mirror of
https://github.com/metabarcoding/obitools4.git
synced 2025-06-29 16:20:46 +00:00
Patch a bug on writing to stdout, and add clearer error on openning data files
This commit is contained in:
144
pkg/obialign/readalign.go
Normal file
144
pkg/obialign/readalign.go
Normal file
@ -0,0 +1,144 @@
|
||||
package obialign
|
||||
|
||||
import (
|
||||
"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obikmer"
|
||||
"git.metabarcoding.org/obitools/obitools4/obitools4/pkg/obiseq"
|
||||
)
|
||||
|
||||
func ReadAlign(seqA, seqB *obiseq.BioSequence,
|
||||
gap, scale float64, delta int, fastScoreRel bool,
|
||||
arena PEAlignArena, shift_buff *map[int]int) (int, []int, int, int, float64, bool) {
|
||||
var score, shift int
|
||||
var startA, startB int
|
||||
var partLen, over int
|
||||
var rawSeqA, qualSeqA []byte
|
||||
var rawSeqB, qualSeqB []byte
|
||||
var extra5, extra3 int
|
||||
|
||||
var path []int
|
||||
|
||||
if !_InitializedDnaScore {
|
||||
_InitDNAScoreMatrix()
|
||||
}
|
||||
|
||||
fastCount := -1
|
||||
fastScore := -1.0
|
||||
|
||||
directAlignment := true
|
||||
|
||||
index := obikmer.Index4mer(seqA,
|
||||
&arena.pointer.fastIndex,
|
||||
&arena.pointer.fastBuffer)
|
||||
|
||||
shift, fastCount, fastScore = obikmer.FastShiftFourMer(index, shift_buff, seqA.Len(), seqB, fastScoreRel, nil)
|
||||
|
||||
seqBR := seqB.ReverseComplement(false)
|
||||
shiftR, fastCountR, fastScoreR := obikmer.FastShiftFourMer(index, shift_buff, seqA.Len(), seqBR, fastScoreRel, nil)
|
||||
|
||||
if fastCount < fastCountR {
|
||||
shift = shiftR
|
||||
fastCount = fastCountR
|
||||
fastScore = fastScoreR
|
||||
seqB = seqBR
|
||||
directAlignment = false
|
||||
}
|
||||
|
||||
if shift > 0 {
|
||||
over = seqA.Len() - shift
|
||||
} else {
|
||||
over = seqB.Len() + shift
|
||||
}
|
||||
|
||||
// At least one mismatch exists in the overlaping region
|
||||
if fastCount+3 < over {
|
||||
|
||||
if shift > 0 {
|
||||
startA = shift - delta
|
||||
if startA < 0 {
|
||||
startA = 0
|
||||
}
|
||||
extra5 = -startA
|
||||
startB = 0
|
||||
|
||||
rawSeqA = seqA.Sequence()[startA:]
|
||||
qualSeqA = seqA.Qualities()[startA:]
|
||||
partLen = len(rawSeqA)
|
||||
if partLen > seqB.Len() {
|
||||
partLen = seqB.Len()
|
||||
}
|
||||
rawSeqB = seqB.Sequence()[0:partLen]
|
||||
qualSeqB = seqB.Qualities()[0:partLen]
|
||||
extra3 = seqB.Len() - partLen
|
||||
score = _FillMatrixPeLeftAlign(
|
||||
rawSeqA, qualSeqA, rawSeqB, qualSeqB, gap, scale,
|
||||
&arena.pointer.scoreMatrix,
|
||||
&arena.pointer.pathMatrix)
|
||||
} else {
|
||||
|
||||
startA = 0
|
||||
startB = -shift - delta
|
||||
if startB < 0 {
|
||||
startB = 0
|
||||
}
|
||||
extra5 = startB
|
||||
rawSeqB = seqB.Sequence()[startB:]
|
||||
qualSeqB = seqB.Qualities()[startB:]
|
||||
partLen = len(rawSeqB)
|
||||
if partLen > seqA.Len() {
|
||||
partLen = seqA.Len()
|
||||
}
|
||||
rawSeqA = seqA.Sequence()[:partLen]
|
||||
qualSeqA = seqA.Qualities()[:partLen]
|
||||
extra3 = partLen - seqA.Len()
|
||||
|
||||
score = _FillMatrixPeRightAlign(
|
||||
rawSeqA, qualSeqA, rawSeqB, qualSeqB, gap, scale,
|
||||
&arena.pointer.scoreMatrix,
|
||||
&arena.pointer.pathMatrix)
|
||||
}
|
||||
|
||||
path = _Backtracking(arena.pointer.pathMatrix,
|
||||
len(rawSeqA), len(rawSeqB),
|
||||
&arena.pointer.path)
|
||||
|
||||
} else {
|
||||
|
||||
// Both overlaping regions are identicals
|
||||
|
||||
if shift > 0 {
|
||||
startA = shift
|
||||
startB = 0
|
||||
extra5 = -startA
|
||||
qualSeqA = seqA.Qualities()[startA:]
|
||||
partLen = len(qualSeqA)
|
||||
qualSeqB = seqB.Qualities()[0:partLen]
|
||||
extra3 = seqB.Len() - partLen
|
||||
score = 0
|
||||
} else {
|
||||
startA = 0
|
||||
startB = -shift
|
||||
extra5 = startB
|
||||
qualSeqB = seqB.Qualities()[startB:]
|
||||
partLen = len(qualSeqB)
|
||||
extra3 = partLen - seqA.Len()
|
||||
qualSeqA = seqA.Qualities()[:partLen]
|
||||
}
|
||||
score = 0
|
||||
for i, qualA := range qualSeqA {
|
||||
qualB := qualSeqB[i]
|
||||
score += _NucScorePartMatchMatch[qualA][qualB]
|
||||
}
|
||||
|
||||
path = arena.pointer.path[:0]
|
||||
path = append(path, 0, partLen)
|
||||
}
|
||||
|
||||
path[0] += extra5
|
||||
if path[len(path)-1] == 0 {
|
||||
path[len(path)-2] += extra3
|
||||
} else {
|
||||
path = append(path, extra3, 0)
|
||||
}
|
||||
|
||||
return score, path, fastCount, over, fastScore, directAlignment
|
||||
}
|
Reference in New Issue
Block a user