X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=pcacommand.cpp;h=27e448bdfa1060a91c701c25fe301907e3529ec7;hb=01f6af90c907264686304a5c684c702e94ff40ae;hp=a91143bbd87c28deedc6edc03af4ffbea95f37de;hpb=ca9ac1d80c62f57270b0dcd49410ebe08a8aecd6;p=mothur.git diff --git a/pcacommand.cpp b/pcacommand.cpp index a91143b..27e448b 100644 --- a/pcacommand.cpp +++ b/pcacommand.cpp @@ -37,10 +37,10 @@ string PCACommand::getHelpString(){ helpString += "The pca command parameters are shared, relabund, label, groups and metric. shared or relabund is required unless you have a valid current file."; helpString += "The label parameter is used to analyze specific labels in your input. Default is the first label in your shared or relabund file. Multiple labels may be separated by dashes.\n"; helpString += "The groups parameter allows you to specify which groups you would like analyzed. Groupnames are separated by dashes.\n"; - helpString += "The metric parameter allows indicate you if would like the pearson correlation coefficient calculated. Default=True"; + helpString += "The metric parameter allows you to indicate if would like the pearson correlation coefficient calculated. Default=True"; helpString += "Example pca(groups=yourGroups).\n"; helpString += "Example pca(groups=A-B-C).\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) { @@ -70,6 +70,7 @@ PCACommand::PCACommand(string option) { //allow user to run help if(option == "help") { help(); abort = true; calledHelp = true; } + else if(option == "citation") { citation(); abort = true; calledHelp = true;} else { vector myArray = setParameters(); @@ -116,12 +117,12 @@ PCACommand::PCACommand(string option) { sharedfile = validParameter.validFile(parameters, "shared", true); if (sharedfile == "not open") { sharedfile = ""; abort = true; } else if (sharedfile == "not found") { sharedfile = ""; } - else { mode = "sharedfile"; inputFile = sharedfile; } + else { mode = "sharedfile"; inputFile = sharedfile; m->setSharedFile(sharedfile); } relabundfile = validParameter.validFile(parameters, "relabund", true); if (relabundfile == "not open") { relabundfile = ""; abort = true; } else if (relabundfile == "not found") { relabundfile = ""; } - else { mode = "relabund"; inputFile = relabundfile; } + else { mode = "relabund"; inputFile = relabundfile; m->setRelAbundFile(relabundfile); } if ((sharedfile == "") && (relabundfile == "")) { @@ -156,7 +157,7 @@ PCACommand::PCACommand(string option) { groups = validParameter.validFile(parameters, "groups", false); if (groups == "not found") { groups = ""; } else { m->splitAtDash(groups, Groups); } - m->Groups = Groups; + m->setGroups(Groups); } @@ -180,7 +181,7 @@ int PCACommand::execute(){ //get first line of shared file vector< vector > matrix; InputData* input; - if (mode == "shared") { + if (mode == "sharedfile") { input = new InputData(inputFile, "sharedfile"); }else if (mode == "relabund") { input = new InputData(inputFile, "relabund"); @@ -202,7 +203,7 @@ int PCACommand::execute(){ //as long as you are not at the end of the file or done wih the lines you want while((lookupFloat[0] != NULL) && (userLabels.size() != 0)) { - if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } delete input; for (int i = 0; i < lookupFloat.size(); i++) { delete lookupFloat[i]; } lookupFloat.clear(); return 0; } + if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } delete input; for (int i = 0; i < lookupFloat.size(); i++) { delete lookupFloat[i]; } lookupFloat.clear(); return 0; } if(labels.count(lookupFloat[0]->getLabel()) == 1){ processedLabels.insert(lookupFloat[0]->getLabel()); @@ -235,7 +236,7 @@ int PCACommand::execute(){ } - if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } delete input; for (int i = 0; i < lookupFloat.size(); i++) { delete lookupFloat[i]; } lookupFloat.clear(); return 0; } + if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } delete input; for (int i = 0; i < lookupFloat.size(); i++) { delete lookupFloat[i]; } lookupFloat.clear(); return 0; } //output error messages about any remaining user labels set::iterator it; @@ -263,7 +264,7 @@ int PCACommand::execute(){ for (int i = 0; i < lookupFloat.size(); i++) { if (lookupFloat[i] != NULL) { delete lookupFloat[i]; } } lookupFloat.clear(); delete input; - if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; } + if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } return 0; } m->mothurOutEndLine(); m->mothurOut("Output File Names: "); m->mothurOutEndLine(); @@ -277,6 +278,7 @@ int PCACommand::execute(){ exit(1); } } + /********************************************************************************************************************** vector< vector > PCACommand::createMatrix(vector lookupFloat){ try { @@ -306,82 +308,80 @@ vector< vector > PCACommand::createMatrix(vector& lookupFloat){ try { m->mothurOut("\nProcessing " + lookupFloat[0]->getLabel()); m->mothurOutEndLine(); + + int numOTUs = lookupFloat[0]->getNumBins(); + int numSamples = lookupFloat.size(); - vector< vector > matrix; matrix.resize(lookupFloat.size()); - - ofstream out; - string temp = outputDir + "matrix.transpose.out"; - m->openOutputFile(temp, out); - out << "matrix" << endl; + vector< vector > matrix(numSamples); + vector colMeans(numOTUs); - //fill matrix with shared files relative abundances + //fill matrix with shared relative abundances, re-center for (int i = 0; i < lookupFloat.size(); i++) { - for (int j = 0; j < lookupFloat[i]->getNumBins(); j++) { - matrix[i].push_back(lookupFloat[i]->getAbundance(j)); - out << lookupFloat[i]->getAbundance(j) << '\t'; - } - out << endl; - } - out << endl << endl << "transpose" << endl; - vector< vector > transposeMatrix; transposeMatrix.resize(matrix[0].size()); - for (int i = 0; i < transposeMatrix.size(); i++) { - for (int j = 0; j < matrix.size(); j++) { - transposeMatrix[i].push_back(matrix[j][i]); - out << matrix[j][i] << '\t'; + matrix[i].resize(numOTUs, 0); + + for (int j = 0; j < numOTUs; j++) { + matrix[i][j] = lookupFloat[i]->getAbundance(j); + colMeans[j] += matrix[i][j]; } - out << endl; } - matrix = linearCalc.matrix_mult(matrix, transposeMatrix); + + for(int j=0;j > centered = matrix; + for(int i=0;i > transpose(numOTUs); + for (int i = 0; i < numOTUs; i++) { + transpose[i].resize(numSamples, 0); + for (int j = 0; j < numSamples; j++) { + transpose[i][j] = centered[j][i]; + } + } + + vector > crossProduct = linearCalc.matrix_mult(transpose, centered); - double offset = 0.0000; vector d; vector e; - vector > G = matrix; - //vector > copy_G; - - for(int count=0;count<2;count++){ - linearCalc.recenter(offset, matrix, G); if (m->control_pressed) { return 0; } - linearCalc.tred2(G, d, e); if (m->control_pressed) { return 0; } - linearCalc.qtli(d, e, G); if (m->control_pressed) { return 0; } - offset = d[d.size()-1]; - if(offset > 0.0) break; - } + + linearCalc.tred2(crossProduct, d, e); if (m->control_pressed) { return 0; } + linearCalc.qtli(d, e, crossProduct); if (m->control_pressed) { return 0; } + + vector > X = linearCalc.matrix_mult(centered, crossProduct); if (m->control_pressed) { return 0; } string fbase = outputDir + m->getRootName(m->getSimpleName(inputFile)); string outputFileName = fbase + lookupFloat[0]->getLabel(); - output(outputFileName, m->Groups, G, d); + output(outputFileName, m->getGroups(), X, d); if (metric) { + vector > observedEuclideanDistance = linearCalc.getObservedEuclideanDistance(centered); + for (int i = 1; i < 4; i++) { - vector< vector > EuclidDists = linearCalc.calculateEuclidianDistance(G, i); //G is the pca file - - if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; } - - double corr = linearCalc.calcPearson(EuclidDists, matrix); //G is the pca file, D is the users distance matrix - - m->mothurOut("Pearson's coefficient using " + toString(i) + " axis: " + toString(corr)); m->mothurOutEndLine(); + vector< vector > PCAEuclidDists = linearCalc.calculateEuclidianDistance(X, i); //G is the pca file + if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } return 0; } + + double corr = linearCalc.calcPearson(PCAEuclidDists, observedEuclideanDistance); + m->mothurOut("Rsq " + toString(i) + " axis: " + toString(corr * corr)); m->mothurOutEndLine(); - if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; } + if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } return 0; } } } @@ -396,14 +396,11 @@ int PCACommand::process(vector& lookupFloat){ void PCACommand::output(string fnameRoot, vector name_list, vector >& G, vector d) { try { - int rank = name_list.size(); + + int numEigenValues = d.size(); double dsum = 0.0000; - for(int i=0;i= 0) { G[i][j] *= pow(d[j],0.5); } - else { G[i][j] = 0.00000; } - } } ofstream pcaData((fnameRoot+".pca.axes").c_str(), ios::trunc); @@ -419,19 +416,19 @@ void PCACommand::output(string fnameRoot, vector name_list, vector