// Return true if any changes made
bool RefineTask::RefineVertP(MSA* msaIn, unsigned uIters)
{
bool bAnyChanges = false;
const unsigned uColCountIn = msaIn->GetColCount();
const unsigned uSeqCountIn = msaIn->GetSeqCount();
if (uColCountIn < 3 || uSeqCountIn < 3)
return false;
unsigned *AnchorCols = new unsigned[uColCountIn];
unsigned uAnchorColCount;
SetMSAWeightsMuscle(*msaIn);
FindAnchorCols(*msaIn, AnchorCols, &uAnchorColCount);
const unsigned uRangeCount = uAnchorColCount + 1;
Range *Ranges = new Range[uRangeCount];
ColsToRanges(AnchorCols, uAnchorColCount, uColCountIn, Ranges);
ListVertSavings(uColCountIn, uAnchorColCount, Ranges, uRangeCount);
delete[] AnchorCols;
MSA msaOut;
msaOut.SetSize(uSeqCountIn, 0);
for (unsigned uSeqIndex = 0; uSeqIndex < uSeqCountIn ; ++uSeqIndex)
{
const char *ptrName = msaIn->GetSeqName(uSeqIndex);
unsigned uId = msaIn->GetSeqId(uSeqIndex);
msaOut.SetSeqName(uSeqIndex, ptrName);
msaOut.SetSeqId(uSeqIndex, uId);
}
MuscleContext* ctx = getMuscleContext();
workpool->uRangeCount = uRangeCount;
for (unsigned uRangeIndex = 0; uRangeIndex < uRangeCount && !ctx->isCanceled(); ++uRangeIndex)
{
workpool->uRangeIndex = uRangeIndex;
MSA msaRange;
const Range &r = Ranges[uRangeIndex];
const unsigned uFromColIndex = r.m_uBestColLeft;
const unsigned uRangeColCount = r.m_uBestColRight - uFromColIndex;
if (0 == uRangeColCount)
continue;
else if (1 == uRangeColCount)
{
MSAFromColRange(*msaIn, uFromColIndex, 1, msaRange);
MSAAppend(msaOut, msaRange);
continue;
}
MSAFromColRange(*msaIn, uFromColIndex, uRangeColCount, msaRange);
bool &bLockLeft = workpool->bLockLeft;
bool &bLockRight = workpool->bLockRight;
bLockLeft = (0 != uRangeIndex);
bLockRight = (uRangeCount - 1 != uRangeIndex);
bool bAnyChangesThisBlock = RefineHorizP(&msaRange, uIters, bLockLeft, bLockRight);
bAnyChanges = (bAnyChanges || bAnyChangesThisBlock);
MSAAppend(msaOut, msaRange);
}
delete[] Ranges;
if (ctx->isCanceled()) {
throw MuscleException("Canceled");
}
#if DEBUG
// Sanity check
AssertMSAEqIgnoreCaseAndGaps(*msaIn, msaOut);
#endif
if (bAnyChanges) {
msaIn->Copy(msaOut);
}
return bAnyChanges;
}
// Return true if any changes made
void AnchoredProfileProfile(MSA &msa1, MSA &msa2, MSA &msaOut)
{
const unsigned uColCountIn = msa1.GetColCount();
const unsigned uSeqCountIn = msa1.GetSeqCount() + msa2.GetSeqCount();
unsigned *AnchorCols = new unsigned[uColCountIn];
unsigned uAnchorColCount;
PrepareMSAforScoring(msa1);
PrepareMSAforScoring(msa2);
FindAnchorColsPP(msa1, msa2, AnchorCols, &uAnchorColCount);
const unsigned uRangeCount = uAnchorColCount + 1;
Range *Ranges = new Range[uRangeCount];
#if TRACE
Log("%u ranges\n", uRangeCount);
#endif
ColsToRanges(AnchorCols, uAnchorColCount, uColCountIn, Ranges);
ListVertSavings(uColCountIn, uAnchorColCount, Ranges, uRangeCount);
#if TRACE
{
Log("Anchor cols: ");
for (unsigned i = 0; i < uAnchorColCount; ++i)
Log(" %u", AnchorCols[i]);
Log("\n");
Log("Ranges:\n");
for (unsigned i = 0; i < uRangeCount; ++i)
Log("%4u - %4u\n", Ranges[i].m_uBestColLeft, Ranges[i].m_uBestColRight);
}
#endif
delete[] AnchorCols;
msaOut.SetSize(uSeqCountIn, 0);
for (unsigned uSeqIndex = 0; uSeqIndex < uSeqCountIn; ++uSeqIndex)
{
const char *ptrName;
unsigned uId;
if( uSeqIndex < msa1.GetSeqCount() )
{
msa1.SetSeqId(uSeqIndex, uSeqIndex);
ptrName = msa1.GetSeqName(uSeqIndex);
}
else
{
msa2.SetSeqId(uSeqIndex-msa1.GetSeqCount(), uSeqIndex);
ptrName = msa2.GetSeqName(uSeqIndex-msa1.GetSeqCount());
}
msaOut.SetSeqName(uSeqIndex, ptrName);
msaOut.SetSeqId(uSeqIndex, uSeqIndex);
}
for (unsigned uRangeIndex = 0; uRangeIndex < uRangeCount; ++uRangeIndex)
{
MSA msaRange1;
MSA msaRange2;
MSA msaRangeOut;
const Range &r = Ranges[uRangeIndex];
const unsigned uFromColIndex = r.m_uBestColLeft;
const unsigned uRangeColCount = r.m_uBestColRight - uFromColIndex;
if (0 == uRangeColCount)
continue;
/* else if (1 == uRangeColCount)
{
MSAFromColRange(msaIn, uFromColIndex, 1, msaRange);
MSAAppend(msaOut, msaRange);
continue;
}
*/
MSAFromColRange(msa1, uFromColIndex, uRangeColCount, msaRange1);
MSAFromColRange(msa2, uFromColIndex, uRangeColCount, msaRange2);
StripGapColumns(msaRange1);
StripGapColumns(msaRange2);
#if TRACE
Log("\n-------------\n");
Log("Range %u - %u count=%u\n", r.m_uBestColLeft, r.m_uBestColRight, uRangeColCount);
Log("Before:\n");
msaRange1.LogMe();
msaRange2.LogMe();
#endif
ProfileProfile(msaRange1, msaRange2, msaRangeOut);
#if TRACE
Log("After:\n");
msaRangeOut.LogMe();
#endif
for (unsigned uSeqIndex = 0; uSeqIndex < uSeqCountIn; ++uSeqIndex)
msaRangeOut.SetSeqId(uSeqIndex, uSeqIndex);
MSAAppend(msaOut, msaRangeOut);
#if TRACE
Log("msaOut after Cat:\n");
msaOut.LogMe();
#endif
}
delete[] Ranges;
}
void AlignTwoMSAsGivenPathSW(const PWPath &Path, const MSA &msaA, const MSA &msaB,
MSA &msaCombined)
{
msaCombined.Clear();
#if TRACE
Log("AlignTwoMSAsGivenPathSW\n");
Log("Template A:\n");
msaA.LogMe();
Log("Template B:\n");
msaB.LogMe();
#endif
const unsigned uColCountA = msaA.GetColCount();
const unsigned uColCountB = msaB.GetColCount();
const unsigned uSeqCountA = msaA.GetSeqCount();
const unsigned uSeqCountB = msaB.GetSeqCount();
msaCombined.SetSeqCount(uSeqCountA + uSeqCountB);
// Copy sequence names into combined MSA
for (unsigned uSeqIndexA = 0; uSeqIndexA < uSeqCountA; ++uSeqIndexA)
{
msaCombined.SetSeqName(uSeqIndexA, msaA.GetSeqName(uSeqIndexA));
msaCombined.SetSeqId(uSeqIndexA, msaA.GetSeqId(uSeqIndexA));
}
for (unsigned uSeqIndexB = 0; uSeqIndexB < uSeqCountB; ++uSeqIndexB)
{
msaCombined.SetSeqName(uSeqCountA + uSeqIndexB, msaB.GetSeqName(uSeqIndexB));
msaCombined.SetSeqId(uSeqCountA + uSeqIndexB, msaB.GetSeqId(uSeqIndexB));
}
unsigned uColIndexA = 0;
unsigned uColIndexB = 0;
unsigned uColIndexCombined = 0;
const unsigned uEdgeCount = Path.GetEdgeCount();
for (unsigned uEdgeIndex = 0; uEdgeIndex < uEdgeCount; ++uEdgeIndex)
{
const PWEdge &Edge = Path.GetEdge(uEdgeIndex);
#if TRACE
Log("\nEdge %u %c%u.%u\n",
uEdgeIndex,
Edge.cType,
Edge.uPrefixLengthA,
Edge.uPrefixLengthB);
#endif
const char cType = Edge.cType;
const unsigned uPrefixLengthA = Edge.uPrefixLengthA;
unsigned uColCountA = 0;
if (uPrefixLengthA > 0)
{
const unsigned uNodeIndexA = uPrefixLengthA - 1;
const unsigned uTplColIndexA = uNodeIndexA;
if (uTplColIndexA > uColIndexA)
uColCountA = uTplColIndexA - uColIndexA;
}
const unsigned uPrefixLengthB = Edge.uPrefixLengthB;
unsigned uColCountB = 0;
if (uPrefixLengthB > 0)
{
const unsigned uNodeIndexB = uPrefixLengthB - 1;
const unsigned uTplColIndexB = uNodeIndexB;
if (uTplColIndexB > uColIndexB)
uColCountB = uTplColIndexB - uColIndexB;
}
AppendUnalignedTerminals(msaA, uColIndexA, uColCountA, msaB, uColIndexB, uColCountB,
uSeqCountA, uSeqCountB, msaCombined, uColIndexCombined);
switch (cType)
{
case 'M':
{
assert(uPrefixLengthA > 0);
assert(uPrefixLengthB > 0);
const unsigned uColA = uPrefixLengthA - 1;
const unsigned uColB = uPrefixLengthB - 1;
assert(uColIndexA == uColA);
assert(uColIndexB == uColB);
AppendMatch(msaA, uColIndexA, msaB, uColIndexB, uSeqCountA, uSeqCountB,
msaCombined, uColIndexCombined);
break;
}
case 'D':
{
assert(uPrefixLengthA > 0);
const unsigned uColA = uPrefixLengthA - 1;
assert(uColIndexA == uColA);
AppendDelete(msaA, uColIndexA, uSeqCountA, uSeqCountB, msaCombined, uColIndexCombined);
break;
}
case 'I':
{
assert(uPrefixLengthB > 0);
const unsigned uColB = uPrefixLengthB - 1;
assert(uColIndexB == uColB);
AppendInsert(msaB, uColIndexB, uSeqCountA, uSeqCountB, msaCombined, uColIndexCombined);
break;
}
default:
assert(false);
}
}
unsigned uInsertColCountA = uColCountA - uColIndexA;
unsigned uInsertColCountB = uColCountB - uColIndexB;
AppendUnalignedTerminals(msaA, uColIndexA, uInsertColCountA, msaB, uColIndexB,
uInsertColCountB, uSeqCountA, uSeqCountB, msaCombined, uColIndexCombined);
}
