X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=matrixoutputcommand.cpp;h=6537963bb58e034ef00da3812cc6af2bb167be71;hb=b9800d7b05252b911b10d36febfb5e6da3271766;hp=effc368e0c7f932b719925a0afb349b5afee4bdf;hpb=ca9ac1d80c62f57270b0dcd49410ebe08a8aecd6;p=mothur.git diff --git a/matrixoutputcommand.cpp b/matrixoutputcommand.cpp index effc368..6537963 100644 --- a/matrixoutputcommand.cpp +++ b/matrixoutputcommand.cpp @@ -8,47 +8,6 @@ */ #include "matrixoutputcommand.h" -#include "sharedsobscollectsummary.h" -#include "sharedchao1.h" -#include "sharedace.h" -#include "sharednseqs.h" -#include "sharedjabund.h" -#include "sharedsorabund.h" -#include "sharedjclass.h" -#include "sharedsorclass.h" -#include "sharedjest.h" -#include "sharedsorest.h" -#include "sharedthetayc.h" -#include "sharedthetan.h" -#include "sharedkstest.h" -#include "whittaker.h" -#include "sharedochiai.h" -#include "sharedanderbergs.h" -#include "sharedkulczynski.h" -#include "sharedkulczynskicody.h" -#include "sharedlennon.h" -#include "sharedmorisitahorn.h" -#include "sharedbraycurtis.h" -#include "sharedjackknife.h" -#include "whittaker.h" -#include "odum.h" -#include "canberra.h" -#include "structeuclidean.h" -#include "structchord.h" -#include "hellinger.h" -#include "manhattan.h" -#include "structpearson.h" -#include "soergel.h" -#include "spearman.h" -#include "structkulczynski.h" -#include "structchi2.h" -#include "speciesprofile.h" -#include "hamming.h" -#include "gower.h" -#include "memchi2.h" -#include "memchord.h" -#include "memeuclidean.h" -#include "mempearson.h" //********************************************************************************************************************** vector MatrixOutputCommand::setParameters(){ @@ -58,6 +17,7 @@ vector MatrixOutputCommand::setParameters(){ CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups); CommandParameter pcalc("calc", "Multiple", "sharedsobs-sharedchao-sharedace-jabund-sorabund-jclass-sorclass-jest-sorest-thetayc-thetan-kstest-sharednseqs-ochiai-anderberg-kulczynski-kulczynskicody-lennon-morisitahorn-braycurtis-whittaker-odum-canberra-structeuclidean-structchord-hellinger-manhattan-structpearson-soergel-spearman-structkulczynski-speciesprofile-hamming-structchi2-gower-memchi2-memchord-memeuclidean-mempearson", "jclass-thetayc", "", "", "",true,false); parameters.push_back(pcalc); CommandParameter poutput("output", "Multiple", "lt-square", "lt", "", "", "",false,false); parameters.push_back(poutput); + CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors); CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir); CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir); @@ -74,7 +34,8 @@ vector MatrixOutputCommand::setParameters(){ string MatrixOutputCommand::getHelpString(){ try { string helpString = ""; - helpString += "The dist.shared command parameters are shared, groups, calc, output and label. shared is a required, unless you have a valid current file.\n"; + ValidCalculators validCalculator; + helpString += "The dist.shared command parameters are shared, groups, calc, output, processors and label. shared is a required, unless you have a valid current file.\n"; helpString += "The groups parameter allows you to specify which of the groups in your groupfile you would like included used.\n"; helpString += "The group names are separated by dashes. The label parameter allows you to select what distance levels you would like distance matrices created for, and is also separated by dashes.\n"; helpString += "The dist.shared command should be in the following format: dist.shared(groups=yourGroups, calc=yourCalcs, label=yourLabels).\n"; @@ -82,9 +43,9 @@ string MatrixOutputCommand::getHelpString(){ helpString += "Example dist.shared(groups=A-B-C, calc=jabund-sorabund).\n"; helpString += "The default value for groups is all the groups in your groupfile.\n"; helpString += "The default value for calc is jclass and thetayc.\n"; - helpString += validCalculator->printCalc("matrix"); + helpString += validCalculator.printCalc("matrix"); helpString += "The dist.shared command outputs a .dist file for each calculator you specify at each distance you choose.\n"; - helpString += "Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n\n"; + helpString += "Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n"; return helpString; } catch(exception& e) { @@ -113,7 +74,8 @@ MatrixOutputCommand::MatrixOutputCommand(string option) { allLines = 1; //allow user to run help - if(option == "help") { validCalculator = new ValidCalculators(); help(); abort = true; calledHelp = true; } + if(option == "help") { help(); abort = true; calledHelp = true; } + else if(option == "citation") { citation(); abort = true; calledHelp = true;} else { vector myArray = setParameters(); @@ -154,6 +116,7 @@ MatrixOutputCommand::MatrixOutputCommand(string option) { if (sharedfile != "") { m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); } else { m->mothurOut("You have no current sharedfile and the shared parameter is required."); m->mothurOutEndLine(); abort = true; } }else if (sharedfile == "not open") { sharedfile = ""; abort = true; } + else { m->setSharedFile(sharedfile); } //if the user changes the output directory command factory will send this info to us in the output parameter outputDir = validParameter.validFile(parameters, "outputdir", false); if (outputDir == "not found"){ @@ -177,8 +140,12 @@ MatrixOutputCommand::MatrixOutputCommand(string option) { if (groups == "not found") { groups = ""; } else { m->splitAtDash(groups, Groups); - m->Groups = Groups; + m->setGroups(Groups); } + + string temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found"){ temp = m->getProcessors(); } + m->setProcessors(temp); + m->mothurConvert(temp, processors); calc = validParameter.validFile(parameters, "calc", false); if (calc == "not found") { calc = "jclass-thetayc"; } @@ -186,14 +153,19 @@ MatrixOutputCommand::MatrixOutputCommand(string option) { if (calc == "default") { calc = "jclass-thetayc"; } } m->splitAtDash(calc, Estimators); + if (m->inUsersGroups("citation", Estimators)) { + ValidCalculators validCalc; validCalc.printCitations(Estimators); + //remove citation from list of calcs + for (int i = 0; i < Estimators.size(); i++) { if (Estimators[i] == "citation") { Estimators.erase(Estimators.begin()+i); break; } } + } if (abort == false) { - validCalculator = new ValidCalculators(); + ValidCalculators validCalculator; int i; for (i=0; iisValidCalculator("matrix", Estimators[i]) == true) { + if (validCalculator.isValidCalculator("matrix", Estimators[i]) == true) { if (Estimators[i] == "sharedsobs") { matrixCalculators.push_back(new SharedSobsCS()); }else if (Estimators[i] == "sharedchao") { @@ -288,11 +260,7 @@ MatrixOutputCommand::MatrixOutputCommand(string option) { //********************************************************************************************************************** -MatrixOutputCommand::~MatrixOutputCommand(){ - if (abort == false) { - delete validCalculator; - } -} +MatrixOutputCommand::~MatrixOutputCommand(){} //********************************************************************************************************************** @@ -315,13 +283,18 @@ int MatrixOutputCommand::execute(){ if (lookup.size() < 2) { m->mothurOut("You have not provided enough valid groups. I cannot run the command."); m->mothurOutEndLine(); delete input; for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; } return 0;} numGroups = lookup.size(); + lines.resize(processors); + for (int i = 0; i < processors; i++) { + lines[i].start = int (sqrt(float(i)/float(processors)) * numGroups); + lines[i].end = int (sqrt(float(i+1)/float(processors)) * numGroups); + } - if (m->control_pressed) { delete input; for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; } m->Groups.clear(); return 0; } + if (m->control_pressed) { delete input; for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; } m->clearGroups(); return 0; } //as long as you are not at the end of the file or done wih the lines you want while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) { - if (m->control_pressed) { outputTypes.clear(); delete input; for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; } for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } m->Groups.clear(); return 0; } + if (m->control_pressed) { outputTypes.clear(); delete input; for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; } for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } m->clearGroups(); return 0; } if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){ m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine(); @@ -354,7 +327,7 @@ int MatrixOutputCommand::execute(){ lookup = input->getSharedRAbundVectors(); } - if (m->control_pressed) { outputTypes.clear(); delete input; for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } m->Groups.clear(); return 0; } + if (m->control_pressed) { outputTypes.clear(); delete input; for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } m->clearGroups(); return 0; } //output error messages about any remaining user labels set::iterator it; @@ -369,7 +342,7 @@ int MatrixOutputCommand::execute(){ } } - if (m->control_pressed) { outputTypes.clear(); delete input; for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } m->Groups.clear(); return 0; } + if (m->control_pressed) { outputTypes.clear(); delete input; for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } m->clearGroups(); return 0; } //run last label if you need to if (needToRun == true) { @@ -381,10 +354,10 @@ int MatrixOutputCommand::execute(){ for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; } } - if (m->control_pressed) { outputTypes.clear(); delete input; for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } m->Groups.clear(); return 0; } + if (m->control_pressed) { outputTypes.clear(); delete input; for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } m->clearGroups(); return 0; } //reset groups parameter - m->Groups.clear(); + m->clearGroups(); //set phylip file as new current phylipfile string current = ""; @@ -407,7 +380,7 @@ int MatrixOutputCommand::execute(){ } } /***********************************************************/ -void MatrixOutputCommand::printSims(ostream& out) { +void MatrixOutputCommand::printSims(ostream& out, vector< vector >& simMatrix) { try { out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint); @@ -441,63 +414,216 @@ void MatrixOutputCommand::printSims(ostream& out) { /***********************************************************/ int MatrixOutputCommand::process(vector thisLookup){ try { + EstOutput data; + vector subset; + vector< vector > calcDists; calcDists.resize(matrixCalculators.size()); //one for each calc, this will be used to make .dist files + - EstOutput data; - vector subset; - - //for each calculator - for(int i = 0 ; i < matrixCalculators.size(); i++) { + if(processors == 1){ + driver(thisLookup, 0, numGroups, calcDists); + }else{ + int process = 1; + vector processIDS; + + #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) + //loop through and create all the processes you want + while (process != processors) { + int pid = fork(); + + if (pid > 0) { + processIDS.push_back(pid); + process++; + }else if (pid == 0){ + driver(thisLookup, lines[process].start, lines[process].end, calcDists); - //initialize simMatrix - simMatrix.clear(); - simMatrix.resize(numGroups); - for (int p = 0; p < simMatrix.size(); p++) { - for (int j = 0; j < simMatrix.size(); j++) { - simMatrix[p].push_back(0.0); + string tempdistFileName = m->getRootName(m->getSimpleName(sharedfile)) + toString(getpid()) + ".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(thisLookup, 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 distSharedData 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 pDataArray; + DWORD dwThreadIdArray[processors-1]; + HANDLE hThreadArray[processors-1]; + + //Create processor worker threads. + for( int i=1; i newLookup; + for (int k = 0; k < thisLookup.size(); k++) { + SharedRAbundVector* temp = new SharedRAbundVector(); + temp->setLabel(thisLookup[k]->getLabel()); + temp->setGroup(thisLookup[k]->getGroup()); + newLookup.push_back(temp); + } + + //for each bin + for (int k = 0; k < thisLookup[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 < thisLookup.size(); j++) { newLookup[j]->push_back(thisLookup[j]->getAbundance(k), thisLookup[j]->getGroup()); } + } + + // Allocate memory for thread data. + distSharedData* tempSum = new distSharedData(m, lines[i].start, lines[i].end, Estimators, newLookup); + pDataArray.push_back(tempSum); + processIDS.push_back(i); + + hThreadArray[i-1] = CreateThread(NULL, 0, MyDistSharedThreadFunction, pDataArray[i-1], 0, &dwThreadIdArray[i-1]); + } + + //parent do your part + driver(thisLookup, 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++){ + 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 + } + + + + for (int i = 0; i < calcDists.size(); i++) { + if (m->control_pressed) { break; } + + //initialize matrix + vector< vector > 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; + float dist = calcDists[i][j].dist; + + matrix[row][column] = dist; + matrix[column][row] = dist; + } + + string distFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + matrixCalculators[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) { + m->errorOut(e, "MatrixOutputCommand", "process"); + exit(1); + } +} +/**************************************************************************************************/ +int MatrixOutputCommand::driver(vector thisLookup, int start, int end, vector< vector >& calcDists) { + try { + + vector subset; + for (int k = start; k < end; k++) { // pass cdd each set of groups to compare + + for (int l = 0; l < k; l++) { - for (int k = 0; k < thisLookup.size(); k++) { - for (int l = k; l < thisLookup.size(); l++) { - if (k != l) { //we dont need to similiarity of a groups to itself - //get estimated similarity between 2 groups - - if (m->control_pressed) { return 0; } - - subset.clear(); //clear out old pair of sharedrabunds - //add new pair of sharedrabunds - subset.push_back(thisLookup[k]); subset.push_back(thisLookup[l]); - - //if this calc needs all groups to calculate the pair load all groups - if (matrixCalculators[i]->getNeedsAll()) { - //load subset with rest of lookup for those calcs that need everyone to calc for a pair - for (int w = 0; w < thisLookup.size(); w++) { - if ((w != k) && (w != l)) { subset.push_back(thisLookup[w]); } - } - } - - data = matrixCalculators[i]->getValues(subset); //saves the calculator outputs - //save values in similarity matrix - simMatrix[k][l] = 1.0 - data[0]; //convert similiarity to distance - simMatrix[l][k] = 1.0 - data[0]; //convert similiarity to distance + if (k != l) { //we dont need to similiarity of a groups to itself + subset.clear(); //clear out old pair of sharedrabunds + //add new pair of sharedrabunds + subset.push_back(thisLookup[k]); subset.push_back(thisLookup[l]); + + for(int i=0;igetNeedsAll()) { + //load subset with rest of lookup for those calcs that need everyone to calc for a pair + for (int w = 0; w < thisLookup.size(); w++) { + if ((w != k) && (w != l)) { subset.push_back(thisLookup[w]); } } } + + vector tempdata = matrixCalculators[i]->getValues(subset); //saves the calculator outputs + + if (m->control_pressed) { return 1; } + + seqDist temp(l, k, tempdata[0]); + calcDists[i].push_back(temp); } - - exportFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + matrixCalculators[i]->getName() + "." + thisLookup[0]->getLabel() + "." + output + ".dist"; - m->openOutputFile(exportFileName, out); - outputNames.push_back(exportFileName); outputTypes["phylip"].push_back(exportFileName); - - printSims(out); - out.close(); - } - - return 0; + } + } + return 0; } catch(exception& e) { - m->errorOut(e, "MatrixOutputCommand", "process"); + m->errorOut(e, "MatrixOutputCommand", "driver"); exit(1); } }