pcoacommand.cpp

   1
   2 /*
   3  *  pcacommand.cpp
   4  *  Mothur
   5  *
   6  *  Created by westcott on 1/4/10.
   7  *  Copyright 2010 Schloss Lab. All rights reserved.
   8  *
   9  */
  10
  11 #include "pcoacommand.h"
  12 #include "readphylipvector.h"
  13
  14
  15 //**********************************************************************************************************************
  16 vector<string> PCOACommand::setParameters(){
  17         try {
  18                 CommandParameter pphylip("phylip", "InputTypes", "", "", "none", "none", "none","pcoa-loadings",false,true,true); parameters.push_back(pphylip);
  19                 CommandParameter pmetric("metric", "Boolean", "", "T", "", "", "","",false,false); parameters.push_back(pmetric);
  20                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
  21                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
  22
  23                 vector<string> myArray;
  24                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  25                 return myArray;
  26         }
  27         catch(exception& e) {
  28                 m->errorOut(e, "PCOACommand", "setParameters");
  29                 exit(1);
  30         }
  31 }
  32 //**********************************************************************************************************************
  33 string PCOACommand::getHelpString(){
  34         try {
  35                 string helpString = "";
  36                 helpString += "The pcoa command parameters are phylip and metric";
  37                 helpString += "The phylip parameter allows you to enter your distance file.";
  38                 helpString += "The metric parameter allows indicate you if would like the pearson correlation coefficient calculated. Default=True";
  39                 helpString += "Example pcoa(phylip=yourDistanceFile).\n";
  40                 helpString += "Note: No spaces between parameter labels (i.e. phylip), '=' and parameters (i.e.yourDistanceFile).\n";
  41                 return helpString;
  42         }
  43         catch(exception& e) {
  44                 m->errorOut(e, "PCOACommand", "getHelpString");
  45                 exit(1);
  46         }
  47 }
  48 //**********************************************************************************************************************
  49 string PCOACommand::getOutputPattern(string type) {
  50     try {
  51         string pattern = "";
  52
  53         if (type == "pcoa") {  pattern = "[filename],pcoa.axes"; }
  54         else if (type == "loadings") {  pattern = "[filename],pcoa.loadings"; }
  55         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  56
  57         return pattern;
  58     }
  59     catch(exception& e) {
  60         m->errorOut(e, "PCOACommand", "getOutputPattern");
  61         exit(1);
  62     }
  63 }
  64
  65
  66 //**********************************************************************************************************************
  67 PCOACommand::PCOACommand(){
  68         try {
  69                 abort = true; calledHelp = true;
  70                 setParameters();
  71                 vector<string> tempOutNames;
  72                 outputTypes["pcoa"] = tempOutNames;
  73                 outputTypes["loadings"] = tempOutNames;
  74         }
  75         catch(exception& e) {
  76                 m->errorOut(e, "PCOACommand", "PCOACommand");
  77                 exit(1);
  78         }
  79 }
  80 //**********************************************************************************************************************
  81
  82 PCOACommand::PCOACommand(string option)  {
  83         try {
  84                 abort = false; calledHelp = false;
  85
  86                 //allow user to run help
  87                 if(option == "help") { help(); abort = true; calledHelp = true; }
  88                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  89
  90                 else {
  91                         vector<string> myArray = setParameters();
  92
  93                         OptionParser parser(option);
  94                         map<string, string> parameters = parser. getParameters();
  95
  96                         ValidParameters validParameter;
  97                         map<string, string>::iterator it;
  98
  99                         //check to make sure all parameters are valid for command
 100                         for (it = parameters.begin(); it != parameters.end(); it++) {
 101                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 102                         }
 103                         //if the user changes the input directory command factory will send this info to us in the output parameter
 104                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 105                         if (inputDir == "not found"){   inputDir = "";          }
 106                         else {
 107                                 string path;
 108                                 it = parameters.find("phylip");
 109                                 //user has given a template file
 110                                 if(it != parameters.end()){
 111                                         path = m->hasPath(it->second);
 112                                         //if the user has not given a path then, add inputdir. else leave path alone.
 113                                         if (path == "") {       parameters["phylip"] = inputDir + it->second;           }
 114                                 }
 115                         }
 116
 117                         //initialize outputTypes
 118                         vector<string> tempOutNames;
 119                         outputTypes["pcoa"] = tempOutNames;
 120                         outputTypes["loadings"] = tempOutNames;
 121
 122                         //required parameters
 123                         phylipfile = validParameter.validFile(parameters, "phylip", true);
 124                         if (phylipfile == "not open") { abort = true; }
 125                         else if (phylipfile == "not found") {
 126                                 //if there is a current phylip file, use it
 127                                 phylipfile = m->getPhylipFile();
 128                                 if (phylipfile != "") { m->mothurOut("Using " + phylipfile + " as input file for the phylip parameter."); m->mothurOutEndLine(); }
 129                                 else {  m->mothurOut("You have no current phylip file and the phylip parameter is required."); m->mothurOutEndLine(); abort = true; }
 130                         }else { m->setPhylipFile(phylipfile); }
 131
 132                         filename = phylipfile;
 133
 134                         //if the user changes the output directory command factory will send this info to us in the output parameter
 135                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 136                                 outputDir = "";
 137                                 outputDir += m->hasPath(phylipfile); //if user entered a file with a path then preserve it
 138                         }
 139
 140                         string temp = validParameter.validFile(parameters, "metric", false);    if (temp == "not found"){       temp = "T";                             }
 141                         metric = m->isTrue(temp);
 142                 }
 143
 144         }
 145         catch(exception& e) {
 146                 m->errorOut(e, "PCOACommand", "PCOACommand");
 147                 exit(1);
 148         }
 149 }
 150 //**********************************************************************************************************************
 151 int PCOACommand::execute(){
 152         try {
 153
 154                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 155
 156                 cout.setf(ios::fixed, ios::floatfield);
 157                 cout.setf(ios::showpoint);
 158                 cerr.setf(ios::fixed, ios::floatfield);
 159                 cerr.setf(ios::showpoint);
 160
 161                 vector<string> names;
 162                 vector<vector<double> > D;
 163
 164                 fbase = outputDir + m->getRootName(m->getSimpleName(filename));
 165
 166                 ReadPhylipVector readFile(filename);
 167                 names = readFile.read(D);
 168
 169                 if (m->control_pressed) { return 0; }
 170
 171                 double offset = 0.0000;
 172                 vector<double> d;
 173                 vector<double> e;
 174                 vector<vector<double> > G = D;
 175                 //vector<vector<double> > copy_G;
 176
 177                 m->mothurOut("\nProcessing...\n");
 178
 179                 for(int count=0;count<2;count++){
 180                         linearCalc.recenter(offset, D, G);              if (m->control_pressed) { return 0; }
 181                         linearCalc.tred2(G, d, e);                              if (m->control_pressed) { return 0; }
 182                         linearCalc.qtli(d, e, G);                               if (m->control_pressed) { return 0; }
 183                         offset = d[d.size()-1];
 184                         if(offset > 0.0) break;
 185                 }
 186
 187                 if (m->control_pressed) { return 0; }
 188
 189                 output(fbase, names, G, d);
 190
 191                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);  } return 0; }
 192
 193                 if (metric) {
 194
 195                         for (int i = 1; i < 4; i++) {
 196
 197                                 vector< vector<double> > EuclidDists = linearCalc.calculateEuclidianDistance(G, i); //G is the pcoa file
 198
 199                                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);  } return 0; }
 200
 201                                 double corr = linearCalc.calcPearson(EuclidDists, D); //G is the pcoa file, D is the users distance matrix
 202
 203                                 m->mothurOut("Rsq " + toString(i) + " axis: " + toString(corr * corr)); m->mothurOutEndLine();
 204
 205                                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);  } return 0; }
 206                         }
 207                 }
 208
 209                 m->mothurOutEndLine();
 210                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 211                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 212                 m->mothurOutEndLine();
 213
 214                 return 0;
 215         }
 216         catch(exception& e) {
 217                 m->errorOut(e, "PCOACommand", "execute");
 218                 exit(1);
 219         }
 220 }
 221 /*********************************************************************************************************************************/
 222
 223 void PCOACommand::get_comment(istream& f, char begin, char end){
 224         try {
 225                 char d=f.get();
 226                 while(d != end){        d = f.get();    }
 227                 d = f.peek();
 228         }
 229         catch(exception& e) {
 230                 m->errorOut(e, "PCOACommand", "get_comment");
 231                 exit(1);
 232         }
 233 }
 234 /*********************************************************************************************************************************/
 235
 236 void PCOACommand::output(string fnameRoot, vector<string> name_list, vector<vector<double> >& G, vector<double> d) {
 237         try {
 238                 int rank = name_list.size();
 239                 double dsum = 0.0000;
 240                 for(int i=0;i<rank;i++){
 241                         dsum += d[i];
 242                         for(int j=0;j<rank;j++){
 243                                 if(d[j] >= 0)   {       G[i][j] *= pow(d[j],0.5);       }
 244                                 else                    {       G[i][j] = 0.00000;                      }
 245                         }
 246                 }
 247
 248                 ofstream pcaData;
 249         map<string, string> variables;
 250         variables["[filename]"] = fnameRoot;
 251         string pcoaDataFile = getOutputFileName("pcoa",variables);
 252         m->openOutputFile(pcoaDataFile, pcaData);
 253                 pcaData.setf(ios::fixed, ios::floatfield);
 254                 pcaData.setf(ios::showpoint);
 255                 outputNames.push_back(pcoaDataFile);
 256                 outputTypes["pcoa"].push_back(pcoaDataFile);
 257
 258                 ofstream pcaLoadings;
 259         string loadingsFile = getOutputFileName("loadings",variables);
 260         m->openOutputFile(loadingsFile, pcaLoadings);
 261                 pcaLoadings.setf(ios::fixed, ios::floatfield);
 262                 pcaLoadings.setf(ios::showpoint);
 263                 outputNames.push_back(loadingsFile);
 264                 outputTypes["loadings"].push_back(loadingsFile);
 265
 266                 pcaLoadings << "axis\tloading\n";
 267                 for(int i=0;i<rank;i++){
 268                         pcaLoadings << i+1 << '\t' << d[i] * 100.0 / dsum << endl;
 269                 }
 270
 271                 pcaData << "group";
 272                 for(int i=0;i<rank;i++){
 273                         pcaData << '\t' << "axis" << i+1;
 274                 }
 275                 pcaData << endl;
 276
 277                 for(int i=0;i<rank;i++){
 278                         pcaData << name_list[i] << '\t';
 279                         for(int j=0;j<rank;j++){
 280                                 pcaData << G[i][j] << '\t';
 281                         }
 282                         pcaData << endl;
 283                 }
 284         }
 285         catch(exception& e) {
 286                 m->errorOut(e, "PCOACommand", "output");
 287                 exit(1);
 288         }
 289 }
 290
 291 /*********************************************************************************************************************************/
 292