2022-01-13 23:27:39 +01:00
|
|
|
package obialign
|
|
|
|
|
|
|
|
import (
|
|
|
|
"math"
|
|
|
|
)
|
|
|
|
|
2022-01-14 15:30:01 +01:00
|
|
|
var _FourBitsCount = []float64{
|
2022-01-13 23:27:39 +01:00
|
|
|
0, // 0000
|
|
|
|
1, // 0001
|
|
|
|
1, // 0010
|
|
|
|
2, // 0011
|
|
|
|
1, // 0100
|
|
|
|
2, // 0101
|
|
|
|
2, // 0110
|
|
|
|
3, // 0111
|
|
|
|
1, // 1000
|
|
|
|
2, // 1001
|
|
|
|
2, // 1010
|
|
|
|
3, // 1011
|
|
|
|
2, // 1100
|
|
|
|
3, // 1101
|
|
|
|
3, // 1110
|
|
|
|
4, // 1111
|
|
|
|
}
|
|
|
|
|
2022-01-14 15:30:01 +01:00
|
|
|
var _InitializedDnaScore = false
|
2022-01-13 23:27:39 +01:00
|
|
|
|
2022-01-14 15:30:01 +01:00
|
|
|
var _NucPartMatch [32][32]float64
|
|
|
|
var _NucScorePartMatchMatch [100][100]int
|
|
|
|
var _NucScorePartMatchMismatch [100][100]int
|
2022-01-13 23:27:39 +01:00
|
|
|
|
2022-01-14 15:30:01 +01:00
|
|
|
func _MatchRatio(a, b byte) float64 {
|
2022-01-13 23:27:39 +01:00
|
|
|
// count of common bits
|
2022-01-14 15:30:01 +01:00
|
|
|
cm := _FourBitsCount[a&b&15]
|
2022-01-13 23:27:39 +01:00
|
|
|
|
2022-01-14 15:30:01 +01:00
|
|
|
ca := _FourBitsCount[a&15]
|
|
|
|
cb := _FourBitsCount[b&15]
|
2022-01-13 23:27:39 +01:00
|
|
|
|
|
|
|
if cm == 0 || ca == 0 || cb == 0 {
|
|
|
|
return float64(0)
|
|
|
|
}
|
|
|
|
|
|
|
|
return float64(cm) / float64(ca) / float64(cb)
|
|
|
|
}
|
|
|
|
|
2022-01-14 15:30:01 +01:00
|
|
|
func _Logaddexp(a, b float64) float64 {
|
2022-01-13 23:27:39 +01:00
|
|
|
if a > b {
|
|
|
|
a, b = b, a
|
|
|
|
}
|
|
|
|
|
|
|
|
return b + math.Log1p(math.Exp(a-b))
|
|
|
|
}
|
|
|
|
|
2022-01-14 15:30:01 +01:00
|
|
|
func _MatchScoreRatio(a, b byte) (float64, float64) {
|
2022-01-13 23:27:39 +01:00
|
|
|
|
|
|
|
l2 := math.Log(2)
|
|
|
|
l3 := math.Log(3)
|
|
|
|
l4 := math.Log(4)
|
|
|
|
l10 := math.Log(10)
|
|
|
|
lE1 := -float64(a)/10*l10 - l4
|
|
|
|
lE2 := -float64(b)/10*l10 - l4
|
|
|
|
lO1 := math.Log1p(-math.Exp(lE1 + l3))
|
|
|
|
lO2 := math.Log1p(-math.Exp(lE2 + l3))
|
|
|
|
lO1O2 := lO1 + lO2
|
|
|
|
lE1E2 := lE1 + lE2
|
|
|
|
lO1E2 := lO1 + lE2
|
|
|
|
lO2E1 := lO2 + lE1
|
|
|
|
|
2022-01-14 15:30:01 +01:00
|
|
|
MM := _Logaddexp(lO1O2, lE1E2+l3) + l4
|
|
|
|
Mm := _Logaddexp(_Logaddexp(lO1E2, lO2E1), lE1E2+l2) + l4
|
2022-01-13 23:27:39 +01:00
|
|
|
|
|
|
|
return MM, Mm
|
|
|
|
}
|
|
|
|
|
2022-01-14 15:30:01 +01:00
|
|
|
func _InitNucPartMatch() {
|
2022-01-13 23:27:39 +01:00
|
|
|
|
2022-01-14 15:36:39 +01:00
|
|
|
for i, a := range _FourBitsBaseCode {
|
|
|
|
for j, b := range _FourBitsBaseCode {
|
2022-01-14 15:30:01 +01:00
|
|
|
_NucPartMatch[i][j] = _MatchRatio(a, b)
|
2022-01-13 23:27:39 +01:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2022-01-14 15:30:01 +01:00
|
|
|
func _InitNucScorePartMatch() {
|
2022-01-13 23:27:39 +01:00
|
|
|
for i := 0; i < 100; i++ {
|
|
|
|
for j := 0; j < 100; j++ {
|
2022-01-14 15:30:01 +01:00
|
|
|
MM, Mm := _MatchScoreRatio(byte(i), byte(j))
|
|
|
|
_NucScorePartMatchMatch[i][j] = int(MM*10 + 0.5)
|
|
|
|
_NucScorePartMatchMismatch[i][j] = int(Mm*10 + 0.5)
|
2022-01-13 23:27:39 +01:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2022-01-14 15:30:01 +01:00
|
|
|
func _InitDNAScoreMatrix() {
|
|
|
|
if !_InitializedDnaScore {
|
|
|
|
_InitNucPartMatch()
|
|
|
|
_InitNucScorePartMatch()
|
|
|
|
_InitializedDnaScore = true
|
2022-01-13 23:27:39 +01:00
|
|
|
}
|
|
|
|
}
|