#include "sharedthetayc.h"
#include "sharedthetan.h"
#include "sharedkstest.h"
-#include "sharedbdiversity.h"
+#include "whittaker.h"
#include "sharedochiai.h"
#include "sharedanderbergs.h"
#include "sharedkulczynski.h"
#include "sharedlennon.h"
#include "sharedmorisitahorn.h"
#include "sharedbraycurtis.h"
+#include "sharedjackknife.h"
+#include "whittaker.h"
+#include "odum.h"
+#include "canberra.h"
+#include "stricteuclidean.h"
+#include "strictchord.h"
+#include "hellinger.h"
+#include "manhattan.h"
+#include "strictpearson.h"
+#include "soergel.h"
+#include "spearman.h"
+#include "strictkulczynski.h"
+#include "speciesprofile.h"
+#include "hamming.h"
-
+//**********************************************************************************************************************
+vector<string> SummarySharedCommand::getValidParameters(){
+ try {
+ string Array[] = {"label","calc","groups","all","outputdir","distance","inputdir", "processors"};
+ vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
+ return myArray;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SummarySharedCommand", "getValidParameters");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+SummarySharedCommand::SummarySharedCommand(){
+ try {
+ abort = true;
+ //initialize outputTypes
+ vector<string> tempOutNames;
+ outputTypes["summary"] = tempOutNames;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SummarySharedCommand", "SummarySharedCommand");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+vector<string> SummarySharedCommand::getRequiredParameters(){
+ try {
+ vector<string> myArray;
+ return myArray;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SummarySharedCommand", "getRequiredParameters");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+vector<string> SummarySharedCommand::getRequiredFiles(){
+ try {
+ string Array[] = {"shared"};
+ vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
+ return myArray;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SummarySharedCommand", "getRequiredFiles");
+ exit(1);
+ }
+}
//**********************************************************************************************************************
-SummarySharedCommand::SummarySharedCommand(){
+SummarySharedCommand::SummarySharedCommand(string option) {
try {
globaldata = GlobalData::getInstance();
- outputFileName = ((getRootName(globaldata->inputFileName)) + "shared.summary");
- openOutputFile(outputFileName, outputFileHandle);
- format = globaldata->getFormat();
- validCalculator = new ValidCalculators();
- util = new SharedUtil();
-
- int i;
- for (i=0; i<globaldata->Estimators.size(); i++) {
- if (validCalculator->isValidCalculator("sharedsummary", globaldata->Estimators[i]) == true) {
- if (globaldata->Estimators[i] == "sharedsobs") {
- sumCalculators.push_back(new SharedSobsCS());
- }else if (globaldata->Estimators[i] == "sharedchao") {
- sumCalculators.push_back(new SharedChao1());
- }else if (globaldata->Estimators[i] == "sharedace") {
- sumCalculators.push_back(new SharedAce());
- }else if (globaldata->Estimators[i] == "jabund") {
- sumCalculators.push_back(new JAbund());
- }else if (globaldata->Estimators[i] == "sorensonabund") {
- sumCalculators.push_back(new SorAbund());
- }else if (globaldata->Estimators[i] == "jclass") {
- sumCalculators.push_back(new Jclass());
- }else if (globaldata->Estimators[i] == "sorclass") {
- sumCalculators.push_back(new SorClass());
- }else if (globaldata->Estimators[i] == "jest") {
- sumCalculators.push_back(new Jest());
- }else if (globaldata->Estimators[i] == "sorest") {
- sumCalculators.push_back(new SorEst());
- }else if (globaldata->Estimators[i] == "thetayc") {
- sumCalculators.push_back(new ThetaYC());
- }else if (globaldata->Estimators[i] == "thetan") {
- sumCalculators.push_back(new ThetaN());
- }else if (globaldata->Estimators[i] == "kstest") {
- sumCalculators.push_back(new KSTest());
- }else if (globaldata->Estimators[i] == "sharednseqs") {
- sumCalculators.push_back(new SharedNSeqs());
- }else if (globaldata->Estimators[i] == "ochiai") {
- sumCalculators.push_back(new Ochiai());
- }else if (globaldata->Estimators[i] == "anderberg") {
- sumCalculators.push_back(new Anderberg());
- }else if (globaldata->Estimators[i] == "kulczynski") {
- sumCalculators.push_back(new Kulczynski());
- }else if (globaldata->Estimators[i] == "kulczynskicody") {
- sumCalculators.push_back(new KulczynskiCody());
- }else if (globaldata->Estimators[i] == "lennon") {
- sumCalculators.push_back(new Lennon());
- }else if (globaldata->Estimators[i] == "morisitahorn") {
- sumCalculators.push_back(new MorHorn());
- }else if (globaldata->Estimators[i] == "braycurtis") {
- sumCalculators.push_back(new BrayCurtis());
- }
- else if (globaldata->Estimators[i] == "bdiversity") {
- sumCalculators.push_back(new BDiversity());
+ abort = false;
+ allLines = 1;
+ labels.clear();
+ Estimators.clear();
+
+ //allow user to run help
+ if(option == "help") { validCalculator = new ValidCalculators(); help(); abort = true; }
+
+ else {
+ //valid paramters for this command
+ string Array[] = {"label","calc","groups","all","outputdir","distance","inputdir", "processors"};
+ vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
+
+ OptionParser parser(option);
+ map<string, string> parameters = parser.getParameters();
+
+ ValidParameters validParameter;
+
+ //check to make sure all parameters are valid for command
+ for (map<string, string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
+ if (validParameter.isValidParameter(it->first, myArray, it->second) != true) { abort = true; }
+ }
+
+ //make sure the user has already run the read.otu command
+ if (globaldata->getSharedFile() == "") {
+ m->mothurOut("You must read a list and a group, or a shared before you can use the summary.shared command."); m->mothurOutEndLine(); abort = true;
+ }
+
+ //initialize outputTypes
+ vector<string> tempOutNames;
+ outputTypes["summary"] = tempOutNames;
+
+ //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"){
+ outputDir = "";
+ outputDir += m->hasPath(globaldata->getSharedFile()); //if user entered a file with a path then preserve it
+ }
+
+ //check for optional parameter and set defaults
+ // ...at some point should added some additional type checking...
+ label = validParameter.validFile(parameters, "label", false);
+ if (label == "not found") { label = ""; }
+ else {
+ if(label != "all") { m->splitAtDash(label, labels); allLines = 0; }
+ else { allLines = 1; }
+ }
+
+ //if the user has not specified any labels use the ones from read.otu
+ if(label == "") {
+ allLines = globaldata->allLines;
+ labels = globaldata->labels;
+ }
+
+ calc = validParameter.validFile(parameters, "calc", false);
+ if (calc == "not found") { calc = "sharedsobs-sharedchao-sharedace-jabund-sorabund-jclass-sorclass-jest-sorest-thetayc-thetan"; }
+ else {
+ if (calc == "default") { calc = "sharedsobs-sharedchao-sharedace-jabund-sorabund-jclass-sorclass-jest-sorest-thetayc-thetan"; }
+ }
+ m->splitAtDash(calc, Estimators);
+
+ groups = validParameter.validFile(parameters, "groups", false);
+ if (groups == "not found") { groups = ""; }
+ else {
+ m->splitAtDash(groups, Groups);
+ globaldata->Groups = Groups;
+ }
+
+ string temp = validParameter.validFile(parameters, "all", false); if (temp == "not found") { temp = "false"; }
+ all = m->isTrue(temp);
+
+ temp = validParameter.validFile(parameters, "distance", false); if (temp == "not found") { temp = "false"; }
+ createPhylip = m->isTrue(temp);
+
+ temp = validParameter.validFile(parameters, "processors", false); if(temp == "not found"){ temp = "1"; }
+ convert(temp, processors);
+
+ if (abort == false) {
+
+ validCalculator = new ValidCalculators();
+ int i;
+
+ for (i=0; i<Estimators.size(); i++) {
+ if (validCalculator->isValidCalculator("sharedsummary", Estimators[i]) == true) {
+ if (Estimators[i] == "sharedsobs") {
+ sumCalculators.push_back(new SharedSobsCS());
+ }else if (Estimators[i] == "sharedchao") {
+ sumCalculators.push_back(new SharedChao1());
+ }else if (Estimators[i] == "sharedace") {
+ sumCalculators.push_back(new SharedAce());
+ }else if (Estimators[i] == "jabund") {
+ sumCalculators.push_back(new JAbund());
+ }else if (Estimators[i] == "sorabund") {
+ sumCalculators.push_back(new SorAbund());
+ }else if (Estimators[i] == "jclass") {
+ sumCalculators.push_back(new Jclass());
+ }else if (Estimators[i] == "sorclass") {
+ sumCalculators.push_back(new SorClass());
+ }else if (Estimators[i] == "jest") {
+ sumCalculators.push_back(new Jest());
+ }else if (Estimators[i] == "sorest") {
+ sumCalculators.push_back(new SorEst());
+ }else if (Estimators[i] == "thetayc") {
+ sumCalculators.push_back(new ThetaYC());
+ }else if (Estimators[i] == "thetan") {
+ sumCalculators.push_back(new ThetaN());
+ }else if (Estimators[i] == "kstest") {
+ sumCalculators.push_back(new KSTest());
+ }else if (Estimators[i] == "sharednseqs") {
+ sumCalculators.push_back(new SharedNSeqs());
+ }else if (Estimators[i] == "ochiai") {
+ sumCalculators.push_back(new Ochiai());
+ }else if (Estimators[i] == "anderberg") {
+ sumCalculators.push_back(new Anderberg());
+ }else if (Estimators[i] == "kulczynski") {
+ sumCalculators.push_back(new Kulczynski());
+ }else if (Estimators[i] == "kulczynskicody") {
+ sumCalculators.push_back(new KulczynskiCody());
+ }else if (Estimators[i] == "lennon") {
+ sumCalculators.push_back(new Lennon());
+ }else if (Estimators[i] == "morisitahorn") {
+ sumCalculators.push_back(new MorHorn());
+ }else if (Estimators[i] == "braycurtis") {
+ sumCalculators.push_back(new BrayCurtis());
+ }else if (Estimators[i] == "whittaker") {
+ sumCalculators.push_back(new Whittaker());
+ }else if (Estimators[i] == "odum") {
+ sumCalculators.push_back(new Odum());
+ }else if (Estimators[i] == "canberra") {
+ sumCalculators.push_back(new Canberra());
+ }else if (Estimators[i] == "stricteuclidean") {
+ sumCalculators.push_back(new StrictEuclidean());
+ }else if (Estimators[i] == "strictchord") {
+ sumCalculators.push_back(new StrictChord());
+ }else if (Estimators[i] == "hellinger") {
+ sumCalculators.push_back(new Hellinger());
+ }else if (Estimators[i] == "manhattan") {
+ sumCalculators.push_back(new Manhattan());
+ }else if (Estimators[i] == "strictpearson") {
+ sumCalculators.push_back(new StrictPearson());
+ }else if (Estimators[i] == "soergel") {
+ sumCalculators.push_back(new Soergel());
+ }else if (Estimators[i] == "spearman") {
+ sumCalculators.push_back(new Spearman());
+ }else if (Estimators[i] == "strictkulczynski") {
+ sumCalculators.push_back(new StrictKulczynski());
+ }else if (Estimators[i] == "speciesprofile") {
+ sumCalculators.push_back(new SpeciesProfile());
+ }else if (Estimators[i] == "hamming") {
+ sumCalculators.push_back(new Hamming());
+ }
+ }
}
+ mult = false;
}
}
- //reset calc for next command
- globaldata->setCalc("");
-
}
catch(exception& e) {
- cout << "Standard Error: " << e.what() << " has occurred in the SummarySharedCommand class Function SummarySharedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+ m->errorOut(e, "SummarySharedCommand", "SummarySharedCommand");
exit(1);
}
- catch(...) {
- cout << "An unknown error has occurred in the SummarySharedCommand class function SummarySharedCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+}
+
+//**********************************************************************************************************************
+
+void SummarySharedCommand::help(){
+ try {
+ m->mothurOut("The summary.shared command can only be executed after a successful read.otu command.\n");
+ m->mothurOut("The summary.shared command parameters are label, calc, distance and all. No parameters are required.\n");
+ m->mothurOut("The summary.shared command should be in the following format: \n");
+ m->mothurOut("summary.shared(label=yourLabel, calc=yourEstimators, groups=yourGroups).\n");
+ m->mothurOut("Example summary.shared(label=unique-.01-.03, groups=B-C, calc=sharedchao-sharedace-jabund-sorensonabund-jclass-sorclass-jest-sorest-thetayc-thetan).\n");
+ validCalculator->printCalc("sharedsummary", cout);
+ m->mothurOut("The default value for calc is sharedsobs-sharedchao-sharedace-jabund-sorensonabund-jclass-sorclass-jest-sorest-thetayc-thetan\n");
+ m->mothurOut("The default value for groups is all the groups in your groupfile.\n");
+ m->mothurOut("The distance parameter allows you to indicate you would like a distance file created for each calculator for each label, default=f.\n");
+ m->mothurOut("The label parameter is used to analyze specific labels in your input.\n");
+ m->mothurOut("The all parameter is used to specify if you want the estimate of all your groups together. This estimate can only be made for sharedsobs and sharedchao calculators. The default is false.\n");
+ m->mothurOut("If you use sharedchao and run into memory issues, set all to false. \n");
+ m->mothurOut("The groups parameter allows you to specify which of the groups in your groupfile you would like analyzed. You must enter at least 2 valid groups.\n");
+ m->mothurOut("Note: No spaces between parameter labels (i.e. label), '=' and parameters (i.e.yourLabel).\n\n");
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SummarySharedCommand", "help");
exit(1);
- }
+ }
}
+
//**********************************************************************************************************************
SummarySharedCommand::~SummarySharedCommand(){
- delete input;
- delete read;
- delete util;
+ if (abort == false) {
+ delete read;
+ delete validCalculator;
+ }
}
//**********************************************************************************************************************
int SummarySharedCommand::execute(){
try {
- int count = 1;
+
+ if (abort == true) { return 0; }
+
+ ofstream outputFileHandle, outAll;
+ string outputFileName = outputDir + m->getRootName(m->getSimpleName(globaldata->inputFileName)) + "shared.summary";
//if the users entered no valid calculators don't execute command
if (sumCalculators.size() == 0) { return 0; }
-
- if (format == "sharedfile") {
- read = new ReadOTUFile(globaldata->inputFileName);
- read->read(&*globaldata);
-
- input = globaldata->ginput;
- order = input->getSharedOrderVector();
- }else {
- //you are using a list and a groupfile
- read = new ReadOTUFile(globaldata->inputFileName);
- read->read(&*globaldata);
-
- input = globaldata->ginput;
- SharedList = globaldata->gSharedList;
- order = SharedList->getSharedOrderVector();
+ //check if any calcs can do multiples
+ else{
+ if (all){
+ for (int i = 0; i < sumCalculators.size(); i++) {
+ if (sumCalculators[i]->getMultiple() == true) { mult = true; }
+ }
+ }
}
- //set users groups
- util->setGroups(globaldata->Groups, globaldata->gGroupmap->namesOfGroups, "summary");
-
+ //read first line
+ read = new ReadOTUFile(globaldata->inputFileName);
+ read->read(&*globaldata);
+
+ input = globaldata->ginput;
+ lookup = input->getSharedRAbundVectors();
+ string lastLabel = lookup[0]->getLabel();
+
+ /******************************************************/
+ //output headings for files
+ /******************************************************/
//output estimator names as column headers
+ m->openOutputFile(outputFileName, outputFileHandle);
outputFileHandle << "label" <<'\t' << "comparison" << '\t';
for(int i=0;i<sumCalculators.size();i++){
outputFileHandle << '\t' << sumCalculators[i]->getName();
+ if (sumCalculators[i]->getCols() == 3) { outputFileHandle << "\t" << sumCalculators[i]->getName() << "_lci\t" << sumCalculators[i]->getName() << "_hci"; }
}
outputFileHandle << endl;
+ outputFileHandle.close();
- while(order != NULL){
+ //create file and put column headers for multiple groups file
+ string outAllFileName = ((m->getRootName(globaldata->inputFileName)) + "sharedmultiple.summary");
+ if (mult == true) {
+ m->openOutputFile(outAllFileName, outAll);
+ outputNames.push_back(outAllFileName);
+
+ outAll << "label" <<'\t' << "comparison" << '\t';
+ for(int i=0;i<sumCalculators.size();i++){
+ if (sumCalculators[i]->getMultiple() == true) {
+ outAll << '\t' << sumCalculators[i]->getName();
+ }
+ }
+ outAll << endl;
+ outAll.close();
+ }
+
+ if (lookup.size() < 2) {
+ m->mothurOut("I cannot run the command without at least 2 valid groups.");
+ for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; }
+
+ //close files and clean up
+ remove(outputFileName.c_str());
+ if (mult == true) { remove(outAllFileName.c_str()); }
+ return 0;
+ //if you only have 2 groups you don't need a .sharedmultiple file
+ }else if ((lookup.size() == 2) && (mult == true)) {
+ mult = false;
+ remove(outAllFileName.c_str());
+ outputNames.pop_back();
+ }
+
+ if (m->control_pressed) {
+ if (mult) { remove(outAllFileName.c_str()); }
+ remove(outputFileName.c_str());
+ delete input; globaldata->ginput = NULL;
+ for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; }
+ for(int i=0;i<sumCalculators.size();i++){ delete sumCalculators[i]; }
+ globaldata->Groups.clear();
+ return 0;
+ }
+ /******************************************************/
+
+
+ /******************************************************/
+ //comparison breakup to be used by different processes later
+ numGroups = globaldata->Groups.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 the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
+ set<string> processedLabels;
+ set<string> userLabels = labels;
+
+ //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) {
+ if (mult) { remove(outAllFileName.c_str()); }
+ remove(outputFileName.c_str());
+ delete input; globaldata->ginput = NULL;
+ for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; }
+ for(int i=0;i<sumCalculators.size();i++){ delete sumCalculators[i]; }
+ globaldata->Groups.clear();
+ return 0;
+ }
+
+
+ if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
+ m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
+ process(lookup, outputFileName, outAllFileName);
+
+ processedLabels.insert(lookup[0]->getLabel());
+ userLabels.erase(lookup[0]->getLabel());
+ }
+
+ if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
+ string saveLabel = lookup[0]->getLabel();
+
+ for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; }
+ lookup = input->getSharedRAbundVectors(lastLabel);
+
+ m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
+ process(lookup, outputFileName, outAllFileName);
+
+ processedLabels.insert(lookup[0]->getLabel());
+ userLabels.erase(lookup[0]->getLabel());
+
+ //restore real lastlabel to save below
+ lookup[0]->setLabel(saveLabel);
+ }
+
+ lastLabel = lookup[0]->getLabel();
+
+ //get next line to process
+ //prevent memory leak
+ for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; }
+ lookup = input->getSharedRAbundVectors();
+ }
+
+ if (m->control_pressed) {
+ if (mult) { remove(outAllFileName.c_str()); }
+ remove(outputFileName.c_str());
+ delete input; globaldata->ginput = NULL;
+ for(int i=0;i<sumCalculators.size();i++){ delete sumCalculators[i]; }
+ globaldata->Groups.clear();
+ return 0;
+ }
+
+ //output error messages about any remaining user labels
+ set<string>::iterator it;
+ bool needToRun = false;
+ for (it = userLabels.begin(); it != userLabels.end(); it++) {
+ m->mothurOut("Your file does not include the label " + *it);
+ if (processedLabels.count(lastLabel) != 1) {
+ m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
+ needToRun = true;
+ }else {
+ m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
+ }
+ }
+
+ //run last label if you need to
+ if (needToRun == true) {
+ for (int i = 0; i < lookup.size(); i++) { if (lookup[i] != NULL) { delete lookup[i]; } }
+ lookup = input->getSharedRAbundVectors(lastLabel);
+
+ m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
+ process(lookup, outputFileName, outAllFileName);
+ for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; }
+ }
- if(globaldata->allLines == 1 || globaldata->lines.count(count) == 1 || globaldata->labels.count(order->getLabel()) == 1){
-
- cout << order->getLabel() << '\t' << count << endl;
- util->getSharedVectors(globaldata->Groups, lookup, order); //fills group vectors from order vector. //fills group vectors from order vector.
- //randomize group order
- if (globaldata->getJumble() == "1") { random_shuffle(lookup.begin(), lookup.end()); }
+ //reset groups parameter
+ globaldata->Groups.clear();
+
+ for(int i=0;i<sumCalculators.size();i++){ delete sumCalculators[i]; }
+ delete input; globaldata->ginput = NULL;
+
+ if (m->control_pressed) {
+ remove(outAllFileName.c_str());
+ remove(outputFileName.c_str());
+ return 0;
+ }
+
+ m->mothurOutEndLine();
+ m->mothurOut("Output File Names: "); m->mothurOutEndLine();
+ m->mothurOut(outputFileName); m->mothurOutEndLine();
+ if (mult) { m->mothurOut(outAllFileName); m->mothurOutEndLine(); outputTypes["summary"].push_back(outAllFileName); }
+ for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); } outputTypes["summary"].push_back(outputFileName);
+ m->mothurOutEndLine();
+
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SummarySharedCommand", "execute");
+ exit(1);
+ }
+}
- int n = 1;
- for (int k = 0; k < (lookup.size() - 1); k++) { // pass cdd each set of groups to commpare
- for (int l = n; l < lookup.size(); l++) {
+/***********************************************************/
+int SummarySharedCommand::process(vector<SharedRAbundVector*> thisLookup, string sumFileName, string sumAllFileName) {
+ try {
+ vector< vector<seqDist> > calcDists; //vector containing vectors that contains the summary results for each group compare
+ calcDists.resize(sumCalculators.size()); //one for each calc, this will be used to make .dist files
+
+ #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+ if(processors == 1){
+ driver(thisLookup, 0, numGroups, sumFileName+".temp", sumAllFileName+".temp", calcDists);
+ m->appendFiles((sumFileName + ".temp"), sumFileName);
+ remove((sumFileName + ".temp").c_str());
+ if (mult) {
+ m->appendFiles((sumAllFileName + ".temp"), sumAllFileName);
+ remove((sumAllFileName + ".temp").c_str());
+ }
+ }else{
+ int process = 1;
+ vector<int> processIDS;
+
+ //loop through and create all the processes you want
+ while (process != processors) {
+ int pid = fork();
- outputFileHandle << order->getLabel() << '\t';
+ if (pid > 0) {
+ processIDS.push_back(pid);
+ process++;
+ }else if (pid == 0){
+ driver(thisLookup, lines[process].start, lines[process].end, sumFileName + toString(getpid()) + ".temp", sumAllFileName + toString(getpid()) + ".temp", calcDists);
+
+ //only do this if you want a distance file
+ if (createPhylip) {
+ string tempdistFileName = m->getRootName(m->getSimpleName(sumFileName)) + 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, sumFileName + toString(getpid()) + ".temp", sumAllFileName + toString(getpid()) + ".temp", calcDists);
+ m->appendFiles((sumFileName + toString(getpid()) + ".temp"), sumFileName);
+ remove((sumFileName + toString(getpid()) + ".temp").c_str());
+ if (mult) { m->appendFiles((sumAllFileName + toString(getpid()) + ".temp"), sumAllFileName); }
- //sort groups to be alphanumeric
- if (lookup[k]->getGroup() > lookup[l]->getGroup()) {
- outputFileHandle << (lookup[l]->getGroup() +'\t' + lookup[k]->getGroup()) << '\t'; //print out groups
- }else{
- outputFileHandle << (lookup[k]->getGroup() +'\t' + lookup[l]->getGroup()) << '\t'; //print out groups
+ //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++) {
+ m->appendFiles((sumFileName + toString(processIDS[i]) + ".temp"), sumFileName);
+ remove((sumFileName + toString(processIDS[i]) + ".temp").c_str());
+ if (mult) { remove((sumAllFileName + toString(processIDS[i]) + ".temp").c_str()); }
+
+ if (createPhylip) {
+ string tempdistFileName = m->getRootName(m->getSimpleName(sumFileName)) + toString(processIDS[i]) + ".dist";
+ ifstream intemp;
+ m->openInputFile(tempdistFileName, intemp);
+
+ for (int i = 0; i < calcDists.size(); i++) {
+ 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[i].push_back(tempDist);
+ }
+ }
+ intemp.close();
+ remove(tempdistFileName.c_str());
}
+ }
+
+ }
+ #else
+ driver(thisLookup, 0, numGroups, (sumFileName + ".temp"), (sumAllFileName + ".temp"), calcDists);
+ m->appendFiles((sumFileName + ".temp"), sumFileName);
+ remove((sumFileName + ".temp").c_str());
+ if (mult) {
+ m->appendFiles((sumAllFileName + ".temp"), sumAllFileName);
+ remove((sumAllFileName + ".temp").c_str());
+ }
+ #endif
+
+ if (createPhylip) {
+ for (int i = 0; i < calcDists.size(); i++) {
+ if (m->control_pressed) { break; }
+
+ string distFileName = outputDir + m->getRootName(m->getSimpleName(sumFileName)) + sumCalculators[i]->getName() + "." + thisLookup[0]->getLabel() + ".dist";
+ outputNames.push_back(distFileName);
+ ofstream outDist;
+ m->openOutputFile(distFileName, outDist);
+ outDist.setf(ios::fixed, ios::floatfield); outDist.setf(ios::showpoint);
+
+ //initialize matrix
+ vector< vector<float> > 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;
- for(int i=0;i<sumCalculators.size();i++){
- sumCalculators[i]->getValues(lookup[k], lookup[l]); //saves the calculator outputs
- outputFileHandle << '\t';
- sumCalculators[i]->print(outputFileHandle);
+ matrix[row][column] = dist;
+ matrix[column][row] = dist;
+ }
+
+ //output to file
+ outDist << thisLookup.size() << endl;
+ for (int r=0; r<thisLookup.size(); r++) {
+ //output name
+ string name = thisLookup[r]->getGroup();
+ if (name.length() < 10) { //pad with spaces to make compatible
+ while (name.length() < 10) { name += " "; }
}
- outputFileHandle << endl;
+ outDist << name << '\t';
+
+ //output distances
+ for (int l = 0; l < r; l++) { outDist << matrix[r][l] << '\t'; }
+ outDist << endl;
}
- n++;
+
+ outDist.close();
}
}
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SummarySharedCommand", "process");
+ exit(1);
+ }
+}
+/**************************************************************************************************/
+int SummarySharedCommand::driver(vector<SharedRAbundVector*> thisLookup, int start, int end, string sumFile, string sumAllFile, vector< vector<seqDist> >& calcDists) {
+ try {
- //get next line to process
- if (format == "sharedfile") {
- order = input->getSharedOrderVector();
- }else {
- //you are using a list and a groupfile
- SharedList = input->getSharedListVector(); //get new list vector to process
- if (SharedList != NULL) {
- order = SharedList->getSharedOrderVector(); //gets new order vector with group info.
- }else {
- break;
+ //loop through calculators and add to file all for all calcs that can do mutiple groups
+ if (mult == true) {
+ ofstream outAll;
+ m->openOutputFile(sumAllFile, outAll);
+
+ //output label
+ outAll << thisLookup[0]->getLabel() << '\t';
+
+ //output groups names
+ string outNames = "";
+ for (int j = 0; j < thisLookup.size(); j++) {
+ outNames += thisLookup[j]->getGroup() + "-";
+ }
+ outNames = outNames.substr(0, outNames.length()-1); //rip off extra '-';
+ outAll << outNames << '\t';
+
+ for(int i=0;i<sumCalculators.size();i++){
+ if (sumCalculators[i]->getMultiple() == true) {
+ sumCalculators[i]->getValues(thisLookup);
+
+ if (m->control_pressed) { outAll.close(); return 1; }
+
+ outAll << '\t';
+ sumCalculators[i]->print(outAll);
}
}
- count++;
+ outAll << endl;
+ outAll.close();
}
- //reset groups parameter
- globaldata->Groups.clear(); globaldata->setGroups("");
+ ofstream outputFileHandle;
+ m->openOutputFile(sumFile, outputFileHandle);
+
+ vector<SharedRAbundVector*> subset;
+ for (int k = start; k < end; k++) { // pass cdd each set of groups to compare
+ for (int l = 0; l < k; l++) {
+
+ outputFileHandle << thisLookup[0]->getLabel() << '\t';
+
+ subset.clear(); //clear out old pair of sharedrabunds
+ //add new pair of sharedrabunds
+ subset.push_back(thisLookup[k]); subset.push_back(thisLookup[l]);
+
+ //sort groups to be alphanumeric
+ if (thisLookup[k]->getGroup() > thisLookup[l]->getGroup()) {
+ outputFileHandle << (thisLookup[l]->getGroup() +'\t' + thisLookup[k]->getGroup()) << '\t'; //print out groups
+ }else{
+ outputFileHandle << (thisLookup[k]->getGroup() +'\t' + thisLookup[l]->getGroup()) << '\t'; //print out groups
+ }
+
+ for(int i=0;i<sumCalculators.size();i++) {
+
+ vector<double> tempdata = sumCalculators[i]->getValues(subset); //saves the calculator outputs
+
+ if (m->control_pressed) { outputFileHandle.close(); return 1; }
+
+ outputFileHandle << '\t';
+ sumCalculators[i]->print(outputFileHandle);
+
+ seqDist temp(l, k, tempdata[0]);
+ calcDists[i].push_back(temp);
+ }
+ outputFileHandle << endl;
+ }
+ }
+
+ outputFileHandle.close();
+
return 0;
}
catch(exception& e) {
- cout << "Standard Error: " << e.what() << " has occurred in the SummarySharedCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+ m->errorOut(e, "SummarySharedCommand", "driver");
exit(1);
}
- catch(...) {
- cout << "An unknown error has occurred in the SummarySharedCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
- exit(1);
- }
}
+/**************************************************************************************************/
+
-/***********************************************************/
projects / mothur.git / blobdiff