// Merge diagonals that are continuations of each other with
// short breaks of up to length g_uMaxDiagBreak.
// In a sorted list of diagonals, we only have to check
// consecutive entries.
void MergeDiags(DiagList &DL)
{
unsigned &g_uMaxDiagBreak = getMuscleContext()->params.g_uMaxDiagBreak;
return;
#if DEBUG
if (!DL.IsSorted())
Quit("MergeDiags: !IsSorted");
#endif
// TODO: Fix this!
// Breaks must be with no offset (no gaps)
const unsigned uCount = DL.GetCount();
if (uCount <= 1)
return;
DiagList NewList;
Diag MergedDiag;
const Diag *ptrPrev = &DL.Get(0);
for (unsigned i = 1; i < uCount; ++i)
{
const Diag *ptrDiag = &DL.Get(i);
unsigned uBreakLength = DiagBreak(*ptrPrev, *ptrDiag);
if (uBreakLength <= g_uMaxDiagBreak)
{
MergedDiag.m_uStartPosA = ptrPrev->m_uStartPosA;
MergedDiag.m_uStartPosB = ptrPrev->m_uStartPosB;
MergedDiag.m_uLength = ptrPrev->m_uLength + ptrDiag->m_uLength
+ uBreakLength;
ptrPrev = &MergedDiag;
}
else
{
NewList.Add(*ptrPrev);
ptrPrev = ptrDiag;
}
}
NewList.Add(*ptrPrev);
DL.Copy(NewList);
}
void FindDiagsNuc(const ProfPos *PX, unsigned uLengthX, const ProfPos *PY,
unsigned uLengthY, DiagList &DL)
{
MuscleContext *ctx = getMuscleContext();
ALPHA &g_Alpha = ctx->alpha.g_Alpha;
unsigned* TuplePos = ctx->finddiagsn.TuplePos;
unsigned &g_uMinDiagLength = ctx->params.g_uMinDiagLength;
if (ALPHA_DNA != g_Alpha && ALPHA_RNA != g_Alpha)
Quit("FindDiagsNuc: requires nucleo alphabet");
DL.Clear();
// 16 is arbitrary slop, no principled reason for this.
if (uLengthX < K + 16 || uLengthY < K + 16)
return;
// Set A to shorter profile, B to longer
const ProfPos *PA;
const ProfPos *PB;
unsigned uLengthA;
unsigned uLengthB;
bool bSwap;
if (uLengthX < uLengthY)
{
bSwap = false;
PA = PX;
PB = PY;
uLengthA = uLengthX;
uLengthB = uLengthY;
}
else
{
bSwap = true;
PA = PY;
PB = PX;
uLengthA = uLengthY;
uLengthB = uLengthX;
}
#if TRACE
Log("FindDiagsNuc(LengthA=%d LengthB=%d\n", uLengthA, uLengthB);
#endif
// Build tuple map for the longer profile, B
if (uLengthB < K)
Quit("FindDiags: profile too short");
memset(TuplePos, EMPTY, sizeof(TuplePos));
for (unsigned uPos = 0; uPos < uLengthB - K; ++uPos)
{
const unsigned uTuple = GetTuple(PB, uPos);
if (EMPTY == uTuple)
continue;
TuplePos[uTuple] = uPos;
}
// Find matches
for (unsigned uPosA = 0; uPosA < uLengthA - K; ++uPosA)
{
const unsigned uTuple = GetTuple(PA, uPosA);
if (EMPTY == uTuple)
continue;
const unsigned uPosB = TuplePos[uTuple];
if (EMPTY == uPosB)
continue;
// This tuple is found in both profiles
unsigned uStartPosA = uPosA;
unsigned uStartPosB = uPosB;
// Try to extend the match forwards
unsigned uEndPosA = uPosA + K - 1;
unsigned uEndPosB = uPosB + K - 1;
for (;;)
{
if (uLengthA - 1 == uEndPosA || uLengthB - 1 == uEndPosB)
break;
const unsigned uAAGroupA = PA[uEndPosA+1].m_uResidueGroup;
if (RESIDUE_GROUP_MULTIPLE == uAAGroupA)
break;
const unsigned uAAGroupB = PB[uEndPosB+1].m_uResidueGroup;
if (RESIDUE_GROUP_MULTIPLE == uAAGroupB)
break;
if (uAAGroupA != uAAGroupB)
break;
++uEndPosA;
++uEndPosB;
}
uPosA = uEndPosA;
#if TRACE
{
Log("Match: A %4u-%4u ", uStartPosA, uEndPosA);
for (unsigned n = uStartPosA; n <= uEndPosA; ++n)
Log("%c", LetterToChar(PA[n].m_uResidueGroup));
Log("\n");
Log(" B %4u-%4u ", uStartPosB, uEndPosB);
for (unsigned n = uStartPosB; n <= uEndPosB; ++n)
Log("%c", LetterToChar(PB[n].m_uResidueGroup));
Log("\n");
}
#endif
const unsigned uLength = uEndPosA - uStartPosA + 1;
assert(uEndPosB - uStartPosB + 1 == uLength);
if (uLength >= g_uMinDiagLength)
{
if (bSwap)
DL.Add(uStartPosB, uStartPosA, uLength);
else
DL.Add(uStartPosA, uStartPosB, uLength);
}
}
}
void FindDiags(const ProfPos *PX, unsigned uLengthX, const ProfPos *PY,
unsigned uLengthY, DiagList &DL)
{
if (ALPHA_Amino != g_Alpha)
Quit("FindDiags: requires amino acid alphabet");
DL.Clear();
if (uLengthX < 12 || uLengthY < 12)
return;
// Set A to shorter profile, B to longer
const ProfPos *PA;
const ProfPos *PB;
unsigned uLengthA;
unsigned uLengthB;
bool bSwap;
if (uLengthX < uLengthY)
{
bSwap = false;
PA = PX;
PB = PY;
uLengthA = uLengthX;
uLengthB = uLengthY;
}
else
{
bSwap = true;
PA = PY;
PB = PX;
uLengthA = uLengthY;
uLengthB = uLengthX;
}
// Build tuple map for the longer profile, B
if (uLengthB < KTUP)
Quit("FindDiags: profile too short");
memset(TuplePos, EMPTY, sizeof(TuplePos));
for (unsigned uPos = 0; uPos < uLengthB - KTUP; ++uPos)
{
const unsigned uTuple = GetTuple(PB, uPos);
if (EMPTY == uTuple)
continue;
TuplePos[uTuple] = uPos;
}
// Find matches
for (unsigned uPosA = 0; uPosA < uLengthA - KTUP; ++uPosA)
{
const unsigned uTuple = GetTuple(PA, uPosA);
if (EMPTY == uTuple)
continue;
const unsigned uPosB = TuplePos[uTuple];
if (EMPTY == uPosB)
continue;
// This tuple is found in both profiles
unsigned uStartPosA = uPosA;
unsigned uStartPosB = uPosB;
// Try to extend the match forwards
unsigned uEndPosA = uPosA + KTUP - 1;
unsigned uEndPosB = uPosB + KTUP - 1;
for (;;)
{
if (uLengthA - 1 == uEndPosA || uLengthB - 1 == uEndPosB)
break;
const unsigned uAAGroupA = PA[uEndPosA+1].m_uResidueGroup;
if (RESIDUE_GROUP_MULTIPLE == uAAGroupA)
break;
const unsigned uAAGroupB = PB[uEndPosB+1].m_uResidueGroup;
if (RESIDUE_GROUP_MULTIPLE == uAAGroupB)
break;
if (uAAGroupA != uAAGroupB)
break;
++uEndPosA;
++uEndPosB;
}
uPosA = uEndPosA;
#if TRACE
{
Log("Match: A %4u-%4u ", uStartPosA, uEndPosA);
for (unsigned n = uStartPosA; n <= uEndPosA; ++n)
Log("%c", 'A' + PA[n].m_uResidueGroup);
Log("\n");
Log(" B %4u-%4u ", uStartPosB, uEndPosB);
for (unsigned n = uStartPosB; n <= uEndPosB; ++n)
Log("%c", 'A' + PB[n].m_uResidueGroup);
Log("\n");
}
#endif
const unsigned uLength = uEndPosA - uStartPosA + 1;
assert(uEndPosB - uStartPosB + 1 == uLength);
if (uLength >= g_uMinDiagLength)
{
if (bSwap)
DL.Add(uStartPosB, uStartPosA, uLength);
else
DL.Add(uStartPosA, uStartPosB, uLength);
}
}
}