From 615301e57c25e241356a9c2380648d117709458d Mon Sep 17 00:00:00 2001 From: Kathryn Iverson Date: Mon, 16 Sep 2013 10:18:12 -0400 Subject: [PATCH] changing command name classify.shared to classifyrf.shared --- classifyrfsharedcommand.cpp | 407 ++++++++++++++++++++++++++++++++++++ classifyrfsharedcommand.h | 54 +++++ commandfactory.cpp | 10 +- 3 files changed, 466 insertions(+), 5 deletions(-) create mode 100755 classifyrfsharedcommand.cpp create mode 100755 classifyrfsharedcommand.h diff --git a/classifyrfsharedcommand.cpp b/classifyrfsharedcommand.cpp new file mode 100755 index 0000000..d2cd9f9 --- /dev/null +++ b/classifyrfsharedcommand.cpp @@ -0,0 +1,407 @@ +// +// classifysharedcommand.cpp +// Mothur +// +// Created by Abu Zaher Md. Faridee on 8/13/12. +// Copyright (c) 2012 Schloss Lab. All rights reserved. +// + +#include "classifyrfsharedcommand.h" +#include "randomforest.hpp" +#include "decisiontree.hpp" +#include "rftreenode.hpp" + +//********************************************************************************************************************** +vector ClassifyRFSharedCommand::setParameters(){ + try { + //CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors); + CommandParameter pshared("shared", "InputTypes", "", "", "none", "none", "none","summary",false,true,true); parameters.push_back(pshared); + CommandParameter pdesign("design", "InputTypes", "", "", "none", "none", "none","",false,true,true); parameters.push_back(pdesign); + CommandParameter potupersplit("otupersplit", "Multiple", "log2-squareroot", "log2", "", "", "","",false,false); parameters.push_back(potupersplit); + CommandParameter psplitcriteria("splitcriteria", "Multiple", "gainratio-infogain", "gainratio", "", "", "","",false,false); parameters.push_back(psplitcriteria); + CommandParameter pnumtrees("numtrees", "Number", "", "100", "", "", "","",false,false); parameters.push_back(pnumtrees); + + // parameters related to pruning + CommandParameter pdopruning("prune", "Boolean", "", "T", "", "", "", "", false, false); parameters.push_back(pdopruning); + CommandParameter ppruneaggrns("pruneaggressiveness", "Number", "", "0.9", "", "", "", "", false, false); parameters.push_back(ppruneaggrns); + CommandParameter pdiscardhetrees("discarderrortrees", "Boolean", "", "T", "", "", "", "", false, false); parameters.push_back(pdiscardhetrees); + CommandParameter phetdiscardthreshold("errorthreshold", "Number", "", "0.4", "", "", "", "", false, false); parameters.push_back(phetdiscardthreshold); + CommandParameter psdthreshold("stdthreshold", "Number", "", "0.0", "", "", "", "", false, false); parameters.push_back(psdthreshold); + // pruning params end + + CommandParameter pgroups("groups", "String", "", "", "", "", "","",false,false); parameters.push_back(pgroups); + CommandParameter plabel("label", "String", "", "", "", "", "","",false,false); parameters.push_back(plabel); + CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir); + CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir); + + vector myArray; + for (int i = 0; i < parameters.size(); i++) { myArray.push_back(parameters[i].name); } + return myArray; + } + catch(exception& e) { + m->errorOut(e, "ClassifySharedCommand", "setParameters"); + exit(1); + } +} +//********************************************************************************************************************** +string ClassifyRFSharedCommand::getHelpString(){ + try { + string helpString = ""; + helpString += "The classify.shared command allows you to ....\n"; + helpString += "The classify.shared command parameters are: shared, design, label, groups, otupersplit.\n"; + helpString += "The label parameter is used to analyze specific labels in your input.\n"; + helpString += "The groups parameter allows you to specify which of the groups in your designfile you would like analyzed.\n"; + helpString += "The classify.shared should be in the following format: \n"; + helpString += "classify.shared(shared=yourSharedFile, design=yourDesignFile)\n"; + return helpString; + } + catch(exception& e) { + m->errorOut(e, "ClassifySharedCommand", "getHelpString"); + exit(1); + } +} +//********************************************************************************************************************** +string ClassifyRFSharedCommand::getOutputPattern(string type) { + try { + string pattern = ""; + + if (type == "summary") { pattern = "[filename],[distance],summary"; } //makes file like: amazon.0.03.fasta + else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true; } + + return pattern; + } + catch(exception& e) { + m->errorOut(e, "ClassifySharedCommand", "getOutputPattern"); + exit(1); + } +} +//********************************************************************************************************************** + +ClassifyRFSharedCommand::ClassifyRFSharedCommand() { + try { + abort = true; calledHelp = true; + setParameters(); + vector tempOutNames; + outputTypes["summary"] = tempOutNames; + } + catch(exception& e) { + m->errorOut(e, "ClassifySharedCommand", "ClassifySharedCommand"); + exit(1); + } +} + +//********************************************************************************************************************** +ClassifyRFSharedCommand::ClassifyRFSharedCommand(string option) { + try { + abort = false; calledHelp = false; + allLines = 1; + + //allow user to run help + if(option == "help") { help(); abort = true; calledHelp = true; } + else if(option == "citation") { citation(); abort = true; calledHelp = true;} + + else { + //valid paramters for this command + vector myArray = setParameters(); + + OptionParser parser(option); + map parameters = parser.getParameters(); + + ValidParameters validParameter; + map::iterator it; + //check to make sure all parameters are valid for command + for (it = parameters.begin(); it != parameters.end(); it++) { + if (validParameter.isValidParameter(it->first, myArray, it->second) != true) { abort = true; } + } + vector tempOutNames; + outputTypes["summary"] = 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 (inputDir == "not found"){ inputDir = ""; } + else { + string path; + it = parameters.find("shared"); + //user has given a shared file + if(it != parameters.end()){ + path = m->hasPath(it->second); + //if the user has not given a path then, add inputdir. else leave path alone. + if (path == "") { parameters["shared"] = inputDir + it->second; } + } + + it = parameters.find("design"); + //user has given a design file + if(it != parameters.end()){ + path = m->hasPath(it->second); + //if the user has not given a path then, add inputdir. else leave path alone. + if (path == "") { parameters["design"] = inputDir + it->second; } + } + + } + //check for parameters + //get shared file, it is required + sharedfile = validParameter.validFile(parameters, "shared", true); + if (sharedfile == "not open") { sharedfile = ""; abort = true; } + else if (sharedfile == "not found") { + //if there is a current shared file, use it + sharedfile = m->getSharedFile(); + 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 { m->setSharedFile(sharedfile); } + + //get design file, it is required + designfile = validParameter.validFile(parameters, "design", true); + if (designfile == "not open") { sharedfile = ""; abort = true; } + else if (designfile == "not found") { + //if there is a current shared file, use it + designfile = m->getDesignFile(); + if (designfile != "") { m->mothurOut("Using " + designfile + " as input file for the design parameter."); m->mothurOutEndLine(); } + else { m->mothurOut("You have no current designfile and the design parameter is required."); m->mothurOutEndLine(); abort = true; } + }else { m->setDesignFile(designfile); } + + + //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 = m->hasPath(sharedfile); //if user entered a file with a path then preserve it + } + + // NEW CODE for OTU per split selection criteria + string temp = validParameter.validFile(parameters, "splitcriteria", false); + if (temp == "not found") { temp = "gainratio"; } + if ((temp == "gainratio") || (temp == "infogain")) { + treeSplitCriterion = temp; + } else { m->mothurOut("Not a valid tree splitting criterio. Valid tree splitting criteria are 'gainratio' and 'infogain'."); + m->mothurOutEndLine(); + abort = true; + } + + temp = validParameter.validFile(parameters, "numtrees", false); if (temp == "not found"){ temp = "100"; } + m->mothurConvert(temp, numDecisionTrees); + + // parameters for pruning + temp = validParameter.validFile(parameters, "prune", false); + if (temp == "not found") { temp = "f"; } + doPruning = m->isTrue(temp); + + temp = validParameter.validFile(parameters, "pruneaggressiveness", false); + if (temp == "not found") { temp = "0.9"; } + m->mothurConvert(temp, pruneAggressiveness); + + temp = validParameter.validFile(parameters, "discarderrortrees", false); + if (temp == "not found") { temp = "f"; } + discardHighErrorTrees = m->isTrue(temp); + + temp = validParameter.validFile(parameters, "errorthreshold", false); + if (temp == "not found") { temp = "0.4"; } + m->mothurConvert(temp, highErrorTreeDiscardThreshold); + + temp = validParameter.validFile(parameters, "otupersplit", false); + if (temp == "not found") { temp = "log2"; } + if ((temp == "squareroot") || (temp == "log2")) { + optimumFeatureSubsetSelectionCriteria = temp; + } else { m->mothurOut("Not a valid OTU per split selection method. Valid OTU per split selection methods are 'log2' and 'squareroot'."); + m->mothurOutEndLine(); + abort = true; + } + + temp = validParameter.validFile(parameters, "stdthreshold", false); + if (temp == "not found") { temp = "0.0"; } + m->mothurConvert(temp, featureStandardDeviationThreshold); + + // end of pruning params + + //Groups must be checked later to make sure they are valid. SharedUtilities has functions of check the validity, just make to so m->setGroups() after the checks. If you are using these with a shared file no need to check the SharedRAbundVector class will call SharedUtilites for you, kinda nice, huh? + string groups = validParameter.validFile(parameters, "groups", false); + if (groups == "not found") { groups = ""; } + else { m->splitAtDash(groups, Groups); } + m->setGroups(Groups); + + //Commonly used to process list, rabund, sabund, shared and relabund files. Look at "smart distancing" examples below in the execute function. + string label = validParameter.validFile(parameters, "label", false); + if (label == "not found") { label = ""; } + else { + if(label != "all") { m->splitAtDash(label, labels); allLines = 0; } + else { allLines = 1; } + } + } + + } + catch(exception& e) { + m->errorOut(e, "ClassifySharedCommand", "ClassifySharedCommand"); + exit(1); + } +} +//********************************************************************************************************************** +int ClassifyRFSharedCommand::execute() { + try { + + if (abort == true) { if (calledHelp) { return 0; } return 2; } + + InputData input(sharedfile, "sharedfile"); + vector lookup = input.getSharedRAbundVectors(); + + //read design file + designMap.readDesignMap(designfile); + + string lastLabel = lookup[0]->getLabel(); + set processedLabels; + set 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) { for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; } return 0; } + + if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){ + + m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine(); + + processSharedAndDesignData(lookup); + + 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(); + processSharedAndDesignData(lookup); + + processedLabels.insert(lookup[0]->getLabel()); + userLabels.erase(lookup[0]->getLabel()); + + //restore real lastlabel to save below + lookup[0]->setLabel(saveLabel); + } + + lastLabel = lookup[0]->getLabel(); + //prevent memory leak + for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; lookup[i] = NULL; } + + if (m->control_pressed) { return 0; } + + //get next line to process + lookup = input.getSharedRAbundVectors(); + } + + if (m->control_pressed) { return 0; } + + //output error messages about any remaining user labels + set::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(); + + processSharedAndDesignData(lookup); + + for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; } + + } + + m->mothurOutEndLine(); + m->mothurOut("Output File Names: "); m->mothurOutEndLine(); + for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); } + m->mothurOutEndLine(); + + return 0; + + } + catch(exception& e) { + m->errorOut(e, "ClassifySharedCommand", "execute"); + exit(1); + } +} +//********************************************************************************************************************** + +void ClassifyRFSharedCommand::processSharedAndDesignData(vector lookup){ + try { +// for (int i = 0; i < designMap->getNamesOfGroups().size(); i++) { +// string groupName = designMap->getNamesOfGroups()[i]; +// cout << groupName << endl; +// } + +// for (int i = 0; i < designMap->getNumSeqs(); i++) { +// string sharedGroupName = designMap->getNamesSeqs()[i]; +// string treatmentName = designMap->getGroup(sharedGroupName); +// cout << sharedGroupName << " : " << treatmentName << endl; +// } + + map treatmentToIntMap; + map intToTreatmentMap; + for (int i = 0; i < designMap.getNumGroups(); i++) { + string treatmentName = designMap.getNamesOfGroups()[i]; + treatmentToIntMap[treatmentName] = i; + intToTreatmentMap[i] = treatmentName; + } + + int numSamples = lookup.size(); + int numFeatures = lookup[0]->getNumBins(); + + int numRows = numSamples; + int numColumns = numFeatures + 1; // extra one space needed for the treatment/outcome + + vector< vector > dataSet(numRows, vector(numColumns, 0)); + + vector names; + + for (int i = 0; i < lookup.size(); i++) { + string sharedGroupName = lookup[i]->getGroup(); + names.push_back(sharedGroupName); + string treatmentName = designMap.getGroup(sharedGroupName); + + int j = 0; + for (; j < lookup[i]->getNumBins(); j++) { + int otuCount = lookup[i]->getAbundance(j); + dataSet[i][j] = otuCount; + } + dataSet[i][j] = treatmentToIntMap[treatmentName]; + } + + RandomForest randomForest(dataSet, numDecisionTrees, treeSplitCriterion, doPruning, pruneAggressiveness, discardHighErrorTrees, highErrorTreeDiscardThreshold, optimumFeatureSubsetSelectionCriteria, featureStandardDeviationThreshold); + + randomForest.populateDecisionTrees(); + randomForest.calcForrestErrorRate(); + randomForest.printConfusionMatrix(intToTreatmentMap); + + map variables; + variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + "RF."; + variables["[distance]"] = lookup[0]->getLabel(); + string filename = getOutputFileName("summary", variables); + outputNames.push_back(filename); outputTypes["summary"].push_back(filename); + randomForest.calcForrestVariableImportance(filename); + + // + map variable; + variable["[filename]"] = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + "misclassifications."; + variable["[distance]"] = lookup[0]->getLabel(); + string mc_filename = getOutputFileName("summary", variable); + outputNames.push_back(mc_filename); outputTypes["summary"].push_back(mc_filename); + randomForest.getMissclassifications(mc_filename, intToTreatmentMap, names); + // + + m->mothurOutEndLine(); + } + catch(exception& e) { + m->errorOut(e, "ClassifySharedCommand", "processSharedAndDesignData"); + exit(1); + } +} +//********************************************************************************************************************** + diff --git a/classifyrfsharedcommand.h b/classifyrfsharedcommand.h new file mode 100755 index 0000000..6a948b2 --- /dev/null +++ b/classifyrfsharedcommand.h @@ -0,0 +1,54 @@ +// +// classifysharedcommand.h +// Mothur +// +// Created by Abu Zaher Md. Faridee on 8/13/12. +// Copyright (c) 2012 Schloss Lab. All rights reserved. +// + +#ifndef __Mothur__classifyrfsharedcommand__ +#define __Mothur__classifyrfsharedcommand__ + +#include "command.hpp" +#include "inputdata.h" + +class ClassifyRFSharedCommand : public Command { +public: + ClassifyRFSharedCommand(); + ClassifyRFSharedCommand(string); + ~ClassifyRFSharedCommand() {}; + + vector setParameters(); + string getCommandName() { return "classifyrf.shared"; } + string getCommandCategory() { return "OTU-Based Approaches"; } + string getHelpString(); + string getOutputPattern(string); + string getCitation() { return "http://www.mothur.org/wiki/Classifyrf.shared\n"; } + string getDescription() { return "implements the random forest machine learning algorithm to identify OTUs that can be used to differentiate between various groups of samples"; } + int execute(); + + void help() { m->mothurOut(getHelpString()); } + +private: + bool abort; + string outputDir; + vector outputNames, Groups; + + string sharedfile, designfile; + set labels; + bool allLines; + + int processors; + bool useTiming; + + GroupMap designMap; + + int numDecisionTrees; + string treeSplitCriterion, optimumFeatureSubsetSelectionCriteria; + bool doPruning, discardHighErrorTrees; + double pruneAggressiveness, highErrorTreeDiscardThreshold, featureStandardDeviationThreshold; + + void processSharedAndDesignData(vector lookup); +}; + +#endif /* defined(__Mothur__classifyrfsharedcommand__) */ diff --git a/commandfactory.cpp b/commandfactory.cpp index 6ceabbf..03c3e49 100644 --- a/commandfactory.cpp +++ b/commandfactory.cpp @@ -135,7 +135,7 @@ #include "makecontigscommand.h" #include "loadlogfilecommand.h" #include "sffmultiplecommand.h" -#include "classifysharedcommand.h" +#include "classifyrfsharedcommand.h" #include "filtersharedcommand.h" #include "primerdesigncommand.h" #include "getdistscommand.h" @@ -306,7 +306,7 @@ CommandFactory::CommandFactory(){ commands["make.table"] = "make.table"; commands["sff.multiple"] = "sff.multiple"; commands["quit"] = "MPIEnabled"; - commands["classify.shared"] = "classify.shared"; + commands["classifyrf.shared"] = "classifyrf.shared"; commands["filter.shared"] = "filter.shared"; commands["primer.design"] = "primer.design"; commands["get.dists"] = "get.dists"; @@ -533,7 +533,7 @@ Command* CommandFactory::getCommand(string commandName, string optionString){ else if(commandName == "make.contigs") { command = new MakeContigsCommand(optionString); } else if(commandName == "load.logfile") { command = new LoadLogfileCommand(optionString); } else if(commandName == "sff.multiple") { command = new SffMultipleCommand(optionString); } - else if(commandName == "classify.shared") { command = new ClassifySharedCommand(optionString); } + else if(commandName == "classifyrf.shared") { command = new ClassifyRFSharedCommand(optionString); } else if(commandName == "filter.shared") { command = new FilterSharedCommand(optionString); } else if(commandName == "primer.design") { command = new PrimerDesignCommand(optionString); } else if(commandName == "get.dists") { command = new GetDistsCommand(optionString); } @@ -701,7 +701,7 @@ Command* CommandFactory::getCommand(string commandName, string optionString, str else if(commandName == "make.contigs") { pipecommand = new MakeContigsCommand(optionString); } else if(commandName == "load.logfile") { pipecommand = new LoadLogfileCommand(optionString); } else if(commandName == "sff.multiple") { pipecommand = new SffMultipleCommand(optionString); } - else if(commandName == "classify.shared") { pipecommand = new ClassifySharedCommand(optionString); } + else if(commandName == "classifyrf.shared") { pipecommand = new ClassifyRFSharedCommand(optionString); } else if(commandName == "filter.shared") { pipecommand = new FilterSharedCommand(optionString); } else if(commandName == "primer.design") { pipecommand = new PrimerDesignCommand(optionString); } else if(commandName == "get.dists") { pipecommand = new GetDistsCommand(optionString); } @@ -855,7 +855,7 @@ Command* CommandFactory::getCommand(string commandName){ else if(commandName == "make.contigs") { shellcommand = new MakeContigsCommand(); } else if(commandName == "load.logfile") { shellcommand = new LoadLogfileCommand(); } else if(commandName == "sff.multiple") { shellcommand = new SffMultipleCommand(); } - else if(commandName == "classify.shared") { shellcommand = new ClassifySharedCommand(); } + else if(commandName == "classifyrf.shared") { shellcommand = new ClassifyRFSharedCommand(); } else if(commandName == "filter.shared") { shellcommand = new FilterSharedCommand(); } else if(commandName == "primer.design") { shellcommand = new PrimerDesignCommand(); } else if(commandName == "get.dists") { shellcommand = new GetDistsCommand(); } -- 2.39.2