catchallcommand.cpp

   1 /*
   2  *  catchallcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 5/11/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "catchallcommand.h"
  11 #include "globaldata.hpp"
  12
  13 //**********************************************************************************************************************
  14 vector<string> CatchAllCommand::getValidParameters(){
  15         try {
  16                 string AlignArray[] =  {"sabund","label","inputdir","outputdir"};
  17                 vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  18                 return myArray;
  19         }
  20         catch(exception& e) {
  21                 m->errorOut(e, "CatchAllCommand", "getValidParameters");
  22                 exit(1);
  23         }
  24 }
  25 //**********************************************************************************************************************
  26 CatchAllCommand::CatchAllCommand(){
  27         try {
  28                 //initialize outputTypes
  29                 vector<string> tempOutNames;
  30                 outputTypes["csv"] = tempOutNames;
  31                 outputTypes["summary"] = tempOutNames;
  32         }
  33         catch(exception& e) {
  34                 m->errorOut(e, "CatchAllCommand", "CatchAllCommand");
  35                 exit(1);
  36         }
  37 }
  38 //**********************************************************************************************************************
  39 vector<string> CatchAllCommand::getRequiredParameters(){
  40         try {
  41                 string AlignArray[] =  {"sabund"};
  42                 vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  43                 return myArray;
  44         }
  45         catch(exception& e) {
  46                 m->errorOut(e, "CatchAllCommand", "getRequiredParameters");
  47                 exit(1);
  48         }
  49 }
  50 //**********************************************************************************************************************
  51 vector<string> CatchAllCommand::getRequiredFiles(){
  52         try {
  53                 vector<string> myArray;
  54                 return myArray;
  55         }
  56         catch(exception& e) {
  57                 m->errorOut(e, "CatchAllCommand", "getRequiredFiles");
  58                 exit(1);
  59         }
  60 }
  61 /**************************************************************************************/
  62 CatchAllCommand::CatchAllCommand(string option)  {
  63         try {
  64                 globaldata = GlobalData::getInstance();
  65                 abort = false;
  66                 allLines = 1;
  67
  68                 //allow user to run help
  69                 if(option == "help") { help(); abort = true; }
  70
  71                 else {
  72                         //valid paramters for this command
  73                         string Array[] =  {"sabund","label","inputdir","outputdir"};
  74                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  75
  76                         OptionParser parser(option);
  77                         map<string,string> parameters = parser.getParameters();
  78
  79                         ValidParameters validParameter;
  80                         map<string, string>::iterator it;
  81
  82                         //check to make sure all parameters are valid for command
  83                         for (it = parameters.begin(); it != parameters.end(); it++) {
  84                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  85                         }
  86
  87                         //initialize outputTypes
  88                         vector<string> tempOutNames;
  89                         outputTypes["csv"] = tempOutNames;
  90                         outputTypes["summary"] = tempOutNames;
  91
  92                         //if the user changes the input directory command factory will send this info to us in the output parameter
  93                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  94                         if (inputDir == "not found"){   inputDir = "";          }
  95                         else {
  96                                 string path;
  97                                 it = parameters.find("sabund");
  98                                 //user has given a template file
  99                                 if(it != parameters.end()){
 100                                         path = m->hasPath(it->second);
 101                                         //if the user has not given a path then, add inputdir. else leave path alone.
 102                                         if (path == "") {       parameters["sabund"] = inputDir + it->second;           }
 103                                 }
 104                         }
 105
 106                         //check for required parameters
 107                         sabundfile = validParameter.validFile(parameters, "sabund", true);
 108                         if (sabundfile == "not open") { sabundfile = ""; abort = true; }
 109                         else if (sabundfile == "not found") { sabundfile = "";  m->mothurOut("You must provide a sabund file for the catchall command."); m->mothurOutEndLine(); abort=true; }
 110                         else { globaldata->setSabundFile(sabundfile); globaldata->setFormat("sabund"); }
 111
 112                         string label = validParameter.validFile(parameters, "label", false);
 113                         if (label == "not found") { label = ""; }
 114                         else {
 115                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 116                                 else { allLines = 1;  }
 117                         }
 118
 119
 120                         //if the user changes the output directory command factory will send this info to us in the output parameter
 121                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(sabundfile);     }
 122                 }
 123
 124         }
 125         catch(exception& e) {
 126                 m->errorOut(e, "CatchAllCommand", "CatchAllCommand");
 127                 exit(1);
 128         }
 129 }
 130 //**********************************************************************************************************************
 131
 132 void CatchAllCommand::help(){
 133         try {
 134                 m->mothurOut("The catchall command interfaces mothur with the catchall program written by Linda Woodard, Sean Connolly and John Bunge.\n");
 135                 m->mothurOut("For more information about catchall refer to http://www.northeastern.edu/catchall/index.html \n");
 136                 m->mothurOut("The catchall executable must be in a folder called catchall in the same folder as your mothur executable, similar to mothur's requirements for using blast. \n");
 137                 m->mothurOut("If you are a MAC or Linux user you must also have installed mono, a link to mono is on the webpage. \n");
 138                 m->mothurOut("The catchall command parameters are sabund and label, sabund is required. \n");
 139                 m->mothurOut("The label parameter is used to analyze specific labels in your input.\n");
 140                 m->mothurOut("The catchall command should be in the following format: \n");
 141                 m->mothurOut("catchall(sabund=yourSabundFile) \n");
 142                 m->mothurOut("Example: catchall(sabund=abrecovery.fn.sabund) \n");
 143         }
 144         catch(exception& e) {
 145                 m->errorOut(e, "CatchAllCommand", "help");
 146                 exit(1);
 147         }
 148 }
 149
 150 /**************************************************************************************/
 151 int CatchAllCommand::execute() {
 152         try {
 153
 154                 if (abort == true) { return 0; }
 155
 156                 //prepare full output directory
 157                 outputDir = m->getFullPathName(outputDir);
 158
 159                 //get location of catchall
 160                 GlobalData* globaldata = GlobalData::getInstance();
 161                 path = globaldata->argv;
 162                 path = path.substr(0, (path.find_last_of('m')));
 163                 path = m->getFullPathName(path);
 164
 165                 string catchAllCommandExe = "";
 166                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 167                         catchAllCommandExe += "mono " + path + "catchall/CatchAllcmdL.exe ";
 168                 #else
 169                         catchAllCommandExe += path + "catchall/CatchAllcmdW.exe ";
 170                 #endif
 171
 172                 read = new ReadOTUFile(sabundfile);
 173                 read->read(&*globaldata);
 174
 175                 SAbundVector* sabund = globaldata->sabund;
 176                 string lastLabel = sabund->getLabel();
 177                 input = globaldata->ginput;
 178
 179                 set<string> processedLabels;
 180                 set<string> userLabels = labels;
 181
 182                 string summaryfilename = outputDir + m->getRootName(m->getSimpleName(sabundfile)) + "catchall.summary";
 183                 summaryfilename = m->getFullPathName(summaryfilename);
 184                 outputNames.push_back(summaryfilename); outputTypes["summary"].push_back(summaryfilename);
 185
 186                 ofstream out;
 187                 m->openOutputFile(summaryfilename, out);
 188
 189                 out << "label\tmodel\testimate\tlci\tuci" << endl;
 190
 191                 //for each label the user selected
 192                 while((sabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 193
 194
 195                         if(allLines == 1 || labels.count(sabund->getLabel()) == 1){
 196                                         m->mothurOut(sabund->getLabel());  m->mothurOutEndLine();
 197
 198                                         //create catchall input file from mothur's inputfile
 199                                         string filename = process(sabund);
 200                                         string outputPath = m->getPathName(filename);
 201
 202                                         //create system command
 203                                         string catchAllCommand = catchAllCommandExe + filename + " " + outputPath + " 1";
 204
 205                                         //run catchall
 206                                         system(catchAllCommand.c_str());
 207
 208                                         remove(filename.c_str());
 209
 210                                         filename = m->getRootName(filename); filename = filename.substr(0, filename.length()-1); //rip off extra .
 211
 212                                         outputNames.push_back(filename + "_Analysis.csv"); outputTypes["csv"].push_back(filename + "_Analysis.csv");
 213                                         outputNames.push_back(filename + "_BestModelsAnalysis.csv"); outputTypes["csv"].push_back(filename + "_BestModelsAnalysis.csv");
 214                                         outputNames.push_back(filename + "_BestModelsFits.csv"); outputTypes["csv"].push_back(filename + "_BestModelsFits.csv");
 215                                         outputNames.push_back(filename + "_BubblePlot.csv"); outputTypes["csv"].push_back(filename + "_BubblePlot.csv");
 216
 217                                         createSummaryFile(filename + "_BestModelsAnalysis.csv", sabund->getLabel(), out);
 218
 219                                         if (m->control_pressed) { out.close(); for (int i = 0; i < outputNames.size(); i++) {remove(outputNames[i].c_str());    } delete read;  delete input; globaldata->ginput = NULL; delete sabund;  return 0; }
 220
 221                                         processedLabels.insert(sabund->getLabel());
 222                                         userLabels.erase(sabund->getLabel());
 223                         }
 224
 225                         if ((m->anyLabelsToProcess(sabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 226                                         string saveLabel = sabund->getLabel();
 227
 228                                         delete sabund;
 229                                         sabund = (input->getSAbundVector(lastLabel));
 230
 231                                         m->mothurOut(sabund->getLabel());  m->mothurOutEndLine();
 232
 233
 234                                         //create catchall input file from mothur's inputfile
 235                                         string filename = process(sabund);
 236                                         string outputPath = m->getPathName(filename);
 237
 238                                         //create system command
 239                                         string catchAllCommand = catchAllCommandExe + filename + " " + outputPath + " 1";
 240
 241                                         //run catchall
 242                                         system(catchAllCommand.c_str());
 243
 244                                         remove(filename.c_str());
 245
 246                                         filename = m->getRootName(filename); filename = filename.substr(0, filename.length()-1); //rip off extra .
 247
 248                                         outputNames.push_back(filename + "_Analysis.csv"); outputTypes["csv"].push_back(filename + "_Analysis.csv");
 249                                         outputNames.push_back(filename + "_BestModelsAnalysis.csv"); outputTypes["csv"].push_back(filename + "_BestModelsAnalysis.csv");
 250                                         outputNames.push_back(filename + "_BestModelsFits.csv"); outputTypes["csv"].push_back(filename + "_BestModelsFits.csv");
 251                                         outputNames.push_back(filename + "_BubblePlot.csv"); outputTypes["csv"].push_back(filename + "_BubblePlot.csv");
 252
 253                                         createSummaryFile(filename + "_BestModelsAnalysis.csv", sabund->getLabel(), out);
 254
 255                                         if (m->control_pressed) { out.close(); for (int i = 0; i < outputNames.size(); i++) {remove(outputNames[i].c_str());    } delete read;  delete input; globaldata->ginput = NULL; delete sabund;  return 0; }
 256
 257                                         processedLabels.insert(sabund->getLabel());
 258                                         userLabels.erase(sabund->getLabel());
 259
 260                                         //restore real lastlabel to save below
 261                                         sabund->setLabel(saveLabel);
 262                         }
 263
 264
 265                         lastLabel = sabund->getLabel();
 266
 267                         delete sabund;
 268                         sabund = (input->getSAbundVector());
 269                 }
 270
 271                 //output error messages about any remaining user labels
 272                 set<string>::iterator it;
 273                 bool needToRun = false;
 274                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 275                         m->mothurOut("Your file does not include the label " + *it);
 276                         if (processedLabels.count(lastLabel) != 1) {
 277                                 m->mothurOut(". I will use " + lastLabel + ".");  m->mothurOutEndLine();
 278                                 needToRun = true;
 279                         }else {
 280                                 m->mothurOut(". Please refer to " + lastLabel + ".");  m->mothurOutEndLine();
 281                         }
 282                 }
 283
 284                 //run last label if you need to
 285                 if (needToRun == true)  {
 286                         if (sabund != NULL) {   delete sabund;  }
 287                         sabund = (input->getSAbundVector(lastLabel));
 288
 289                         m->mothurOut(sabund->getLabel());  m->mothurOutEndLine();
 290
 291                         //create catchall input file from mothur's inputfile
 292                         string filename = process(sabund);
 293                         string outputPath = m->getPathName(filename);
 294
 295                         //create system command
 296                         string catchAllCommand = catchAllCommandExe + filename + " " + outputPath + " 1";
 297
 298                         //run catchall
 299                         system(catchAllCommand.c_str());
 300
 301                         remove(filename.c_str());
 302
 303                         filename = m->getRootName(filename); filename = filename.substr(0, filename.length()-1); //rip off extra .
 304
 305                         outputNames.push_back(filename + "_Analysis.csv"); outputTypes["csv"].push_back(filename + "_Analysis.csv");
 306                         outputNames.push_back(filename + "_BestModelsAnalysis.csv"); outputTypes["csv"].push_back(filename + "_BestModelsAnalysis.csv");
 307                         outputNames.push_back(filename + "_BestModelsFits.csv"); outputTypes["csv"].push_back(filename + "_BestModelsFits.csv");
 308                         outputNames.push_back(filename + "_BubblePlot.csv"); outputTypes["csv"].push_back(filename + "_BubblePlot.csv");
 309
 310                         createSummaryFile(filename + "_BestModelsAnalysis.csv", sabund->getLabel(), out);
 311
 312                         delete sabund;
 313                 }
 314
 315                 out.close();
 316                 delete read;
 317                 delete input; globaldata->ginput = NULL;
 318
 319                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {remove(outputNames[i].c_str()); } return 0; }
 320
 321                 m->mothurOutEndLine();
 322                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 323                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 324                 m->mothurOutEndLine();
 325
 326
 327                 return 0;
 328         }
 329         catch(exception& e) {
 330                 m->errorOut(e, "CatchAllCommand", "execute");
 331                 exit(1);
 332         }
 333 }
 334 //**********************************************************************************************************************
 335 string CatchAllCommand::process(SAbundVector* sabund) {
 336         try {
 337                 string filename = outputDir + m->getRootName(m->getSimpleName(sabundfile)) + sabund->getLabel() + ".csv";
 338                 filename = m->getFullPathName(filename);
 339
 340                 ofstream out;
 341                 m->openOutputFile(filename, out);
 342
 343                 for (int i = 1; i <= sabund->getMaxRank(); i++) {
 344                         int temp = sabund->get(i);
 345
 346                         if (temp != 0) {
 347                                 out << i << "," << temp << endl;
 348                         }
 349                 }
 350                 out.close();
 351
 352                 return filename;
 353
 354         }
 355         catch(exception& e) {
 356                 m->errorOut(e, "CatchAllCommand", "process");
 357                 exit(1);
 358         }
 359 }
 360 //**********************************************************************************************************************
 361 int CatchAllCommand::createSummaryFile(string file1, string label, ofstream& out) {
 362         try {
 363
 364                 ifstream in;
 365                 m->openInputFile(file1, in);
 366
 367                 if (!in.eof()) {
 368
 369                         string header = m->getline(in); m->gobble(in);
 370
 371                         int pos = header.find("Total Number of Observed Species =");
 372                         string numString = "";
 373
 374
 375                         if (pos == string::npos) { m->mothurOut("[ERROR]: cannot parse " + file1); m->mothurOutEndLine(); }
 376                         else {
 377                                 //pos will be the position of the T in total, so we want to count to the position of =
 378                                 pos += 34;
 379                                 char c=header[pos];
 380                                 while (c != ','){
 381                                         if (c != ' ') {
 382                                                 numString += c;
 383                                         }
 384                                         pos++;
 385                                         c=header[pos];
 386
 387                                         //sanity check
 388                                         if (pos > header.length()) { m->mothurOut("Cannot find number of OTUs in " + file1); m->mothurOutEndLine(); in.close(); return 0; }
 389                                 }
 390                         }
 391
 392                         string firstline = m->getline(in); m->gobble(in);
 393                         vector<string> values;
 394                         m->splitAtComma(firstline, values);
 395
 396                         values.pop_back(); //last value is always a blank string since the last character in the line is always a ','
 397
 398                         if (values.size() == 1) { //grab next line if firstline didn't have what you wanted
 399                                 string secondline = m->getline(in); m->gobble(in);
 400                                 values.clear();
 401                                 m->splitAtComma(secondline, values);
 402
 403                                 values.pop_back(); //last value is always a blank string since the last character in the line is always a ','
 404                         }
 405
 406                         if (values.size() == 1) { //still not what we wanted fill values with numOTUs
 407                                 values.resize(8, "");
 408                                 values[1] = "Sobs";
 409                                 values[4] = numString;
 410                                 values[6] = numString;
 411                                 values[7] = numString;
 412                         }
 413
 414                         if (values.size() < 8) { values.resize(8, ""); }
 415
 416                         out << label << '\t' << values[1] << '\t' << values[4] << '\t' << values[6] << '\t' << values[7] << endl;
 417                 }
 418
 419                 in.close();
 420
 421                 return 0;
 422
 423         }
 424         catch(exception& e) {
 425                 m->errorOut(e, "CatchAllCommand", "createSummaryFile");
 426                 exit(1);
 427         }
 428 }
 429 /**************************************************************************************/
 430
 431
 432