From: Sarah Westcott Date: Mon, 11 Jun 2012 16:13:55 +0000 (-0400) Subject: fixed bug with dist.shared subsampling. added mode parameter to dist.shared so... X-Git-Url: https://git.donarmstrong.com/?p=mothur.git;a=commitdiff_plain;h=2c97dd48b8e27ee0a6a86c7a082f4c504c3357c6 fixed bug with dist.shared subsampling. added mode parameter to dist.shared so you can select average or median for the results. added some debug code as I sorted out some bug reports to mothur.bugs. --- diff --git a/chimerauchimecommand.cpp b/chimerauchimecommand.cpp index be0421a..b1d996b 100644 --- a/chimerauchimecommand.cpp +++ b/chimerauchimecommand.cpp @@ -994,7 +994,8 @@ int ChimeraUchimeCommand::driverGroups(SequenceParser& parser, string outputFNam if (m->control_pressed) { return 0; } //remove file made for uchime - m->mothurRemove(filename); + if (!m->debug) { m->mothurRemove(filename); } + else { m->mothurOut("[DEBUG]: saving file: " + filename + ".\n"); } //append files m->appendFiles((outputFName+groups[i]), outputFName); m->mothurRemove((outputFName+groups[i])); diff --git a/matrixoutputcommand.cpp b/matrixoutputcommand.cpp index 05cd18a..e1c8a98 100644 --- a/matrixoutputcommand.cpp +++ b/matrixoutputcommand.cpp @@ -19,6 +19,7 @@ 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 pmode("mode", "Multiple", "average-median", "average", "", "", "",false,false); parameters.push_back(pmode); CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors); CommandParameter piters("iters", "Number", "", "1000", "", "", "",false,false); parameters.push_back(piters); CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir); @@ -38,13 +39,14 @@ string MatrixOutputCommand::getHelpString(){ try { string helpString = ""; ValidCalculators validCalculator; - helpString += "The dist.shared command parameters are shared, groups, calc, output, processors, subsample, iters and label. shared is a required, unless you have a valid current file.\n"; + helpString += "The dist.shared command parameters are shared, groups, calc, output, processors, subsample, iters, mode, 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 iters parameter allows you to choose the number of times you would like to run the subsample.\n"; helpString += "The subsample parameter allows you to enter the size pergroup of the sample or you can set subsample=T and mothur will use the size of your smallest group.\n"; helpString += "The dist.shared command should be in the following format: dist.shared(groups=yourGroups, calc=yourCalcs, label=yourLabels).\n"; helpString += "The output parameter allows you to specify format of your distance matrix. Options are lt, and square. The default is lt.\n"; + helpString += "The mode parameter allows you to specify if you want the average or the median values reported when subsampling. Options are average, and median. The default is average.\n"; 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"; @@ -140,6 +142,9 @@ MatrixOutputCommand::MatrixOutputCommand(string option) { output = validParameter.validFile(parameters, "output", false); if(output == "not found"){ output = "lt"; } if ((output != "lt") && (output != "square")) { m->mothurOut(output + " is not a valid output form. Options are lt and square. I will use lt."); m->mothurOutEndLine(); output = "lt"; } + + mode = validParameter.validFile(parameters, "mode", false); if(mode == "not found"){ mode = "average"; } + if ((mode != "average") && (mode != "median")) { m->mothurOut(mode + " is not a valid mode. Options are average and medina. I will use average."); m->mothurOutEndLine(); output = "average"; } groups = validParameter.validFile(parameters, "groups", false); if (groups == "not found") { groups = ""; } @@ -620,11 +625,16 @@ int MatrixOutputCommand::process(vector thisLookup){ } if (subsample && (thisIter != 0)) { + if((thisIter) % 100 == 0){ m->mothurOut(toString(thisIter)); m->mothurOutEndLine(); } calcDistsTotals.push_back(calcDists); + for (int i = 0; i < calcDists.size(); i++) { + for (int j = 0; j < calcDists[i].size(); j++) { + if (m->debug) { m->mothurOut("[DEBUG]: Results: iter = " + toString(thisIter) + ", " + thisLookup[calcDists[i][j].seq1]->getGroup() + " - " + thisLookup[calcDists[i][j].seq2]->getGroup() + " distance = " + toString(calcDists[i][j].dist) + ".\n"); } + } + } //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 { //print results for whole dataset for (int i = 0; i < calcDists.size(); i++) { if (m->control_pressed) { break; } @@ -654,6 +664,7 @@ int MatrixOutputCommand::process(vector thisLookup){ outDist.close(); } } + for (int i = 0; i < calcDists.size(); i++) { calcDists[i].clear(); } } if (iters != 1) { @@ -664,34 +675,46 @@ int MatrixOutputCommand::process(vector thisLookup){ calcAverages[i].resize(calcDistsTotals[0][i].size()); for (int j = 0; j < calcAverages[i].size(); j++) { - calcAverages[i][j].seq1 = calcDists[i][j].seq1; - calcAverages[i][j].seq2 = calcDists[i][j].seq2; + calcAverages[i][j].seq1 = calcDistsTotals[0][i][j].seq1; + calcAverages[i][j].seq2 = calcDistsTotals[0][i][j].seq2; calcAverages[i][j].dist = 0.0; } } - - for (int thisIter = 0; thisIter < iters; thisIter++) { //sum all groups dists for each calculator - for (int i = 0; i < calcAverages.size(); i++) { //initialize sums to zero. + if (mode == "average") { + for (int thisIter = 0; thisIter < iters; thisIter++) { //sum all groups dists for each calculator + for (int i = 0; i < calcAverages.size(); i++) { //initialize sums to zero. + for (int j = 0; j < calcAverages[i].size(); j++) { + calcAverages[i][j].dist += calcDistsTotals[thisIter][i][j].dist; + if (m->debug) { m->mothurOut("[DEBUG]: Totaling for average calc: iter = " + toString(thisIter) + ", " + thisLookup[calcDistsTotals[thisIter][i][j].seq1]->getGroup() + " - " + thisLookup[calcDistsTotals[thisIter][i][j].seq2]->getGroup() + " distance = " + toString(calcDistsTotals[thisIter][i][j].dist) + ". New total = " + toString(calcAverages[i][j].dist) + ".\n"); } + } + } + } + + for (int i = 0; i < calcAverages.size(); i++) { //finds average. for (int j = 0; j < calcAverages[i].size(); j++) { - calcAverages[i][j].dist += calcDistsTotals[thisIter][i][j].dist; + calcAverages[i][j].dist /= (float) iters; } } - } - - for (int i = 0; i < calcAverages.size(); i++) { //finds average. - for (int j = 0; j < calcAverages[i].size(); j++) { - calcAverages[i][j].dist /= (float) iters; + }else { //find median + for (int i = 0; i < calcAverages.size(); i++) { //for each calc + for (int j = 0; j < calcAverages[i].size(); j++) { //for each comparison + vector dists; + for (int thisIter = 0; thisIter < iters; thisIter++) { //for each subsample + dists.push_back(calcDistsTotals[thisIter][i][j].dist); + } + sort(dists.begin(), dists.end()); + calcAverages[i][j].dist = dists[(iters/2)]; + } } } - //find standard deviation vector< vector > stdDev; stdDev.resize(matrixCalculators.size()); for (int i = 0; i < stdDev.size(); i++) { //initialize sums to zero. stdDev[i].resize(calcDistsTotals[0][i].size()); for (int j = 0; j < stdDev[i].size(); j++) { - stdDev[i][j].seq1 = calcDists[i][j].seq1; - stdDev[i][j].seq2 = calcDists[i][j].seq2; + stdDev[i][j].seq1 = calcDistsTotals[0][i][j].seq1; + stdDev[i][j].seq2 = calcDistsTotals[0][i][j].seq2; stdDev[i][j].dist = 0.0; } } @@ -768,6 +791,7 @@ int MatrixOutputCommand::process(vector thisLookup){ 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++) { diff --git a/matrixoutputcommand.h b/matrixoutputcommand.h index 8af539b..594fe30 100644 --- a/matrixoutputcommand.h +++ b/matrixoutputcommand.h @@ -101,7 +101,7 @@ private: bool abort, allLines, subsample; set labels; //holds labels to be used - string outputFile, calc, groups, label, outputDir; + string outputFile, calc, groups, label, outputDir, mode; vector Estimators, Groups, outputNames; //holds estimators to be used int process(vector, string, string); int driver(vector, int, int, vector< vector >&); diff --git a/sequenceparser.cpp b/sequenceparser.cpp index fd94b24..3eb508d 100644 --- a/sequenceparser.cpp +++ b/sequenceparser.cpp @@ -37,13 +37,16 @@ SequenceParser::SequenceParser(string groupFile, string fastaFile, string nameFi m->openInputFile(fastaFile, in); map seqName; //stores name -> sequence string so we can make new "unique" sequences when we parse the name file + int fastaCount = 0; while (!in.eof()) { if (m->control_pressed) { break; } Sequence seq(in); m->gobble(in); + fastaCount++; + if (m->debug) { if((fastaCount) % 1000 == 0){ m->mothurOut("[DEBUG]: reading seq " + toString(fastaCount) + "\n."); } } - if (seq.getName() != "") { + if (seq.getName() != "") { string group = groupMap->getGroup(seq.getName()); if (group == "not found") { error = 1; m->mothurOut("[ERROR]: " + seq.getName() + " is in your fasta file and not in your groupfile, please correct."); m->mothurOutEndLine(); } @@ -133,7 +136,7 @@ SequenceParser::SequenceParser(string groupFile, string fastaFile, string nameFi inName.close(); if (error == 1) { m->control_pressed = true; } - + if (countName != (groupMap->getNumSeqs())) { vector groupseqsnames = groupMap->getNamesSeqs(); @@ -253,6 +256,7 @@ vector SequenceParser::getSeqs(string g){ m->mothurOut("[ERROR]: No sequences available for group " + g + ", please correct."); m->mothurOutEndLine(); }else { seqForThisGroup = it->second; + if (m->debug) { m->mothurOut("[DEBUG]: group " + g + " fasta file has " + toString(seqForThisGroup.size()) + " sequences."); } } return seqForThisGroup; @@ -346,6 +350,7 @@ map SequenceParser::getNameMap(string g){ m->mothurOut("[ERROR]: No nameMap available for group " + g + ", please correct."); m->mothurOutEndLine(); }else { nameMapForThisGroup = it->second; + if (m->debug) { m->mothurOut("[DEBUG]: group " + g + " name file has " + toString(nameMapForThisGroup.size()) + " unique sequences."); } } return nameMapForThisGroup; diff --git a/shhhercommand.cpp b/shhhercommand.cpp index 08bb017..8ae76d9 100644 --- a/shhhercommand.cpp +++ b/shhhercommand.cpp @@ -938,6 +938,8 @@ void ShhherCommand::initPyroCluster(){ try{ if (numOTUs < processors) { processors = 1; } + if (m->debug) { m->mothurOut("[DEBUG]: numSeqs = " + toString(numSeqs) + " numOTUS = " + toString(numOTUs) + " about to alloc a dist vector with size = " + toString((numSeqs * numOTUs)) + ".\n"); } + dist.assign(numSeqs * numOTUs, 0); change.assign(numOTUs, 1); centroids.assign(numOTUs, -1); @@ -947,6 +949,8 @@ void ShhherCommand::initPyroCluster(){ nSeqsBreaks.assign(processors+1, 0); nOTUsBreaks.assign(processors+1, 0); + if (m->debug) { m->mothurOut("[DEBUG]: made it through the memory allocation.\n"); } + nSeqsBreaks[0] = 0; for(int i=0;i filenames, string thisCompositeFASTAFil vector uniqueLengths; int numFlowCells; + if (m->debug) { m->mothurOut("[DEBUG]: About to read flowgrams.\n"); } int numSeqs = getFlowData(flowFileName, seqNameVector, lengths, flowDataIntI, nameMap, numFlowCells); if (m->control_pressed) { break; } @@ -2193,6 +2198,8 @@ int ShhherCommand::driver(vector filenames, string thisCompositeFASTAFil vector nSeqsBreaks; vector nOTUsBreaks; + if (m->debug) { m->mothurOut("[DEBUG]: numSeqs = " + toString(numSeqs) + " numOTUS = " + toString(numOTUs) + " about to alloc a dist vector with size = " + toString((numSeqs * numOTUs)) + ".\n"); } + dist.assign(numSeqs * numOTUs, 0); change.assign(numOTUs, 1); centroids.assign(numOTUs, -1); @@ -2206,6 +2213,8 @@ int ShhherCommand::driver(vector filenames, string thisCompositeFASTAFil nSeqsBreaks[1] = numSeqs; nOTUsBreaks[1] = numOTUs; + if (m->debug) { m->mothurOut("[DEBUG]: done allocating memory, about to denoise.\n"); } + if (m->control_pressed) { break; } double maxDelta = 0; @@ -2336,17 +2345,21 @@ int ShhherCommand::getFlowData(string filename, vector& thisSeqNameVecto thisNameMap.clear(); flowFile >> numFlowCells; + if (m->debug) { m->mothurOut("[DEBUG]: numFlowCells = " + toString(numFlowCells) + ".\n"); } int index = 0;//pcluster while(!flowFile.eof()){ if (m->control_pressed) { break; } flowFile >> seqName >> currentNumFlowCells; + thisLengths.push_back(currentNumFlowCells); thisSeqNameVector.push_back(seqName); thisNameMap[seqName] = index++;//pcluster - + + if (m->debug) { m->mothurOut("[DEBUG]: seqName = " + seqName + " length = " + toString(currentNumFlowCells) + " index = " + toString(index) + "\n"); } + for(int i=0;i> intensity; if(intensity > 9.99) { intensity = 9.99; } @@ -2634,6 +2647,8 @@ int ShhherCommand::getOTUData(int numSeqs, string fileName, vector& otuDat listFile >> label >> numOTUs; + if (m->debug) { m->mothurOut("[DEBUG]: Getting OTU Data...\n"); } + otuData.assign(numSeqs, 0); cumNumSeqs.assign(numOTUs, 0); nSeqsPerOTU.assign(numOTUs, 0); @@ -2648,6 +2663,7 @@ int ShhherCommand::getOTUData(int numSeqs, string fileName, vector& otuDat for(int i=0;icontrol_pressed) { break; } + if (m->debug) { m->mothurOut("[DEBUG]: processing OTU " + toString(i) + ".\n"); } listFile >> singleOTU; diff --git a/subsample.cpp b/subsample.cpp index f7da25f..261297d 100644 --- a/subsample.cpp +++ b/subsample.cpp @@ -250,7 +250,7 @@ vector SubSample::getSample(vector& thislookup, int //subsampling may have created some otus with no sequences in them eliminateZeroOTUS(thislookup); - + if (m->control_pressed) { return m->currentBinLabels; } //save mothurOut's binLabels to restore for next label diff --git a/summarysharedcommand.cpp b/summarysharedcommand.cpp index ed02804..6792f19 100644 --- a/summarysharedcommand.cpp +++ b/summarysharedcommand.cpp @@ -744,7 +744,6 @@ int SummarySharedCommand::process(vector thisLookup, string //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++) { @@ -776,6 +775,7 @@ int SummarySharedCommand::process(vector thisLookup, string } } } + for (int i = 0; i < calcDists.size(); i++) { calcDists[i].clear(); } } if (iters != 1) {