 {
log.Fatal("Sequence A has a length of 0")
}
if seqB.Length() == 0 {
log.Fatal("Sequence B has a length of 0")
}
// swapped := false
if seqA.Length() < seqB.Length() {
seqA, seqB = seqB, seqA
// swapped = true
}
la := seqA.Length()
lb := seqB.Length()
sa := seqA.Sequence()
sb := seqB.Sequence()
matrix = NewFullLCSMatrix(matrix, la, lb)
for i := 1; i <= matrix.l; i++ {
for j := 1; j <= matrix.ll; j++ {
sd, ld := matrix.Get(i-1, j-1)
if sb[i-1] == sa[j-1] {
sd++
}
sup, lup := matrix.Get(i-1, j)
sleft, lleft := matrix.Get(i, j-1)
switch {
case sd >= sup && sd >= sleft:
matrix.Set(i, j, sd, ld+1)
case sup >= sleft && sup >= sd:
matrix.Set(i, j, sup, lup+1)
default:
matrix.Set(i, j, sleft, lleft+1)
}
}
}
s, l := matrix.Get(lb, la)
return int(s), int(l)
}

39
pkg/obialign/lcs.go
View File

@ -96,8 +96,18 @@ func (matrix *LCSMatrix) Set(i, j int, score, length int16) {
}
}
// Computes the LCS between two DNA sequences and the length of the
// corresponding alignment
func LCSScore(seqA, seqB *obiseq.BioSequence, maxError int,
matrix *LCSMatrix) (int, int) {
if seqA.Length() == 0 {
log.Fatal("Sequence A has a length of 0")
}
if seqB.Length() == 0 {
log.Fatal("Sequence B has a length of 0")
}
// swapped := false
if seqA.Length() < seqB.Length() {
@ -105,18 +115,39 @@ func LCSScore(seqA, seqB *obiseq.BioSequence, maxError int,
// swapped = true
}
if (seqA.Length() - seqB.Length()) > maxError {
return -1, -1
}
matrix = NewLCSMatrix(matrix, seqA.Length(), seqB.Length(), maxError)
la := seqA.Length()
lb := seqB.Length()
sa := seqA.Sequence()
sb := seqB.Sequence()
matrix = NewLCSMatrix(matrix, la, lb, maxError)
for i := 1; i <= matrix.l; i++ {
ij := max(1, i-matrix.extra)
sj := min(i+matrix.delta+matrix.extra, matrix.ll)
for j := ij; j <= sj; j++ {
sd, ld := matrix.Get(i-1, j-1)
if seqB.Sequence()[i-1] == seqA.Sequence()[j-1] {
if i > lb {
log.Println("Error on seq B ", 1, matrix.l)
log.Println(i)
log.Println(seqB.Length(), "/", lb)
log.Println(string(sa))
log.Fatalln(string(sb))
}
if j > la {
log.Println("Error on seq A ", ij, sj)
log.Println(j)
log.Println(seqA.Length(), "/", la)
log.Println(string(sa))
log.Fatalln(string(sb))
}
if sb[i-1] == sa[j-1] {
sd++
}
sup, lup := matrix.Get(i-1, j)
@ -132,11 +163,13 @@ func LCSScore(seqA, seqB *obiseq.BioSequence, maxError int,
}
}
s, l := matrix.Get(seqB.Length(), seqA.Length())
s, l := matrix.Get(lb, la)
if (l - s) > int16(maxError) {
// log.Println(l,s,l-s,maxError)
return -1, -1
}
return int(s), int(l)
}