}
}
//**********************************************************************************************************************
+string SummarySharedCommand::getOutputFileNameTag(string type, string inputName=""){
+ try {
+ string outputFileName = "";
+ map<string, vector<string> >::iterator it;
+
+ //is this a type this command creates
+ it = outputTypes.find(type);
+ if (it == outputTypes.end()) { m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
+ else {
+ if (type == "summary") { outputFileName = "shared.summary"; }
+ else if (type == "phylip") { outputFileName = "dist"; }
+ else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true; }
+ }
+ return outputFileName;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SummarySharedCommand", "getOutputFileNameTag");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
SummarySharedCommand::SummarySharedCommand(){
try {
abort = true; calledHelp = true;
setParameters();
vector<string> tempOutNames;
outputTypes["summary"] = tempOutNames;
+ outputTypes["phylip"] = tempOutNames;
}
catch(exception& e) {
m->errorOut(e, "SummarySharedCommand", "SummarySharedCommand");
//initialize outputTypes
vector<string> tempOutNames;
outputTypes["summary"] = tempOutNames;
+ outputTypes["phylip"] = tempOutNames;
//if the user changes the input directory command factory will send this info to us in the output parameter
string inputDir = validParameter.validFile(parameters, "inputdir", false);
if (abort == true) { if (calledHelp) { return 0; } return 2; }
ofstream outputFileHandle, outAll;
- string outputFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + "shared.summary";
+ string outputFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + getOutputFileNameTag("summary");
//if the users entered no valid calculators don't execute command
if (sumCalculators.size() == 0) { return 0; }
outputFileHandle.close();
//create file and put column headers for multiple groups file
- string outAllFileName = ((m->getRootName(sharedfile)) + "sharedmultiple.summary");
+ string outAllFileName = ((m->getRootName(sharedfile)) + "multiple." + getOutputFileNameTag("summary"));
if (mult == true) {
m->openOutputFile(outAllFileName, outAll);
outputNames.push_back(outAllFileName);
return 0;
}
/******************************************************/
-
+ if (subsample) {
+ if (subsampleSize == -1) { //user has not set size, set size = smallest samples size
+ subsampleSize = lookup[0]->getNumSeqs();
+ for (int i = 1; i < lookup.size(); i++) {
+ int thisSize = lookup[i]->getNumSeqs();
+
+ if (thisSize < subsampleSize) { subsampleSize = thisSize; }
+ }
+ }else {
+ m->clearGroups();
+ Groups.clear();
+ vector<SharedRAbundVector*> temp;
+ for (int i = 0; i < lookup.size(); i++) {
+ if (lookup[i]->getNumSeqs() < subsampleSize) {
+ m->mothurOut(lookup[i]->getGroup() + " contains " + toString(lookup[i]->getNumSeqs()) + ". Eliminating."); m->mothurOutEndLine();
+ delete lookup[i];
+ }else {
+ Groups.push_back(lookup[i]->getGroup());
+ temp.push_back(lookup[i]);
+ }
+ }
+ lookup = temp;
+ m->setGroups(Groups);
+ }
+
+ if (lookup.size() < 2) { m->mothurOut("You have not provided enough valid groups. I cannot run the command."); m->mothurOutEndLine(); m->control_pressed = true; delete input; return 0; }
+ }
+
/******************************************************/
//comparison breakup to be used by different processes later
- numGroups = m->getNumGroups();
+ numGroups = lookup.size();
lines.resize(processors);
for (int i = 0; i < processors; i++) {
lines[i].start = int (sqrt(float(i)/float(processors)) * numGroups);
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(sumCalculators.size());
- for (int thisIter = 0; thisIter < iters; thisIter++) {
+ for (int thisIter = 0; thisIter < iters+1; thisIter++) {
vector<SharedRAbundVector*> thisItersLookup = thisLookup;
- if (subsample) {
+ if (subsample && (thisIter != 0)) { //we want the summary results for the whole dataset, then the subsampling
SubSample sample;
vector<string> tempLabels; //dont need since we arent printing the sampled sharedRabunds
#endif
}
- calcDistsTotals.push_back(calcDists);
- if (subsample) {
+ if (subsample && (thisIter != 0)) { //we want the summary results for the whole dataset, then the subsampling
+ calcDistsTotals.push_back(calcDists);
//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(); }
+ }else {
+ if (createPhylip) {
+ 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;
+ }
+
+ string distFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + sumCalculators[i]->getName() + "." + thisLookup[0]->getLabel() + "." + output + "." + getOutputFileNameTag("phylip");
+ outputNames.push_back(distFileName); outputTypes["phylip"].push_back(distFileName);
+ ofstream outDist;
+ m->openOutputFile(distFileName, outDist);
+ outDist.setf(ios::fixed, ios::floatfield); outDist.setf(ios::showpoint);
+
+ printSims(outDist, matrix);
+
+ outDist.close();
+ }
+ }
}
+ for (int i = 0; i < calcDists.size(); i++) { calcDists[i].clear(); }
}
if (iters != 1) {
stdmatrix[column][row] = stdDist;
}
- string distFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + sumCalculators[i]->getName() + "." + thisLookup[0]->getLabel() + "." + output + ".ave.dist";
+ string distFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + sumCalculators[i]->getName() + "." + thisLookup[0]->getLabel() + "." + output + ".ave." + getOutputFileNameTag("phylip");
outputNames.push_back(distFileName); outputTypes["phylip"].push_back(distFileName);
ofstream outAve;
m->openOutputFile(distFileName, outAve);
outAve.close();
- distFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + sumCalculators[i]->getName() + "." + thisLookup[0]->getLabel() + "." + output + ".std.dist";
+ distFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + sumCalculators[i]->getName() + "." + thisLookup[0]->getLabel() + "." + output + ".std." + getOutputFileNameTag("phylip");
outputNames.push_back(distFileName); outputTypes["phylip"].push_back(distFileName);
ofstream outSTD;
m->openOutputFile(distFileName, outSTD);
outSTD.close();
}
- }else {
- if (createPhylip) {
- 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;
- }
-
- string distFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + sumCalculators[i]->getName() + "." + thisLookup[0]->getLabel() + "." + output + ".dist";
- outputNames.push_back(distFileName); outputTypes["phylip"].push_back(distFileName);
- ofstream outDist;
- m->openOutputFile(distFileName, outDist);
- outDist.setf(ios::fixed, ios::floatfield); outDist.setf(ios::showpoint);
-
- printSims(outDist, matrix);
-
- outDist.close();
- }
- }
}
-
+
return 0;
}
catch(exception& e) {
projects / mothur.git / blobdiff