int main(int argc, char **argv)
{
if (argc != NUM_ARGS+1) {
usage();
}
srand(time(0));
Graph *graph = new Graph(argv[1]);
Dendrogram *dendrogram = new Dendrogram(graph,argv[2]);
int sampleSpread = atoi(argv[3]);
int sampleCount = atoi(argv[4]);
int sampleEvery = sampleSpread / sampleCount;
const char *outputPrefix = argv[5];
double threshold = atof(argv[6]);
graph->threshold = threshold;
DendrogramSet samples;
for (int sampleIndex = 0; sampleIndex < sampleSpread; sampleIndex++) {
dendrogram->sample();
if (sampleIndex % sampleEvery == 0) {
cout << "Sampling at " << sampleIndex << " / " << sampleSpread << endl;
samples.insert(new Dendrogram(dendrogram));
}
}
Consensus *consensus = new Consensus(samples,graph);
// Print to console
cout << consensus->toString() << endl;
// Save (in various forms) to disk
char filename[80];
sprintf(filename,"%s.consensus.txt",outputPrefix);
cerr << filename << endl;
ofstream stringFile(filename);
stringFile << consensus->toString() << endl;
stringFile.close();
sprintf(filename,"%s.consensus.dot",outputPrefix);
cerr << filename << endl;
ofstream dotFile(filename);
dotFile << consensus->toDot() << endl;
dotFile.close();
sprintf(filename,"%s.consensus.matrix",outputPrefix);
cerr << filename << endl;
ofstream matrixFile(filename);
matrixFile << consensus->toMatrix() << endl;
matrixFile.close();
}
int TreeGroupCommand::process(vector<SharedRAbundVector*> thisLookup) {
try{
vector< vector< vector<seqDist> > > calcDistsTotals; //each iter, one for each calc, then each groupCombos dists. this will be used to make .dist files
vector< vector<seqDist> > calcDists; calcDists.resize(treeCalculators.size());
for (int thisIter = 0; thisIter < iters; thisIter++) {
vector<SharedRAbundVector*> thisItersLookup = thisLookup;
if (subsample) {
SubSample sample;
vector<string> tempLabels; //dont need since we arent printing the sampled sharedRabunds
//make copy of lookup so we don't get access violations
vector<SharedRAbundVector*> newLookup;
for (int k = 0; k < thisItersLookup.size(); k++) {
SharedRAbundVector* temp = new SharedRAbundVector();
temp->setLabel(thisItersLookup[k]->getLabel());
temp->setGroup(thisItersLookup[k]->getGroup());
newLookup.push_back(temp);
}
//for each bin
for (int k = 0; k < thisItersLookup[0]->getNumBins(); k++) {
if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) { delete newLookup[j]; } return 0; }
for (int j = 0; j < thisItersLookup.size(); j++) { newLookup[j]->push_back(thisItersLookup[j]->getAbundance(k), thisItersLookup[j]->getGroup()); }
}
tempLabels = sample.getSample(newLookup, subsampleSize);
thisItersLookup = newLookup;
}
if(processors == 1){
driver(thisItersLookup, 0, numGroups, calcDists);
}else{
int process = 1;
vector<int> processIDS;
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
//loop through and create all the processes you want
while (process != processors) {
pid_t pid = fork();
if (pid > 0) {
processIDS.push_back(pid);
process++;
}else if (pid == 0){
driver(thisItersLookup, lines[process].start, lines[process].end, calcDists);
string tempdistFileName = m->getRootName(m->getSimpleName(sharedfile)) + m->mothurGetpid(process) + ".dist";
ofstream outtemp;
m->openOutputFile(tempdistFileName, outtemp);
for (int i = 0; i < calcDists.size(); i++) {
outtemp << calcDists[i].size() << endl;
for (int j = 0; j < calcDists[i].size(); j++) {
outtemp << calcDists[i][j].seq1 << '\t' << calcDists[i][j].seq2 << '\t' << calcDists[i][j].dist << endl;
}
}
outtemp.close();
exit(0);
}else {
m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
exit(0);
}
}
//parent do your part
driver(thisItersLookup, lines[0].start, lines[0].end, calcDists);
//force parent to wait until all the processes are done
for (int i = 0; i < processIDS.size(); i++) {
int temp = processIDS[i];
wait(&temp);
}
for (int i = 0; i < processIDS.size(); i++) {
string tempdistFileName = m->getRootName(m->getSimpleName(sharedfile)) + toString(processIDS[i]) + ".dist";
ifstream intemp;
m->openInputFile(tempdistFileName, intemp);
for (int k = 0; k < calcDists.size(); k++) {
int size = 0;
intemp >> size; m->gobble(intemp);
for (int j = 0; j < size; j++) {
int seq1 = 0;
int seq2 = 0;
float dist = 1.0;
intemp >> seq1 >> seq2 >> dist; m->gobble(intemp);
seqDist tempDist(seq1, seq2, dist);
calcDists[k].push_back(tempDist);
}
}
intemp.close();
m->mothurRemove(tempdistFileName);
}
#else
//////////////////////////////////////////////////////////////////////////////////////////////////////
//Windows version shared memory, so be careful when passing variables through the treeSharedData struct.
//Above fork() will clone, so memory is separate, but that's not the case with windows,
//Taking advantage of shared memory to pass results vectors.
//////////////////////////////////////////////////////////////////////////////////////////////////////
vector<treeSharedData*> pDataArray;
DWORD dwThreadIdArray[processors-1];
HANDLE hThreadArray[processors-1];
//Create processor worker threads.
for( int i=1; i<processors; i++ ){
//make copy of lookup so we don't get access violations
vector<SharedRAbundVector*> newLookup;
for (int k = 0; k < thisItersLookup.size(); k++) {
SharedRAbundVector* temp = new SharedRAbundVector();
temp->setLabel(thisItersLookup[k]->getLabel());
temp->setGroup(thisItersLookup[k]->getGroup());
newLookup.push_back(temp);
}
//for each bin
for (int k = 0; k < thisItersLookup[0]->getNumBins(); k++) {
if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) { delete newLookup[j]; } return 0; }
for (int j = 0; j < thisItersLookup.size(); j++) { newLookup[j]->push_back(thisItersLookup[j]->getAbundance(k), thisItersLookup[j]->getGroup()); }
}
// Allocate memory for thread data.
treeSharedData* tempSum = new treeSharedData(m, lines[i].start, lines[i].end, Estimators, newLookup);
pDataArray.push_back(tempSum);
processIDS.push_back(i);
hThreadArray[i-1] = CreateThread(NULL, 0, MyTreeSharedThreadFunction, pDataArray[i-1], 0, &dwThreadIdArray[i-1]);
}
//parent do your part
driver(thisItersLookup, lines[0].start, lines[0].end, calcDists);
//Wait until all threads have terminated.
WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
//Close all thread handles and free memory allocations.
for(int i=0; i < pDataArray.size(); i++){
if (pDataArray[i]->count != (pDataArray[i]->end-pDataArray[i]->start)) {
m->mothurOut("[ERROR]: process " + toString(i) + " only processed " + toString(pDataArray[i]->count) + " of " + toString(pDataArray[i]->end-pDataArray[i]->start) + " groups assigned to it, quitting. \n"); m->control_pressed = true;
}
for (int j = 0; j < pDataArray[i]->thisLookup.size(); j++) { delete pDataArray[i]->thisLookup[j]; }
for (int k = 0; k < calcDists.size(); k++) {
int size = pDataArray[i]->calcDists[k].size();
for (int j = 0; j < size; j++) { calcDists[k].push_back(pDataArray[i]->calcDists[k][j]); }
}
CloseHandle(hThreadArray[i]);
delete pDataArray[i];
}
#endif
}
calcDistsTotals.push_back(calcDists);
if (subsample) {
//clean up memory
for (int i = 0; i < thisItersLookup.size(); i++) { delete thisItersLookup[i]; }
thisItersLookup.clear();
for (int i = 0; i < calcDists.size(); i++) { calcDists[i].clear(); }
}
if (m->debug) { m->mothurOut("[DEBUG]: iter = " + toString(thisIter) + ".\n"); }
}
if (m->debug) { m->mothurOut("[DEBUG]: done with iters.\n"); }
if (iters != 1) {
//we need to find the average distance and standard deviation for each groups distance
vector< vector<seqDist> > calcAverages = m->getAverages(calcDistsTotals);
if (m->debug) { m->mothurOut("[DEBUG]: found averages.\n"); }
//create average tree for each calc
for (int i = 0; i < calcDists.size(); i++) {
vector< vector<double> > matrix; //square matrix to represent the distance
matrix.resize(thisLookup.size());
for (int k = 0; k < thisLookup.size(); k++) { matrix[k].resize(thisLookup.size(), 0.0); }
for (int j = 0; j < calcAverages[i].size(); j++) {
int row = calcAverages[i][j].seq1;
int column = calcAverages[i][j].seq2;
float dist = calcAverages[i][j].dist;
matrix[row][column] = dist;
matrix[column][row] = dist;
}
//create a new filename
map<string, string> variables;
variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(inputfile));
variables["[calc]"] = treeCalculators[i]->getName();
variables["[distance]"] = thisLookup[0]->getLabel();
variables["[tag]"] = "ave";
string outputFile = getOutputFileName("tree",variables);
outputNames.push_back(outputFile); outputTypes["tree"].push_back(outputFile);
//creates tree from similarity matrix and write out file
Tree* newTree = createTree(matrix);
if (newTree != NULL) { writeTree(outputFile, newTree); }
}
if (m->debug) { m->mothurOut("[DEBUG]: done averages trees.\n"); }
//create all trees for each calc and find their consensus tree
for (int i = 0; i < calcDists.size(); i++) {
if (m->control_pressed) { break; }
//create a new filename
//create a new filename
map<string, string> variables;
variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(inputfile));
variables["[calc]"] = treeCalculators[i]->getName();
variables["[distance]"] = thisLookup[0]->getLabel();
variables["[tag]"] = "all";
string outputFile = getOutputFileName("tree",variables);
outputNames.push_back(outputFile); outputTypes["tree"].push_back(outputFile);
ofstream outAll;
m->openOutputFile(outputFile, outAll);
vector<Tree*> trees;
for (int myIter = 0; myIter < iters; myIter++) {
if(m->control_pressed) { break; }
//initialize matrix
vector< vector<double> > matrix; //square matrix to represent the distance
matrix.resize(thisLookup.size());
for (int k = 0; k < thisLookup.size(); k++) { matrix[k].resize(thisLookup.size(), 0.0); }
for (int j = 0; j < calcDistsTotals[myIter][i].size(); j++) {
int row = calcDistsTotals[myIter][i][j].seq1;
int column = calcDistsTotals[myIter][i][j].seq2;
double dist = calcDistsTotals[myIter][i][j].dist;
matrix[row][column] = dist;
matrix[column][row] = dist;
}
//creates tree from similarity matrix and write out file
Tree* newTree = createTree(matrix);
if (newTree != NULL) {
newTree->print(outAll);
trees.push_back(newTree);
}
}
outAll.close();
if (m->control_pressed) { for (int k = 0; k < trees.size(); k++) { delete trees[k]; } }
if (m->debug) { m->mothurOut("[DEBUG]: done all trees.\n"); }
Consensus consensus;
//clear old tree names if any
m->Treenames.clear(); m->Treenames = m->getGroups(); //may have changed if subsample eliminated groups
Tree* conTree = consensus.getTree(trees);
if (m->debug) { m->mothurOut("[DEBUG]: done cons tree.\n"); }
//create a new filename
variables["[tag]"] = "cons";
string conFile = getOutputFileName("tree",variables);
outputNames.push_back(conFile); outputTypes["tree"].push_back(conFile);
ofstream outTree;
m->openOutputFile(conFile, outTree);
if (conTree != NULL) { conTree->print(outTree, "boot"); delete conTree; }
}
}else {
for (int i = 0; i < calcDists.size(); i++) {
if (m->control_pressed) { break; }
//initialize matrix
vector< vector<double> > matrix; //square matrix to represent the distance
matrix.resize(thisLookup.size());
for (int k = 0; k < thisLookup.size(); k++) { matrix[k].resize(thisLookup.size(), 0.0); }
for (int j = 0; j < calcDists[i].size(); j++) {
int row = calcDists[i][j].seq1;
int column = calcDists[i][j].seq2;
double dist = calcDists[i][j].dist;
matrix[row][column] = dist;
matrix[column][row] = dist;
}
//create a new filename
map<string, string> variables;
variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(inputfile));
variables["[calc]"] = treeCalculators[i]->getName();
variables["[distance]"] = thisLookup[0]->getLabel();
variables["[tag]"] = "";
string outputFile = getOutputFileName("tree",variables);
outputNames.push_back(outputFile); outputTypes["tree"].push_back(outputFile);
//creates tree from similarity matrix and write out file
Tree* newTree = createTree(matrix);
if (newTree != NULL) { writeTree(outputFile, newTree); delete newTree; }
}
}
return 0;
}
catch(exception& e) {
m->errorOut(e, "TreeGroupCommand", "process");
exit(1);
}
}
int main(int argc, char* argv[]) {
// initialise random
// Random::Random();
string SampleName;
string ChainName;
string OutGroupFile;
string Path = "";
int catgtrpoint = 0;
int rates = 0;
int modes = 0;
int sitestat = 0;
int grepmode = 0;
int coaff = 0;
int verbose = 0;
int extract = 0;
int burnin = 0;
int every = 1;
int until = -1;
int ncat = 20;
int bf = 0;
int rr = 0;
int step = 10;
int save = 0;
int truelength = 0;
int sitebranch = 0;
int clock = 0;
int ps = 0;
int clustermodes = 0;
double mindist = 0.03;
int minsize = 10;
double cutoff = 0.5;
int nrep = 1;
int popeff = 0;
int branchfreqs = 0;
int benchmark = 0;
int intlnl = 0;
double entropytimemax = 0;
int entropyN = 0;
int constcorrect = 0;
double rho = 1;
int sitelogl = 0;
// read arguments
try {
if (argc == 1) {
throw(0);
}
int i = 1;
while (i < argc) {
string s = argv[i];
if (s == "-c") {
i++;
if (i == argc) throw(0);
if (! IsFloat(argv[i])) throw(0);
cutoff = atof(argv[i]);
}
else if (s == "-v") {
verbose = 2;
}
else if (s == "-rr") {
rr = 1;
}
else if (s == "-lnl") {
intlnl = 1;
}
else if (s == "-sitelogl") {
sitelogl = 1;
}
else if (s == "-cst") {
constcorrect = 1;
i++;
rho = atof(argv[i]);
}
else if (s == "-finitetime") {
i++;
entropytimemax = atof(argv[i]);
i++;
entropyN = atoi(argv[i]);
}
else if (s == "-catcoaff") {
coaff = 1;
}
else if (s == "-ratecoaff") {
coaff = 2;
}
else if (s == "-mblcoaff") {
coaff = 3;
}
else if (s == "-popeff") {
popeff = 1;
}
else if (s == "-branchfreqs") {
branchfreqs = 1;
}
else if (s == "-step") {
i++;
step = atoi(argv[i]);
}
else if (s == "-div") {
clock = 1;
}
else if (s == "-p") {
i++;
Path = argv[i];
}
else if (s == "-ps") {
ps = 1;
}
else if (s == "-bf") {
bf = 1;
}
else if (s == "-l") {
truelength = 1;
}
else if (s == "-ll") {
truelength = 1;
sitebranch = 1;
}
else if (s == "-gm") {
grepmode = 1;
}
else if (s == "-bench") {
benchmark = 1;
}
else if (s == "-nrep") {
i++;
if (i == argc) throw(0);
s = argv[i];
if (! IsInt(s)) {
throw(0);
}
nrep = atoi(argv[i]);
}
else if (s == "-s") {
save = 1;
}
else if (s == "-cl") {
clustermodes = 1;
}
else if ((s == "-sz") || (s == "-ms")) {
i++;
if (i == argc) throw(0);
s = argv[i];
if (! IsInt(s)) {
throw(0);
}
minsize = atoi(argv[i]);
}
else if ((s == "-ds") || (s == "-md")) {
i++;
if (i == argc) throw(0);
s = argv[i];
if (! IsFloat(s)) {
throw(0);
}
mindist = atof(argv[i]);
}
else if (s == "-ncat") {
i++;
if (i == argc) throw(0);
s = argv[i];
if (! IsInt(s)) {
throw(0);
}
ncat = atoi(argv[i]);
}
else if ( (s == "-m") || (s == "-modes") ) {
modes = 1;
}
else if ( (s == "-r") || (s == "-rates") ) {
rates = 1;
}
else if ( (s == "-ss") || (s == "-sitestat") ) {
sitestat = 1;
}
else if (s == "-catgtrpoint") {
catgtrpoint = 1;
}
else if ( (s == "-x") || (s == "-extract") ) {
extract = 1;
i++;
if (i == argc) throw(0);
s = argv[i];
if (! IsInt(s)) {
throw(0);
}
burnin = atoi(argv[i]);
i++;
if (i == argc) throw(0);
s = argv[i];
if (IsInt(s)) {
every = atoi(argv[i]);
i++;
if (i == argc) throw(0);
s = argv[i];
if (IsInt(s)) {
until = atoi(argv[i]);
}
else {
i--;
}
}
else {
i--;
}
}
else {
if (i != (argc -1)) {
throw(0);
}
ChainName = argv[i];
}
i++;
}
if ((SampleName == "") && (ChainName == "")) {
throw(0);
}
}
catch(...) {
cerr << "readpb [-x <burnin> <every> <until>] <chainname> \n";
cerr << '\n';
cerr << "\tdefaults : burnin = 0, every = 1, until the end\n";
cerr << '\n';
cerr << "additional options:\n";
cerr << "\t-c <cutoff> : collapses all groups with posterior probability lower than cutoff\n";
cerr << "\t-m : posterior distribution of the number of modes\n";
cerr << "\t-ss : mean posterior site-specific stationaries\n";
cerr << "\t-r : mean posterior site-specific rates (continuous gamma only)\n";
cerr << '\n';
cerr << "\t-ncat <n> : defines number of bins for rate histogram (default 20)\n";
cerr << '\n';
cerr << "\t-cl : mode clustering\n";
cerr << "\t\t-ms: cluster min size (default : 10)\n";
cerr << "\t\t-md: aggregating distance threshold (default : 0.03)\n";
cerr << '\n';
cerr << "\t-ps : postscript output for tree (requires LateX), or for site-specific profiles\n";
cerr << '\n';
exit(1);
}
if (SampleName == "") {
SampleName = ChainName + "_sample";
}
try {
string name = ChainName + ".param";
MCParameters* mParam2 = 0;
if (ifstream(name.c_str())) {
mParam2 = new MCParameters;
ifstream param_is(name.c_str());
param_is >> *mParam2;
}
if (mParam2 && mParam2->NormalApprox) {
clock = 1;
}
if ((! catgtrpoint) && (! truelength) && (! sitestat) && (!modes) && (!rates) && (! clock) && (! clustermodes) && ((!mParam2) || (! mParam2->FixTopo)) ) {
string name = ChainName + ".treelist";
if (! ifstream(name.c_str())) {
name = ChainName;
if (! ifstream(name.c_str())) {
cerr << "error: non existing chain\n";
exit(1);
}
}
BipartitionList bplist(name,burnin,every,until);
if (!bplist.Ntree) {
cerr << "empty tree list\n";
exit(1);
}
ofstream bplist_os((SampleName + ".bplist").c_str());
bplist.WriteToStream(bplist_os);
bplist_os.close();
cout << bplist.Ntree << " trees were read\n";
cout.flush();
Consensus* cons = new Consensus(&bplist,cutoff);
ofstream cons_os((SampleName + ".con.tre").c_str());
// cons->Trichotomise();
cons->Phylip(cons_os, 1, 1, 1, 0);
cons_os.close();
cerr << '\n';
cerr << "bipartition list in\t" << SampleName << ".bplist\n";
cerr << "consensus in\t\t" << SampleName << ".con.tre\n";
if (ps) {
cons->ToPS((string) (SampleName + ".con.tre"),12,20,1,1,1,0);
cerr << "postscript in\t\t" << SampleName << ".con.tre.ps\n";
}
cerr << '\n';
if ((! mParam2) || (!mParam2->SaveAll)) {
exit(1);
}
}
// delete mParam2;
Sample* sample = new Sample(ChainName,burnin,every,until,Path);
if (save) {
sample->ToFile(SampleName);
}
MCParameters* mParam = sample->GetParameters();
// initialising
int Nsite = mParam->Nsite;
int Ntaxa = mParam->Ntaxa;
cout << '\n';
cout << "Nsite : " << Nsite << '\n';
cout << "Ntaxa : " << Ntaxa << '\n';
cout << sample->GetSize() << " points to read\n";
cout << '\n';
cout.flush();
if (mParam->NormalApprox) {
if (clustermodes) {
cerr << "no mode clustering under normal approx\n";
cerr << '\n';
exit(1);
}
if (modes) {
cerr << "no mode analysis under normal approx\n";
cerr << '\n';
exit(1);
}
if (rates) {
cerr << "no rate analysis under normal approx\n";
cerr << '\n';
exit(1);
}
if (sitestat) {
cerr << "no site-specific profile analysis under normal approx\n";
cerr << '\n';
exit(1);
}
}
if (clock) {
sample->Dating(ps);
sample->Reset();
}
else if (sitelogl) {
sample->ReadSiteLogLikelihood();
}
else if (intlnl) {
sample->ReadSummedLogLikelihood();
}
else if (entropyN) {
sample->ReadFiniteTimeEntropy(entropytimemax,entropyN);
}
else if (catgtrpoint) {
sample->ReadCATGTRPoint();
}
else if (popeff) {
sample->ReadPopEff(ps);
}
else if (branchfreqs) {
sample->ReadBranchFreqs();
}
else if (coaff) {
sample->Coaff(coaff,step);
}
else if (rr) {
sample->MeanRR();
}
else if (benchmark) {
sample->ReadBench();
}
else if (clustermodes) {
if (! mParam->SaveAll) {
cerr << "error : biochemical profiles were not saved. cannot cluster them\n";
exit(1);
}
sample->ClusterModes(minsize, mindist, 0);
}
else if (grepmode) {
sample->GrepModes();
}
else if (truelength) {
sample->TrueLength(nrep,sitebranch);
}
else if (constcorrect) {
sample->ConstantSiteCorrection(rho);
}
else {
sample->Read(rates, modes, sitestat, cutoff, ncat, ps);
}
}
int TreeGroupCommand::createProcesses(SharedRAbundVectors*& thisLookup, CountTable& ct){
try {
vector<string> groupNames = thisLookup->getNamesGroups();
Treenames = groupNames; //may have changed if subsample eliminated groups
vector<int> lines;
if (processors > (iters+1)) { processors = iters+1; }
//figure out how many sequences you have to process
int numItersPerProcessor = (iters+1) / processors;
for (int i = 0; i < processors; i++) {
if(i == (processors - 1)){ numItersPerProcessor = (iters+1) - i * numItersPerProcessor; }
lines.push_back(numItersPerProcessor);
}
//create array of worker threads
vector<thread*> workerThreads;
vector<treeSharedData*> data;
//Lauch worker threads
for (int i = 0; i < processors-1; i++) {
//make copy of lookup so we don't get access violations
SharedRAbundVectors* newLookup = new SharedRAbundVectors(*thisLookup);
treeSharedData* dataBundle = new treeSharedData(lines[i+1], false, subsample, withReplacement, subsampleSize, Estimators, newLookup);
data.push_back(dataBundle);
workerThreads.push_back(new thread(process, dataBundle));
}
//make copy of lookup so we don't get access violations
SharedRAbundVectors* newLookup = new SharedRAbundVectors(*thisLookup);
treeSharedData* dataBundle = new treeSharedData(lines[0], true, subsample, withReplacement, subsampleSize, Estimators, newLookup);
process(dataBundle);
delete newLookup;
Estimators.clear(); Estimators = dataBundle->Estimators;
vector< vector< vector<seqDist> > > calcDistsTotals = dataBundle->calcDistsTotals;
vector< vector< vector<double> > > matrices = dataBundle->matrices;
for (int i = 0; i < processors-1; i++) {
workerThreads[i]->join();
//get calcDistsTotal info - one entry per iter
for (int j = 0; j < data[i]->calcDistsTotals.size(); j++) { calcDistsTotals.push_back(data[i]->calcDistsTotals[j]); }
delete data[i]->thisLookup;
delete data[i];
delete workerThreads[i];
}
delete dataBundle;
if (iters != 1) {
//we need to find the average distance and standard deviation for each groups distance
vector< vector<seqDist> > calcAverages = util.getAverages(calcDistsTotals);
if (m->getDebug()) { m->mothurOut("[DEBUG]: found averages.\n"); }
//create average tree for each calc
for (int i = 0; i < Estimators.size(); i++) {
vector< vector<double> > matrix; //square matrix to represent the distance
matrix.resize(thisLookup->size());
for (int k = 0; k < thisLookup->size(); k++) { matrix[k].resize(thisLookup->size(), 0.0); }
for (int j = 0; j < calcAverages[i].size(); j++) {
int row = calcAverages[i][j].seq1;
int column = calcAverages[i][j].seq2;
float dist = calcAverages[i][j].dist;
matrix[row][column] = dist;
matrix[column][row] = dist;
}
//create a new filename
map<string, string> variables;
variables["[filename]"] = outputDir + util.getRootName(util.getSimpleName(inputfile));
variables["[calc]"] = Estimators[i];
variables["[distance]"] = thisLookup->getLabel();
variables["[tag]"] = "ave";
string outputFile = getOutputFileName("tree",variables);
outputNames.push_back(outputFile); outputTypes["tree"].push_back(outputFile);
//creates tree from similarity matrix and write out file
Tree* newTree = new Tree(&ct, matrix, Treenames);
if (m->getControl_pressed()) { delete newTree; newTree = NULL; }
else { newTree->assembleTree(); }
if (newTree != NULL) { newTree->createNewickFile(outputFile); delete newTree; }
}
if (m->getDebug()) { m->mothurOut("[DEBUG]: done averages trees.\n"); }
//create all trees for each calc and find their consensus tree
for (int i = 0; i < Estimators.size(); i++) {
if (m->getControl_pressed()) { break; }
//create a new filename
map<string, string> variables;
variables["[filename]"] = outputDir + util.getRootName(util.getSimpleName(inputfile));
variables["[calc]"] = Estimators[i];
variables["[distance]"] = thisLookup->getLabel();
variables["[tag]"] = "all";
string outputFile = getOutputFileName("tree",variables);
outputNames.push_back(outputFile); outputTypes["tree"].push_back(outputFile);
ofstream outAll;
util.openOutputFile(outputFile, outAll);
vector<Tree*> trees;
for (int myIter = 0; myIter < iters; myIter++) {
if(m->getControl_pressed()) { break; }
//initialize matrix
vector< vector<double> > matrix; //square matrix to represent the distance
matrix.resize(thisLookup->size());
for (int k = 0; k < thisLookup->size(); k++) { matrix[k].resize(thisLookup->size(), 0.0); }
for (int j = 0; j < calcDistsTotals[myIter][i].size(); j++) {
int row = calcDistsTotals[myIter][i][j].seq1;
int column = calcDistsTotals[myIter][i][j].seq2;
double dist = calcDistsTotals[myIter][i][j].dist;
matrix[row][column] = dist;
matrix[column][row] = dist;
}
//creates tree from similarity matrix and write out file
Tree* newTree = new Tree(&ct, matrix, Treenames);
if (m->getControl_pressed()) { delete newTree; newTree = NULL; }
else { newTree->assembleTree(); }
if (newTree != NULL) {
newTree->print(outAll);
trees.push_back(newTree);
}
}
outAll.close();
if (m->getControl_pressed()) { for (int k = 0; k < trees.size(); k++) { delete trees[k]; } }
if (m->getDebug()) { m->mothurOut("[DEBUG]: done all trees.\n"); }
Consensus consensus;
Tree* conTree = consensus.getTree(trees);
if (m->getDebug()) { m->mothurOut("[DEBUG]: done cons tree.\n"); }
//create a new filename
variables["[tag]"] = "cons";
string conFile = getOutputFileName("tree",variables);
outputNames.push_back(conFile); outputTypes["tree"].push_back(conFile);
ofstream outTree;
util.openOutputFile(conFile, outTree);
if (conTree != NULL) { conTree->print(outTree, "boot"); delete conTree; }
}
}else {
for (int i = 0; i < matrices.size(); i++) {
if (m->getControl_pressed()) { break; }
//initialize matrix
vector< vector<double> > matrix = matrices[i]; //square matrix to represent the distance
//create a new filename
map<string, string> variables;
variables["[filename]"] = outputDir + util.getRootName(util.getSimpleName(inputfile));
variables["[calc]"] = Estimators[i];
variables["[distance]"] = thisLookup->getLabel();
variables["[tag]"] = "";
string outputFile = getOutputFileName("tree",variables);
outputNames.push_back(outputFile); outputTypes["tree"].push_back(outputFile);
//creates tree from similarity matrix and write out file
Tree* newTree = new Tree(&ct, matrix, Treenames);
if (m->getControl_pressed()) { delete newTree; newTree = NULL; }
else { newTree->assembleTree(); }
if (newTree != NULL) { newTree->createNewickFile(outputFile); delete newTree; }
}
}
return 0;
}
catch(exception& e) {
m->errorOut(e, "TreeGroupCommand", "createProcesses");
exit(1);
}
}