fix issue in WFBacktrace and change format to proper CIGAR,

add test to ensure CIGAR correctness in the case of different traceback results,
add DecodeCIGAR function to exports
This commit is contained in:
2024-11-07 19:01:01 +00:00
parent 3da3ddf10c
commit cde429cb80
6 changed files with 248 additions and 64 deletions

View File

@@ -3,6 +3,7 @@ package tests
import (
"bufio"
"encoding/json"
"log"
"math/rand/v2"
"os"
"strconv"
@@ -46,6 +47,71 @@ func TestWavefrontPacking(t *testing.T) {
}
}
func GetScoreFromCIGAR(CIGAR string, penalties wfa.Penalty) int {
unpackedCIGAR := wfa.RunLengthDecode(CIGAR)
previousOp := '~'
score := 0
for _, Op := range unpackedCIGAR {
if Op == 'M' {
score = score + penalties.M
} else if Op == 'X' {
score = score + penalties.X
} else if (Op == 'I' && previousOp != 'I') || (Op == 'D' && previousOp != 'D') {
score = score + penalties.O + penalties.E
} else if (Op == 'I' && previousOp == 'I') || (Op == 'D' && previousOp == 'D') {
score = score + penalties.E
}
previousOp = Op
}
return score
}
func CheckCIGARCorrectness(s1 string, s2 string, CIGAR string) bool {
unpackedCIGAR := wfa.RunLengthDecode(CIGAR)
i := 0
j := 0
s1Aligned := strings.Builder{}
alignment := strings.Builder{}
s2Aligned := strings.Builder{}
for c := 0; c < len(unpackedCIGAR); c++ {
Op := unpackedCIGAR[c]
if Op == 'M' {
s1Aligned.WriteByte(s1[i])
alignment.WriteRune('|')
s2Aligned.WriteByte(s2[j])
i++
j++
} else if Op == 'X' {
s1Aligned.WriteByte(s1[i])
alignment.WriteRune(' ')
s2Aligned.WriteByte(s2[j])
i++
j++
} else if Op == 'I' {
s1Aligned.WriteRune('-')
alignment.WriteRune(' ')
s2Aligned.WriteByte(s2[j])
j++
} else if Op == 'D' {
s1Aligned.WriteByte(s1[i])
alignment.WriteRune('|')
s2Aligned.WriteRune('-')
i++
}
}
if i == len(s1) && j == len(s2) {
return true
} else {
log.Printf("\n%s\n%s\n%s\n i=%d, j=%d, |s1|=%d, |s2|=%d\n", s1Aligned.String(), alignment.String(), s2Aligned.String(), i, j, len(s1), len(s2))
return false
}
}
func TestWFA(t *testing.T) {
content, _ := os.ReadFile(testJsonPath)
@@ -73,7 +139,9 @@ func TestWFA(t *testing.T) {
for solutions.Scan() {
solution := solutions.Text()
expectedScore, _ := strconv.Atoi(strings.Split(solution, "\t")[0])
expectedCIGAR := strings.Split(solution, "\t")[1]
sequences.Scan()
s1 := sequences.Text()
@@ -85,9 +153,25 @@ func TestWFA(t *testing.T) {
x := wfa.WFAlign(s1, s2, testPenalties, true)
gotScore := x.Score
gotCIGAR := x.CIGAR
if gotScore != -1*expectedScore {
t.Errorf(`test: %s#%d, s1: %s, s2: %s, got: %d, expected: %d\n`, testName, idx, s1, s2, gotScore, expectedScore)
t.Errorf(`test: %s#%d, s1: %s, s2: %s, got: %d, expected: %d`, testName, idx, s1, s2, gotScore, expectedScore)
os.Exit(1)
}
if gotCIGAR != expectedCIGAR {
checkScore := GetScoreFromCIGAR(gotCIGAR, testPenalties)
CIGARCorrectness := CheckCIGARCorrectness(s1, s2, gotCIGAR)
if checkScore != gotScore && checkScore != -1*expectedScore { // nonequivalent alignment
t.Errorf(`test: %s#%d, s1: %s, s2: %s, got: [%s], expected: [%s]`, testName, idx, s1, s2, gotCIGAR, expectedCIGAR)
t.Errorf(`test: %s#%d, recalculated score: %d`, testName, idx, checkScore)
os.Exit(1)
}
if !CIGARCorrectness {
t.Errorf(`test: %s#%d, s1: %s, s2: %s, got: [%s], expected: [%s]`, testName, idx, s1, s2, gotCIGAR, expectedCIGAR)
os.Exit(1)
}
}
idx++