From 17d3033f832c6b8fff5875eb4de29f44a926d4dd Mon Sep 17 00:00:00 2001 From: westcott Date: Tue, 8 Feb 2011 15:42:06 +0000 Subject: [PATCH] added homova command --- Mothur.xcodeproj/project.pbxproj | 6 + amovacommand.cpp | 2 +- commandfactory.cpp | 5 + homovacommand.cpp | 959 +++++++++++++++++++++++++++++++ homovacommand.h | 66 +++ 5 files changed, 1037 insertions(+), 1 deletion(-) create mode 100644 homovacommand.cpp create mode 100644 homovacommand.h diff --git a/Mothur.xcodeproj/project.pbxproj b/Mothur.xcodeproj/project.pbxproj index c3a3d00..7ddaecc 100644 --- a/Mothur.xcodeproj/project.pbxproj +++ b/Mothur.xcodeproj/project.pbxproj @@ -14,6 +14,7 @@ A713EBED12DC7C5E000092AC /* nmdscommand.cpp in Sources */ = {isa = PBXBuildFile; fileRef = A713EBEC12DC7C5E000092AC /* nmdscommand.cpp */; }; A71FE12C12EDF72400963CA7 /* mergegroupscommand.cpp in Sources */ = {isa = PBXBuildFile; fileRef = A71FE12B12EDF72400963CA7 /* mergegroupscommand.cpp */; }; A727864412E9E28C00F86ABA /* removerarecommand.cpp in Sources */ = {isa = PBXBuildFile; fileRef = A727864312E9E28C00F86ABA /* removerarecommand.cpp */; }; + A75790591301749D00A30DAB /* homovacommand.cpp in Sources */ = {isa = PBXBuildFile; fileRef = A75790581301749D00A30DAB /* homovacommand.cpp */; }; A7A61F2D130062E000E05B6B /* amovacommand.cpp in Sources */ = {isa = PBXBuildFile; fileRef = A7A61F2C130062E000E05B6B /* amovacommand.cpp */; }; A7E9B88112D37EC400DA6239 /* ace.cpp in Sources */ = {isa = PBXBuildFile; fileRef = A7E9B64F12D37EC300DA6239 /* ace.cpp */; }; A7E9B88212D37EC400DA6239 /* aligncommand.cpp in Sources */ = {isa = PBXBuildFile; fileRef = A7E9B65112D37EC300DA6239 /* aligncommand.cpp */; }; @@ -316,6 +317,8 @@ A71FE12B12EDF72400963CA7 /* mergegroupscommand.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = mergegroupscommand.cpp; sourceTree = ""; }; A727864212E9E28C00F86ABA /* removerarecommand.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = removerarecommand.h; sourceTree = ""; }; A727864312E9E28C00F86ABA /* removerarecommand.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = removerarecommand.cpp; sourceTree = ""; }; + A75790571301749D00A30DAB /* homovacommand.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = homovacommand.h; sourceTree = ""; }; + A75790581301749D00A30DAB /* homovacommand.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = homovacommand.cpp; sourceTree = ""; }; A7A61F1A130035C800E05B6B /* LICENSE */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; path = LICENSE; sourceTree = ""; }; A7A61F2B130062E000E05B6B /* amovacommand.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = amovacommand.h; sourceTree = ""; }; A7A61F2C130062E000E05B6B /* amovacommand.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = amovacommand.cpp; sourceTree = ""; }; @@ -1112,6 +1115,8 @@ A7E9B72312D37EC400DA6239 /* heatmapsimcommand.h */, A7E9B72812D37EC400DA6239 /* helpcommand.cpp */, A7E9B72912D37EC400DA6239 /* helpcommand.h */, + A75790571301749D00A30DAB /* homovacommand.h */, + A75790581301749D00A30DAB /* homovacommand.cpp */, A7E9B72B12D37EC400DA6239 /* indicatorcommand.cpp */, A7E9B72C12D37EC400DA6239 /* indicatorcommand.h */, A7E9B73B12D37EC400DA6239 /* libshuffcommand.cpp */, @@ -1909,6 +1914,7 @@ A71FE12C12EDF72400963CA7 /* mergegroupscommand.cpp in Sources */, 7E6BE10A12F710D8007ADDBE /* refchimeratest.cpp in Sources */, A7A61F2D130062E000E05B6B /* amovacommand.cpp in Sources */, + A75790591301749D00A30DAB /* homovacommand.cpp in Sources */, ); runOnlyForDeploymentPostprocessing = 0; }; diff --git a/amovacommand.cpp b/amovacommand.cpp index d46cf74..c0b9aec 100644 --- a/amovacommand.cpp +++ b/amovacommand.cpp @@ -324,7 +324,7 @@ void AmovaCommand::help(){ m->mothurOut("The amova command parameters are phylip, iters, groups, label, design, sets and processors. The design parameter is required.\n"); m->mothurOut("The design parameter allows you to assign your groups to sets when you are running amova. It is required. \n"); m->mothurOut("The design file looks like the group file. It is a 2 column tab delimited file, where the first column is the group name and the second column is the set the group belongs to.\n"); - m->mothurOut("The sets parameter allows you to specify which of the sets in your designfile you would like to analyze. The set names are separated by dashes. THe default is all sets in the designfile.\n"); + m->mothurOut("The sets parameter allows you to specify which of the sets in your designfile you would like to analyze. The set names are separated by dashes. THe default is all sets in the designfile. To run the pairwise comparisons of all sets use sets=all.\n"); m->mothurOut("The iters parameter allows you to set number of randomization for the P value. The default is 1000. \n"); m->mothurOut("The groups parameter allows you to specify which of the groups you would like included. The group names are separated by dashes. groups=all will find all pairwise comparisons. \n"); m->mothurOut("The label parameter allows you to select what distance levels you would like, and are also separated by dashes.\n"); diff --git a/commandfactory.cpp b/commandfactory.cpp index 42567e4..924e3f7 100644 --- a/commandfactory.cpp +++ b/commandfactory.cpp @@ -110,6 +110,7 @@ #include "removerarecommand.h" #include "mergegroupscommand.h" #include "amovacommand.h" +#include "homovacommand.h" /*******************************************************/ @@ -222,6 +223,7 @@ CommandFactory::CommandFactory(){ commands["nmds"] = "nmds"; commands["remove.rare"] = "remove.rare"; commands["amova"] = "amova"; + commands["homova"] = "homova"; commands["merge.groups"] = "merge.groups"; commands["pairwise.seqs"] = "MPIEnabled"; commands["pipeline.pds"] = "MPIEnabled"; @@ -385,6 +387,7 @@ Command* CommandFactory::getCommand(string commandName, string optionString){ else if(commandName == "remove.rare") { command = new RemoveRareCommand(optionString); } else if(commandName == "merge.groups") { command = new MergeGroupsCommand(optionString); } else if(commandName == "amova") { command = new AmovaCommand(optionString); } + else if(commandName == "homova") { command = new HomovaCommand(optionString); } else { command = new NoCommand(optionString); } return command; @@ -513,6 +516,7 @@ Command* CommandFactory::getCommand(string commandName, string optionString, str else if(commandName == "remove.rare") { pipecommand = new RemoveRareCommand(optionString); } else if(commandName == "merge.groups") { pipecommand = new MergeGroupsCommand(optionString); } else if(commandName == "amova") { pipecommand = new AmovaCommand(optionString); } + else if(commandName == "homova") { pipecommand = new HomovaCommand(optionString); } else { pipecommand = new NoCommand(optionString); } return pipecommand; @@ -629,6 +633,7 @@ Command* CommandFactory::getCommand(string commandName){ else if(commandName == "remove.rare") { shellcommand = new RemoveRareCommand(); } else if(commandName == "merge.groups") { shellcommand = new MergeGroupsCommand(); } else if(commandName == "amova") { shellcommand = new AmovaCommand(); } + else if(commandName == "homova") { shellcommand = new HomovaCommand(); } else { shellcommand = new NoCommand(); } return shellcommand; diff --git a/homovacommand.cpp b/homovacommand.cpp new file mode 100644 index 0000000..f9bfb10 --- /dev/null +++ b/homovacommand.cpp @@ -0,0 +1,959 @@ +/* + * homovacommand.cpp + * mothur + * + * Created by westcott on 2/8/11. + * Copyright 2011 Schloss Lab. All rights reserved. + * + */ + +#include "homovacommand.h" +#include "sharedutilities.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 HomovaCommand::getValidParameters(){ + try { + string Array[] = {"groups","label","outputdir","iters","phylip","design","sets","processors","inputdir"}; + vector myArray (Array, Array+(sizeof(Array)/sizeof(string))); + return myArray; + } + catch(exception& e) { + m->errorOut(e, "HomovaCommand", "getValidParameters"); + exit(1); + } +} +//********************************************************************************************************************** +HomovaCommand::HomovaCommand(){ + try { + abort = true; + //initialize outputTypes + vector tempOutNames; + outputTypes["homova"] = tempOutNames; + } + catch(exception& e) { + m->errorOut(e, "HomovaCommand", "HomovaCommand"); + exit(1); + } +} +//********************************************************************************************************************** +vector HomovaCommand::getRequiredParameters(){ + try { + string Array[] = {"design"}; + vector myArray (Array, Array+(sizeof(Array)/sizeof(string))); + return myArray; + } + catch(exception& e) { + m->errorOut(e, "HomovaCommand", "getRequiredParameters"); + exit(1); + } +} +//********************************************************************************************************************** +vector HomovaCommand::getRequiredFiles(){ + try { + string Array[] = {}; + vector myArray (Array, Array+(sizeof(Array)/sizeof(string))); + return myArray; + } + catch(exception& e) { + m->errorOut(e, "HomovaCommand", "getRequiredFiles"); + exit(1); + } +} +//********************************************************************************************************************** + +HomovaCommand::HomovaCommand(string option) { + try { + globaldata = GlobalData::getInstance(); + abort = false; + allLines = 1; + labels.clear(); + + //allow user to run help + if(option == "help") { help(); abort = true; } + + else { + //valid paramters for this command + string AlignArray[] = {"groups","label","outputdir","iters","phylip","design","sets","processors","inputdir"}; + vector myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string))); + + OptionParser parser(option); + map parameters = parser.getParameters(); + + ValidParameters validParameter; + + //check to make sure all parameters are valid for command + map::iterator it; + for (it = parameters.begin(); it != parameters.end(); it++) { + if (validParameter.isValidParameter(it->first, myArray, it->second) != true) { abort = true; } + } + + //initialize outputTypes + vector tempOutNames; + outputTypes["homova"] = 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 = ""; } + + //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("design"); + //user has given a template 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; } + } + + it = parameters.find("phylip"); + //user has given a template 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["phylip"] = inputDir + it->second; } + } + } + + phylipfile = validParameter.validFile(parameters, "phylip", true); + if (phylipfile == "not open") { phylipfile = ""; abort = true; } + else if (phylipfile == "not found") { phylipfile = ""; } + else { globaldata->newRead(); format = "phylip"; globaldata->setPhylipFile(phylipfile); } + + //check for required parameters + designfile = validParameter.validFile(parameters, "design", true); + if (designfile == "not open") { abort = true; } + else if (designfile == "not found") { designfile = ""; m->mothurOut("You must provide an design file."); m->mothurOutEndLine(); abort = true; } + + //make sure the user has already run the read.otu command + if ((globaldata->getSharedFile() == "")) { + if ((phylipfile == "")) { + m->mothurOut("You must read a list and a group, a shared file or provide a distance matrix before you can use the homova command."); m->mothurOutEndLine(); abort = true; + } + }else { sharedfile = globaldata->getSharedFile(); } + + //use distance matrix if one is provided + if ((sharedfile != "") && (phylipfile != "")) { sharedfile = ""; } + + //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; + } + + groups = validParameter.validFile(parameters, "groups", false); + if (groups == "not found") { groups = ""; } + else { + m->splitAtDash(groups, Groups); + globaldata->Groups = Groups; + } + + sets = validParameter.validFile(parameters, "sets", false); + if (sets == "not found") { sets = ""; pickedGroups = false; } + else { + pickedGroups = true; + m->splitAtDash(sets, Sets); + } + + + string temp = validParameter.validFile(parameters, "iters", false); if (temp == "not found") { temp = "1000"; } + convert(temp, iters); + + temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found"){ temp = "1"; } + convert(temp, processors); + + vector Estimators; + calc = validParameter.validFile(parameters, "calc", false); + if (calc == "not found") { calc = "morisitahorn"; } + m->splitAtDash(calc, Estimators); + + if (abort == false) { + + ValidCalculators* validCalculator = new ValidCalculators(); + + for (int i=0; iisValidCalculator("treegroup", Estimators[i]) == true) { + if (Estimators[i] == "sharedsobs") { + calculators.push_back(new SharedSobsCS()); + }else if (Estimators[i] == "sharedchao") { + calculators.push_back(new SharedChao1()); + }else if (Estimators[i] == "sharedace") { + calculators.push_back(new SharedAce()); + }else if (Estimators[i] == "jabund") { + calculators.push_back(new JAbund()); + }else if (Estimators[i] == "sorabund") { + calculators.push_back(new SorAbund()); + }else if (Estimators[i] == "jclass") { + calculators.push_back(new Jclass()); + }else if (Estimators[i] == "sorclass") { + calculators.push_back(new SorClass()); + }else if (Estimators[i] == "jest") { + calculators.push_back(new Jest()); + }else if (Estimators[i] == "sorest") { + calculators.push_back(new SorEst()); + }else if (Estimators[i] == "thetayc") { + calculators.push_back(new ThetaYC()); + }else if (Estimators[i] == "thetan") { + calculators.push_back(new ThetaN()); + }else if (Estimators[i] == "kstest") { + calculators.push_back(new KSTest()); + }else if (Estimators[i] == "sharednseqs") { + calculators.push_back(new SharedNSeqs()); + }else if (Estimators[i] == "ochiai") { + calculators.push_back(new Ochiai()); + }else if (Estimators[i] == "anderberg") { + calculators.push_back(new Anderberg()); + }else if (Estimators[i] == "kulczynski") { + calculators.push_back(new Kulczynski()); + }else if (Estimators[i] == "kulczynskicody") { + calculators.push_back(new KulczynskiCody()); + }else if (Estimators[i] == "lennon") { + calculators.push_back(new Lennon()); + }else if (Estimators[i] == "morisitahorn") { + calculators.push_back(new MorHorn()); + }else if (Estimators[i] == "braycurtis") { + calculators.push_back(new BrayCurtis()); + }else if (Estimators[i] == "whittaker") { + calculators.push_back(new Whittaker()); + }else if (Estimators[i] == "odum") { + calculators.push_back(new Odum()); + }else if (Estimators[i] == "canberra") { + calculators.push_back(new Canberra()); + }else if (Estimators[i] == "structeuclidean") { + calculators.push_back(new StructEuclidean()); + }else if (Estimators[i] == "structchord") { + calculators.push_back(new StructChord()); + }else if (Estimators[i] == "hellinger") { + calculators.push_back(new Hellinger()); + }else if (Estimators[i] == "manhattan") { + calculators.push_back(new Manhattan()); + }else if (Estimators[i] == "structpearson") { + calculators.push_back(new StructPearson()); + }else if (Estimators[i] == "soergel") { + calculators.push_back(new Soergel()); + }else if (Estimators[i] == "spearman") { + calculators.push_back(new Spearman()); + }else if (Estimators[i] == "structkulczynski") { + calculators.push_back(new StructKulczynski()); + }else if (Estimators[i] == "speciesprofile") { + calculators.push_back(new SpeciesProfile()); + }else if (Estimators[i] == "hamming") { + calculators.push_back(new Hamming()); + }else if (Estimators[i] == "structchi2") { + calculators.push_back(new StructChi2()); + }else if (Estimators[i] == "gower") { + calculators.push_back(new Gower()); + }else if (Estimators[i] == "memchi2") { + calculators.push_back(new MemChi2()); + }else if (Estimators[i] == "memchord") { + calculators.push_back(new MemChord()); + }else if (Estimators[i] == "memeuclidean") { + calculators.push_back(new MemEuclidean()); + }else if (Estimators[i] == "mempearson") { + calculators.push_back(new MemPearson()); + } + } + } + } + } + + } + catch(exception& e) { + m->errorOut(e, "HomovaCommand", "HomovaCommand"); + exit(1); + } +} + +//********************************************************************************************************************** + +void HomovaCommand::help(){ + try { + m->mothurOut("The homova command can only be executed after a successful read.otu command of a list and group or shared file, or by providing a phylip formatted distance matrix.\n"); + m->mothurOut("The homova command outputs a .homova file. \n"); + m->mothurOut("The homova command parameters are phylip, iters, groups, label, design, sets and processors. The design parameter is required.\n"); + m->mothurOut("The design parameter allows you to assign your groups to sets when you are running amova. It is required. \n"); + m->mothurOut("The design file looks like the group file. It is a 2 column tab delimited file, where the first column is the group name and the second column is the set the group belongs to.\n"); + m->mothurOut("The sets parameter allows you to specify which of the sets in your designfile you would like to analyze. The set names are separated by dashes. THe default is all sets in the designfile. To run the pairwise comparisons of all sets use sets=all.\n"); + m->mothurOut("The iters parameter allows you to set number of randomization for the P value. The default is 1000. \n"); + m->mothurOut("The groups parameter allows you to specify which of the groups you would like included. The group names are separated by dashes. groups=all will find all pairwise comparisons. \n"); + m->mothurOut("The label parameter allows you to select what distance levels you would like, and are also separated by dashes.\n"); + m->mothurOut("The processors parameter allows you to specify how many processors you would like to use. The default is 1. \n"); + m->mothurOut("The homova command should be in the following format: homova(design=yourDesignFile).\n"); + m->mothurOut("Example amova(design=temp.design, groups=A-B-C).\n"); + m->mothurOut("Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n\n"); + + } + catch(exception& e) { + m->errorOut(e, "HomovaCommand", "help"); + exit(1); + } +} + +//********************************************************************************************************************** + +HomovaCommand::~HomovaCommand(){} + +//********************************************************************************************************************** + +int HomovaCommand::execute(){ + try { + + if (abort == true) { return 0; } + + //read design file + designMap = new GroupMap(designfile); + designMap->readDesignMap(); + + //make sure sets are all in designMap + SharedUtil* util = new SharedUtil(); + util->setGroups(Sets, designMap->namesOfGroups); + delete util; + + //if the user pickedGroups or set sets=all, then we want to figure out how to split the pairwise comparison between processors + int numGroups = Sets.size(); + if (pickedGroups) { + for (int a=0; a groups; groups.push_back(Sets[a]); groups.push_back(Sets[l]); + namesOfGroupCombos.push_back(groups); + } + } + }else { //one giant compare + vector groups; + for (int a=0; amothurOut("Not enough sets, I need at least 2 valid sets. Unable to complete command."); m->mothurOutEndLine(); m->control_pressed = true; } + +#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) + if(processors != 1){ + int numPairs = namesOfGroupCombos.size(); + int numPairsPerProcessor = numPairs / processors; + + for (int i = 0; i < processors; i++) { + int startPos = i * numPairsPerProcessor; + if(i == processors - 1){ + numPairsPerProcessor = numPairs - i * numPairsPerProcessor; + } + lines.push_back(linePair(startPos, numPairsPerProcessor)); + } + } +#endif + + if (sharedfile != "") { //create distance matrix for each label + + if (outputDir == "") { outputDir = m->hasPath(sharedfile); } + + //for each calculator + for(int i = 0 ; i < calculators.size(); i++) { + + //create a new filename + ofstream out; + string outputFile = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + calculators[i]->getName() + ".homova"; + m->openOutputFile(outputFile, out); + outputNames.push_back(outputFile); outputTypes["homova"].push_back(outputFile); + + //print headers + out << "label\tgroupsCompared\tBValue\tpValue" << endl; + out.close(); + } + + InputData input(sharedfile, "sharedfile"); + vector lookup = input.getSharedRAbundVectors(); + string lastLabel = lookup[0]->getLabel(); + + //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label. + set processedLabels; + set userLabels = labels; + + //sanity check between shared file groups and design file groups + for (int i = 0; i < lookup.size(); i++) { + string group = designMap->getGroup(lookup[i]->getGroup()); + + if (group == "not found") { m->control_pressed = true; m->mothurOut("[ERROR]: " + lookup[i]->getGroup() + " is not in your design file, please correct."); m->mothurOutEndLine(); } + } + + //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]; } globaldata->Groups.clear(); delete designMap; for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; } + + if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){ + + m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine(); + process(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(); + + process(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) { globaldata->Groups.clear(); delete designMap; for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; } + + //get next line to process + lookup = input.getSharedRAbundVectors(); + } + + if (m->control_pressed) { globaldata->Groups.clear(); delete designMap; for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } 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(); + + process(lookup); + + for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; } + } + + //reset groups parameter + globaldata->Groups.clear(); + + + }else { //user provided distance matrix + + if (outputDir == "") { outputDir = m->hasPath(phylipfile); } + + //create a new filename + ofstream out; + string outputFile = outputDir + m->getRootName(m->getSimpleName(phylipfile)) + "homova"; + m->openOutputFile(outputFile, out); + outputNames.push_back(outputFile); outputTypes["homova"].push_back(outputFile); + + //print headers + out << "groupsCompared\tBValue\tpValue" << endl; + out.close(); + + ReadPhylipVector readMatrix(phylipfile); + vector< vector > completeMatrix; + vector names = readMatrix.read(completeMatrix); + +#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) + if(processors == 1){ + driver(0, namesOfGroupCombos.size(), names, "", completeMatrix); + }else{ + int process = 1; + vector processIDS; + + //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(lines[process].start, lines[process].num, names, toString(getpid()), completeMatrix); + 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); + } + } + + //do my part + driver(lines[0].start, lines[0].num, names, "", completeMatrix); + + //force parent to wait until all the processes are done + for (int i=0;i<(processors-1);i++) { + int temp = processIDS[i]; + wait(&temp); + } + + //append files + string outputFile = outputDir + m->getRootName(m->getSimpleName(phylipfile)) + "homova"; + for (int j = 0; j < processIDS.size(); j++) { + m->appendFiles((outputFile + toString(processIDS[j])), outputFile); + remove((outputFile + toString(processIDS[j])).c_str()); + } + + } +#else + driver(0, namesOfGroupCombos.size(), names, "", completeMatrix); +#endif + + } + + delete designMap; + + if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; } + + 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, "HomovaCommand", "execute"); + exit(1); + } +} +//********************************************************************************************************************** + +int HomovaCommand::process(vector thisLookup) { + try{ + +#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) + if(processors == 1){ + driver(0, namesOfGroupCombos.size(), thisLookup, ""); + }else{ + int process = 1; + vector processIDS; + + //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(lines[process].start, lines[process].num, thisLookup, toString(getpid())); + 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); + } + } + + //do my part + driver(lines[0].start, lines[0].num, thisLookup, ""); + + //force parent to wait until all the processes are done + for (int i=0;i<(processors-1);i++) { + int temp = processIDS[i]; + wait(&temp); + } + + //append files + for(int i = 0 ; i < calculators.size(); i++) { + string outputFile = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + calculators[i]->getName() + ".homova"; + + for (int j = 0; j < processIDS.size(); j++) { + m->appendFiles((outputFile + toString(processIDS[j])), outputFile); + remove((outputFile + toString(processIDS[j])).c_str()); + } + } + } +#else + driver(0, namesOfGroupCombos.size(), thisLookUp, ""); +#endif + + return 0; + } + catch(exception& e) { + m->errorOut(e, "HomovaCommand", "process"); + exit(1); + } +} +//********************************************************************************************************************** + +int HomovaCommand::driver(int start, int num, vector thisLookup, string pidValue) { + try { + vector subset; + EstOutput data; + + //for each calculator + for(int i = 0 ; i < calculators.size(); i++) { + + //create a new filename + ofstream out; + string outputFile = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + calculators[i]->getName() + ".homova" + pidValue; + m->openOutputFileAppend(outputFile, out); + out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint); + + //for each combo + for (int c = start; c < (start+num); c++) { + + if (m->control_pressed) { out.close(); return 0; } + + //get set names + vector setNames; + for (int j = 0; j < namesOfGroupCombos[c].size(); j++) { setNames.push_back(namesOfGroupCombos[c][j]); } + + vector thisCombosLookup; + vector thisCombosLookupSets; //what set each sharedRabund is from to be used when calculating SSWithin + for (int k = 0; k < thisLookup.size(); k++) { + string thisGroup = thisLookup[k]->getGroup(); + + //is this group for a set we want to compare?? + if (m->inUsersGroups(designMap->getGroup(thisGroup), setNames)) { + thisCombosLookup.push_back(thisLookup[k]); + thisCombosLookupSets.push_back(designMap->getGroup(thisGroup)); + } + + } + + int numGroups = thisCombosLookup.size(); + + //calc the distance matrix + matrix.clear(); + matrix.resize(numGroups); + for (int k = 0; k < matrix.size(); k++) { for (int j = 0; j < matrix.size(); j++) { matrix[k].push_back(1.0); } } + + if (thisCombosLookup.size() == 0) { + m->mothurOut("[ERROR]: Missing shared info for sets. Skipping comparison."); m->mothurOutEndLine(); + }else{ + + out << thisLookup[0]->getLabel() << '\t'; + if (setNames.size() == 2) { out << setNames[0] << "-" << setNames[1] << '\t'; } + else { out << "all" << '\t'; } + + for (int k = 0; k < thisCombosLookup.size(); k++) { + for (int l = k; l < thisCombosLookup.size(); l++) { + + if (m->control_pressed) { out.close(); return 0; } + + if (k != l) { //we dont need to similiarity of a groups to itself + //get estimated similarity between 2 groups + subset.clear(); //clear out old pair of sharedrabunds + //add new pair of sharedrabunds + subset.push_back(thisCombosLookup[k]); subset.push_back(thisCombosLookup[l]); + + //if this calc needs all groups to calculate the pair load all groups + if (calculators[i]->getNeedsAll()) { + //load subset with rest of lookup for those calcs that need everyone to calc for a pair + for (int w = 0; w < thisCombosLookup.size(); w++) { + if ((w != k) && (w != l)) { subset.push_back(thisCombosLookup[w]); } + } + } + + data = calculators[i]->getValues(subset); //saves the calculator outputs + + //save values in similarity matrix + matrix[k][l] = 1.0 - data[0]; + matrix[l][k] = 1.0 - data[0]; + } + } + } + + //calc homova + calcHomova(out, setNames.size(), thisCombosLookupSets); + } + } + + out.close(); + } + + return 0; + + } + catch(exception& e) { + m->errorOut(e, "HomovaCommand", "driver"); + exit(1); + } +} +//********************************************************************************************************************** + +int HomovaCommand::driver(int start, int num, vector names, string pidValue, vector< vector >& completeMatrix) { + try { + + //create a new filename + ofstream out; + string outputFile = outputDir + m->getRootName(m->getSimpleName(phylipfile)) + "homova" + pidValue; + m->openOutputFileAppend(outputFile, out); + out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint); + + //for each combo + for (int c = start; c < (start+num); c++) { + + if (m->control_pressed) { out.close(); return 0; } + + //get set names + vector setNames; + for (int j = 0; j < namesOfGroupCombos[c].size(); j++) { setNames.push_back(namesOfGroupCombos[c][j]); } + + vector thisCombosSets; //what set each line in the distance matrix is from to be used when calculating SSWithin + set indexes; + for (int k = 0; k < names.size(); k++) { + //is this group for a set we want to compare?? + if (m->inUsersGroups(designMap->getGroup(names[k]), setNames)) { + thisCombosSets.push_back(designMap->getGroup(names[k])); + indexes.insert(k); //save indexes of valid rows in matrix for submatrix + } + } + + int numGroups = thisCombosSets.size(); + + //calc the distance matrix + matrix.clear(); + matrix.resize(numGroups); + + for (int k = 0; k < matrix.size(); k++) { for (int j = 0; j < matrix.size(); j++) { matrix[k].push_back(1.0); } } + + if (thisCombosSets.size() == 0) { + m->mothurOut("[ERROR]: Missing distance info for sets. Skipping comparison."); m->mothurOutEndLine(); + }else{ + + if (setNames.size() == 2) { out << setNames[0] << "-" << setNames[1] << '\t'; } + else { out << "all" << '\t'; } + + //fill submatrix + int rowCount = 0; + for (int j = 0; j < completeMatrix.size(); j++) { + + if (indexes.count(j) != 0) { //we want at least part of this row + int columnCount = 0; + for (int k = 0; k < completeMatrix[j].size(); k++) { + + if (indexes.count(k) != 0) { //we want this distance + matrix[rowCount][columnCount] = completeMatrix[j][k]; + matrix[columnCount][rowCount] = completeMatrix[j][k]; + columnCount++; + } + } + rowCount++; + } + } + + //calc homova + calcHomova(out, setNames.size(), thisCombosSets); + } + } + + out.close(); + + + return 0; + + } + catch(exception& e) { + m->errorOut(e, "HomovaCommand", "driver"); + exit(1); + } +} +//********************************************************************************************************************** +int HomovaCommand::calcHomova(ofstream& out, int numTreatments, vector thisCombosLookupSets) { + try { + + double SSTotal, BValue, pValue; + map SSWithin; + map::iterator it; + + //SSTotal = 0.0; + SSTotal = calcTotal(numTreatments); + cout << "sstotal = " << SSTotal << " (n-P) = " << (matrix.size() - numTreatments) << endl; + int numSamples = matrix.size(); + + //calc BValue + map counts; + SSWithin = calcWithin(matrix, numTreatments, thisCombosLookupSets, counts); + + double sum = 0.0; + double sumDenom = 0.0; + for (it = SSWithin.begin(); it != SSWithin.end(); it++) { + cout << it->first << '\t' << it->second << '\t' << (counts[it->first] - numTreatments) << endl; + double temp2 = (it->second) / (double) (counts[it->first] - 1); + cout << "sumTerm = " << temp2 << '\t' << "ln sumTerm = " << log(temp2) << endl; + sum += ((counts[it->first] - 1) * log(temp2)); + sumDenom += ((1 / (double) (counts[it->first] - 1)) - ( 1 / (double) (numSamples - numTreatments) )); + } + + double temp = SSTotal / (double) (numSamples - numTreatments); + double numeratorTerm1 = (numSamples - numTreatments) * log(temp); + double numerator = numeratorTerm1 - sum; + double denom = 1 + ((1 / (double) (3 * (numTreatments - 1))) * sumDenom); + + BValue = numerator / denom; + cout << "numeratorTerm1 = " << numeratorTerm1 << " sum = " << sum << " sumDenom = " << sumDenom << " numerator = " << numerator << " denom = " << denom << " B = " << BValue << endl; + + //calc Pvalue + int count = 0; + for (int i = 0; i < iters; i++) { + if (m->control_pressed) { break; } + + //randomly shuffle names to randomize the matrix + vector copyNames = thisCombosLookupSets; + random_shuffle(copyNames.begin(), copyNames.end()); + + counts.clear(); + map randomSSWithin = calcWithin(matrix, numTreatments, copyNames, counts); + + sum = 0.0; + sumDenom = 0.0; + for (it = randomSSWithin.begin(); it != randomSSWithin.end(); it++) { + double temp2 = (it->second) / (double) (counts[it->first] - numTreatments); + sum += ((counts[it->first] - 1) * log(temp2)); + sumDenom += ((1 / (double) (counts[it->first] - 1)) - ( 1 / (double) (numSamples - numTreatments) )); + } + + numerator = numeratorTerm1 - sum; + denom = 1 + ((1 / (double) (3 * (numTreatments - 1))) * sumDenom); + + double randomBValue = numerator / denom; + + if (randomBValue >= BValue) { count++; } + } + + pValue = count / (float) iters; + + out << BValue << '\t' << pValue << endl; + + return 0; + } + catch(exception& e) { + m->errorOut(e, "HomovaCommand", "calcHomova"); + exit(1); + } +} +//********************************************************************************************************************** +map HomovaCommand::calcWithin(vector< vector >& thisMatrix, int numTreatments, vector thisCombosLookupSets, map& count) { + try { + map within; //maps treatment to within value + map::iterator it; + map::iterator itCount; + + for (int i = 0; i < thisCombosLookupSets.size(); i++) { + itCount = count.find(thisCombosLookupSets[i]); + + if (itCount == count.end()) { //first time we have seen this treatment + count[thisCombosLookupSets[i]] = 1; + }else { + count[thisCombosLookupSets[i]]++; + } + + //initialize within + within[thisCombosLookupSets[i]] = 0.0; + } + + + //traverse lower triangle + for (int k = 0; k < thisMatrix.size(); k++) { + for (int l = k; l < thisMatrix[k].size(); l++) { + + //if you are from the same treatment then eij is 1 so add, else eij = 0 + if (thisCombosLookupSets[k] == thisCombosLookupSets[l]) { + within[thisCombosLookupSets[k]] += (thisMatrix[k][l] * thisMatrix[k][l]); //dij^2 + } + } + } + + //1 / (numSamples in this treatment) + for (it = within.begin(); it != within.end(); it++) { + (it->second) *= (1.0 / (float) count[it->first]); + } + + return within; + } + catch(exception& e) { + m->errorOut(e, "HomovaCommand", "calcWithin"); + exit(1); + } +} +//********************************************************************************************************************** +double HomovaCommand::calcTotal(int numTreatments) { + try { + double total = 0.0; + + //traverse lower triangle + for (int k = 0; k < matrix.size(); k++) { + for (int l = k; l < matrix[k].size(); l++) { + total += (matrix[k][l] * matrix[k][l]); //dij^2 + } + } + + //1 / numSamples + total *= (1.0 / (float) matrix.size()); + + return total; + } + catch(exception& e) { + m->errorOut(e, "HomovaCommand", "calcTotal"); + exit(1); + } +} +//********************************************************************************************************************** + + diff --git a/homovacommand.h b/homovacommand.h new file mode 100644 index 0000000..1ac2aa0 --- /dev/null +++ b/homovacommand.h @@ -0,0 +1,66 @@ +#ifndef HOMOVACOMMAND_H +#define HOMOVACOMMAND_H + +/* + * homovacommand.h + * mothur + * + * Created by westcott on 2/8/11. + * Copyright 2011 Schloss Lab. All rights reserved. + * + */ + + +#include "command.hpp" +#include "inputdata.h" +#include "sharedrabundvector.h" +#include "validcalculator.h" +#include "readphylipvector.h" + +class GlobalData; + +class HomovaCommand : public Command { + +public: + HomovaCommand(string); + HomovaCommand(); + ~HomovaCommand(); + vector getRequiredParameters(); + vector getValidParameters(); + vector getRequiredFiles(); + map > getOutputFiles() { return outputTypes; } + int execute(); + void help(); + +private: + struct linePair { + int start; + int num; + linePair(int i, int j) : start(i), num(j) {} + }; + vector lines; + + GlobalData* globaldata; + GroupMap* designMap; + map > outputTypes; + + vector< vector > matrix; + bool abort, allLines, pickedGroups; + set labels; //holds labels to be used + string format, groups, label, outputDir, inputDir, designfile, sets, phylipfile, calc, sharedfile; + vector Groups, outputNames, Sets; + vector< vector > namesOfGroupCombos; + int iters, processors; + vector calculators; + + int driver(int, int, vector, string); + int driver(int, int, vector, string, vector< vector >&); + int process(vector); + int calcHomova(ofstream&, int, vector); + map calcWithin(vector< vector >&, int, vector, map&); + double calcTotal(int); +}; + +#endif + + -- 2.39.2