void ArgumentList::add(const std::string& ArgumentName, const MyMatrix& value)
{
ArgumentType thisOne = matrix;
std::pair<std::string, ArgumentType> thisPair(ArgumentName,thisOne);
ArgumentNames.push_back(thisPair);
MatrixArguments.insert(std::make_pair(ArgumentName,value));
RegisterName(ArgumentName, thisOne);
}
void ArgumentList::addList(const std::string& ArgumentName, const CellMatrix& values)
{
ArgumentType thisOne = list;
std::pair<std::string, ArgumentType> thisPair(ArgumentName,thisOne);
ArgumentNames.push_back(thisPair);
ListArguments.insert(std::make_pair(ArgumentName,values));
RegisterName(ArgumentName, thisOne);
}
void ArgumentList::add(const std::string& ArgumentName, bool value)
{
ArgumentType thisOne = boolean;
std::pair<std::string, ArgumentType> thisPair(ArgumentName,thisOne);
ArgumentNames.push_back(thisPair);
BoolArguments.insert(std::make_pair(ArgumentName,value));
RegisterName(ArgumentName, thisOne);
}
void ArgumentList::add(const std::string& ArgumentName, const std::string& value)
{
ArgumentType thisOne = string;
std::pair<std::string, ArgumentType> thisPair(ArgumentName,thisOne);
ArgumentNames.push_back(thisPair);
std::pair<std::string,std::string> valuePair(ArgumentName,value);
StringArguments.insert(valuePair);
RegisterName(ArgumentName, thisOne);
}
void ArgumentList::add(const std::string& ArgumentName, double value)
{
ArgumentType thisOne = number;
std::pair<std::string, ArgumentType> thisPair(ArgumentName,thisOne);
ArgumentNames.push_back(thisPair);
std::pair<std::string,double> valuePair(ArgumentName,value);
DoubleArguments.insert(valuePair);
RegisterName(ArgumentName, thisOne);
}
map<Range,vector<int> > Plink::mkBlks(int null1, int null2 )
{
// First SNP, vector of SNPs (inc. first)
map< int, vector<int> > blocks;
// Some constants
const double cutHighCI = 0.98;
const double cutLowCI = 0.70;
const double cutLowCIVar [5] = {0,0,0.80,0.50,0.50};
const double maxDist [5] = {0,0,20000,30000,1000000};
const double recHighCI = 0.90;
const double informFrac = 0.95;
const double fourGameteCutoff = 0.01;
const double mafThresh = 0.05;
// Set to skip SNPs with low MAFs
// Uses genome-wide reference number: need to allocate for all SNPs here
vector<bool> skipMarker(nl_all,false);
for (int x = 0; x < nl_all; x++)
skipMarker[x] = locus[x]->freq < mafThresh;
// Consider each chromosome one at a time; skip X for now
int startChromosome = locus[ 0 ]->chr;
int finalChromosome = locus[ nl_all - 1 ]->chr;
for (int chr = startChromosome ; chr <= finalChromosome; chr++)
{
if ( scaffold.find(chr) == scaffold.end() )
continue;
int fromPosition = scaffold[chr].lstart;
int toPosition = scaffold[chr].lstop;
int nsnps = toPosition - fromPosition + 1;
/////////////////////////////////////////////////////////////////////////
// Make a list of marker pairs in "strong LD", sorted by distance apart
set<LDPair,Pair_cmp> strongPairs;
map<int2,DPrime> dpStore;
int numStrong = 0;
int numRec = 0;
int numInGroup = 0;
// Each pair of markers
for (int x = fromPosition; x < toPosition; x++)
{
if ( ! par::silent )
{
std::cerr << "Chromosome " << locus[x]->chr
<< ", position " << locus[x]->bp/1000000.0
<< "Mb \r";
}
for (int y = x+1; y <= toPosition; y++)
{
if ( locus[x]->chr != locus[y]->chr )
continue;
if ( ( locus[y]->bp - locus[x]->bp ) > par::disp_r_window_kb )
{
continue;
}
if ( locus[x]->freq == 0 || locus[y]->freq == 0 )
continue;
PairwiseLinkage thisPair(x,y);
thisPair.calculateLD();
thisPair.calculateCI();
double lod = thisPair.lod;
double lowCI = thisPair.dp_lower;
double highCI = thisPair.dp_upper;
int2 t(x,y);
DPrime d;
d.dp = thisPair.dp;
d.dpl = lowCI;
d.dpu = highCI;
d.lod = lod;
dpStore.insert( make_pair( t,d ) );
// Is this pair in strong LD?
if (lod < -90) continue; //missing data
if (highCI < cutHighCI || lowCI < cutLowCI)
continue; //must pass "strong LD" test
// Store this pair
LDPair p(x,y, abs( locus[x]->bp - locus[y]->bp ) );
strongPairs.insert( p );
}
}
// Now we have a list of SNPs in strong LD within this region
// Now construct blocks based on this
set<int> used;
// #blocks:
vector<vector<int> > blockArray;
int cnt = 0;
for ( set<LDPair>::reverse_iterator i = strongPairs.rbegin();
i != strongPairs.rend();
++i )
{
int numStrong = 0;
int numRec = 0;
int numInGroup = 0;
vector<int> thisBlock;
int first = i->s1;
int last = i->s2;
long sep = i->dist;
// See if this block overlaps with another:
if ( used.find(first) != used.end()
|| used.find(last) != used.end() )
{
continue;
}
// Next, count the number of markers in the block.
// (nb. assume all SNPs belong)
for (int x = first; x <=last ; x++)
{
if( !skipMarker[x] )
numInGroup++;
}
// Skip it if it is too long in bases for it's size in markers
if (numInGroup < 4 && sep > maxDist[numInGroup])
{
continue;
}
// Add first SNP
thisBlock.push_back( first );
// Test block: requires 95% of informative markers to be "strong"
for (int y = first+1; y <= last; y++)
{
if (skipMarker[y])
{
continue;
}
thisBlock.push_back(y);
//loop over columns in row y
for (int x = first; x < y; x++)
{
if (skipMarker[x])
continue;
double lod;
double lowCI;
double highCI;
map<int2,DPrime>::iterator l = dpStore.find( int2(x,y) );
if ( l == dpStore.end() )
{
// Recalculate
PairwiseLinkage thisPair(x,y);
thisPair.calculateLD();
thisPair.calculateCI();
lod = thisPair.lod;
lowCI = thisPair.dp_lower;
highCI = thisPair.dp_upper;
}
else
{
// Get the right bits
lod = l->second.lod;
lowCI = l->second.dpl;
highCI = l->second.dpu;
}
// Monomorphic marker error
if ( lod < -90)
continue;
// Skip bad markers
if ( lod == 0 && lowCI == 0 && highCI == 0)
continue;
// For small blocks use different CI cutoffs
if (numInGroup < 5)
{
if (lowCI > cutLowCIVar[numInGroup] && highCI >= cutHighCI)
numStrong++;
}
else
{
if (lowCI > cutLowCI && highCI >= cutHighCI)
numStrong++; //strong LD
}
if (highCI < recHighCI)
numRec++; //recombination
}
}
// Change the definition somewhat for small blocks
if (numInGroup > 3)
{
if (numStrong + numRec < 6)
{
continue;
}
}
else if (numInGroup > 2)
{
if (numStrong + numRec < 3)
{
continue;
}
}
else
{
if (numStrong + numRec < 1)
{
continue;
}
}
// If this qualifies as a block, add to the block list, but in
// order by first marker number:
if ( (double)numStrong/(double)(numStrong + numRec) > informFrac)
{
blocks.insert( make_pair( first , thisBlock ));
// Track that these SNPs belong to a block
for (int u = first; u <= last; u++)
used.insert(u);
}
}
// Next chromosome
}
if ( ! par::silent )
cerr << "\n";
map<int,vector<int> >::iterator j = blocks.begin();
printLOG(int2str( blocks.size() )
+ " blocks called, writing list to [ "
+ par::output_file_name + ".blocks ]\n");
ofstream O1( (par::output_file_name+".blocks").c_str() , ios::out );
printLOG("Writing extra block details to [ " +
par::output_file_name + ".blocks.det ]\n");
ofstream O2( (par::output_file_name+".blocks.det").c_str() , ios::out );
O2 << setw(4) << "CHR" << " "
<< setw(12) << "BP1" << " "
<< setw(12) << "BP2" << " "
<< setw(12) << "KB" << " "
<< setw(6) << "NSNPS" << " "
<< setw(4) << "SNPS" << "\n";
while ( j != blocks.end() )
{
O1 << "*";
vector<int> & b = j->second;
for (int k=0; k<b.size(); k++)
O1 << " " << PP->locus[b[k]]->name;
O1 << "\n";
O2 << setw(4) << PP->locus[b[0]]->chr << " "
<< setw(12) << PP->locus[b[0]]->bp << " "
<< setw(12) << PP->locus[b[b.size()-1]]->bp << " "
<< setw(12) << (double)(PP->locus[b[b.size()-1]]->bp - PP->locus[b[0]]->bp + 1)/1000.0 << " "
<< setw(6) << b.size() << " ";
for (int k=0; k<b.size(); k++)
{
if ( k>0 )
O2 << "|" << PP->locus[b[k]]->name;
else
O2 << PP->locus[b[k]]->name;
}
O2 << "\n";
++j;
}
O1.close();
O2.close();
// List of blocks created here
// (dummy; not used)
map<Range,vector<int> > blocks0;
return blocks0;
}
void fillFiberSetContainerFromPositionsAndConnections(FiberSetContainer& outFiberSetContainer, const XMarkerList& inLinePositions, const IndexPairList& inLineConnections)
{
outFiberSetContainer.deleteAllFiberSets();
std::set<int> typeIDsSet;
std::map<int, int> numPositionsPerType;
std::map<int, std::string> typeLabels;
// Step 1: Scan all positions and get all type ID from them
for (XMarkerList::const_iterator it = inLinePositions.cbegin(); it != inLinePositions.cend(); ++it)
{
XMarker thisMarker = *it;
typeIDsSet.insert(thisMarker.type);
numPositionsPerType[thisMarker.type]++;
typeLabels[thisMarker.type] = thisMarker.name();
}
// Step 2: Now create fibers
for (std::set<int>::const_iterator typeIdIterator = typeIDsSet.cbegin(); typeIdIterator != typeIDsSet.cend(); ++typeIdIterator)
{
int thisTypeID = *typeIdIterator;
FiberSetContainer::FiberSet* newFiberSet = outFiberSetContainer.createFiberSet();
newFiberSet->setColor(Vector3(1, 0, 0));
newFiberSet->setLabel(typeLabels[thisTypeID]);
// Add all connections to the temporary connections list if type ID matches
XMarkerList workPositions;
for (XMarkerList::const_iterator outPositionsIterator = inLinePositions.cbegin(); outPositionsIterator != inLinePositions.cend(); ++outPositionsIterator)
{
XMarker thisMarker = *outPositionsIterator;
if (thisMarker.type == thisTypeID)
{
workPositions.push_back(thisMarker);
}
}
MLint workPositionsSize = (MLint)workPositions.size();
// Add all connections to the temporary connections list if type ID matches
IndexPairList workConnections;
for (IndexPairList::const_iterator outConnectionsIterator = inLineConnections.cbegin(); outConnectionsIterator != inLineConnections.cend(); ++outConnectionsIterator)
{
IndexPair thisPair = *outConnectionsIterator;
if (thisPair.type == thisTypeID)
{
workConnections.push_back(thisPair);
}
}
// If temporary connections list is still empty at this point: create default list
if (workConnections.size() == 0)
{
int numPositionsOfThisType = numPositionsPerType[thisTypeID];
for (int p = 0; p < numPositionsOfThisType - 1; p++)
{
IndexPair thisPair(p, p + 1);
workConnections.push_back(thisPair);
}
}
// Now finally create fibers from connections
for (IndexPairList::const_iterator workListIterator = workConnections.cbegin(); workListIterator != workConnections.cend(); ++workListIterator)
{
IndexPair thisWorkPair = *workListIterator;
MLint startIndex = thisWorkPair.index1;
MLint endIndex = thisWorkPair.index2;
if ((startIndex < workPositionsSize) && (endIndex < workPositionsSize))
{
XMarker startMarker = workPositions[startIndex];
XMarker endMarker = workPositions[endIndex];
FiberSetContainer::FiberSet::Fiber* newFiber = newFiberSet->createFiber();
FiberSetContainer::FiberSet::Fiber::FiberPoint startPoint;
startPoint.setCoordinates(startMarker.x(), startMarker.y(), startMarker.z());
FiberSetContainer::FiberSet::Fiber::FiberPoint endPoint;
endPoint.setCoordinates(endMarker.x(), endMarker.y(), endMarker.z());
newFiber->appendPoint(startPoint);
newFiber->appendPoint(endPoint);
newFiber->setLabel(1.0);
}
}
workPositions.clearList();
workConnections.clearList();
}
}
