libshuffcommand.cpp

   1 /*
   2  *  libshuffcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 3/9/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 /* This class is designed to implement an integral form of the Cramer-von Mises statistic.
  11         you may refer to the "Integration of Microbial Ecology and Statistics: A Test To Compare Gene Libraries"
  12         paper in Applied and Environmental Microbiology, Sept. 2004, p. 5485-5492 0099-2240/04/$8.00+0
  13         DOI: 10.1128/AEM.70.9.5485-5492.2004 Copyright 2004 American Society for Microbiology for more information. */
  14
  15
  16 #include "libshuffcommand.h"
  17 #include "libshuff.h"
  18 #include "slibshuff.h"
  19 #include "dlibshuff.h"
  20
  21 //**********************************************************************************************************************
  22 vector<string> LibShuffCommand::getValidParameters(){
  23         try {
  24                 string Array[] =  {"iters","groups","step","form","cutoff","outputdir","inputdir"};
  25                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  26                 return myArray;
  27         }
  28         catch(exception& e) {
  29                 m->errorOut(e, "LibShuffCommand", "getValidParameters");
  30                 exit(1);
  31         }
  32 }
  33 //**********************************************************************************************************************
  34 LibShuffCommand::LibShuffCommand(){
  35         try {
  36                 //initialize outputTypes
  37                 vector<string> tempOutNames;
  38                 outputTypes["coverage"] = tempOutNames;
  39                 outputTypes["libshuffsummary"] = tempOutNames;
  40         }
  41         catch(exception& e) {
  42                 m->errorOut(e, "LibShuffCommand", "LibShuffCommand");
  43                 exit(1);
  44         }
  45 }
  46 //**********************************************************************************************************************
  47 vector<string> LibShuffCommand::getRequiredParameters(){
  48         try {
  49                 vector<string> myArray;
  50                 return myArray;
  51         }
  52         catch(exception& e) {
  53                 m->errorOut(e, "LibShuffCommand", "getRequiredParameters");
  54                 exit(1);
  55         }
  56 }
  57 //**********************************************************************************************************************
  58 vector<string> LibShuffCommand::getRequiredFiles(){
  59         try {
  60                 string Array[] =  {"phylip","group"};
  61                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  62                 return myArray;
  63         }
  64         catch(exception& e) {
  65                 m->errorOut(e, "LibShuffCommand", "getRequiredFiles");
  66                 exit(1);
  67         }
  68 }
  69 //**********************************************************************************************************************
  70
  71 LibShuffCommand::LibShuffCommand(string option)  {
  72         try {
  73                 globaldata = GlobalData::getInstance();
  74                 abort = false;
  75                 Groups.clear();
  76
  77                 //allow user to run help
  78                 if(option == "help") { help(); abort = true; }
  79
  80                 else {
  81                         //valid paramters for this command
  82                         string Array[] =  {"iters","groups","step","form","cutoff","outputdir","inputdir"};
  83                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  84
  85                         OptionParser parser(option);
  86                         map<string, string> parameters = parser.getParameters();
  87
  88                         ValidParameters validParameter;
  89
  90                         //check to make sure all parameters are valid for command
  91                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  92                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  93                         }
  94
  95                         //initialize outputTypes
  96                         vector<string> tempOutNames;
  97                         outputTypes["coverage"] = tempOutNames;
  98                         outputTypes["libshuffsummary"] = tempOutNames;
  99
 100                         //if the user changes the output directory command factory will send this info to us in the output parameter
 101                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 102                                 outputDir = "";
 103                                 outputDir += m->hasPath(globaldata->getPhylipFile()); //if user entered a file with a path then preserve it
 104                         }
 105
 106                         //make sure the user has already run the read.dist command
 107                         if ((globaldata->gMatrix == NULL) || (globaldata->gGroupmap == NULL)) {
 108                                 m->mothurOut("You must read in a matrix and groupfile using the read.dist command, before you use the libshuff command. "); m->mothurOutEndLine(); abort = true;;
 109                         }
 110
 111                         //check for optional parameter and set defaults
 112                         // ...at some point should added some additional type checking...
 113                         groups = validParameter.validFile(parameters, "groups", false);
 114                         if (groups == "not found") { groups = ""; savegroups = groups; }
 115                         else {
 116                                 savegroups = groups;
 117                                 m->splitAtDash(groups, Groups);
 118                                 globaldata->Groups = Groups;
 119                         }
 120
 121                         string temp;
 122                         temp = validParameter.validFile(parameters, "iters", false);                            if (temp == "not found") { temp = "10000"; }
 123                         convert(temp, iters);
 124
 125                         temp = validParameter.validFile(parameters, "cutoff", false);                           if (temp == "not found") { temp = "1.0"; }
 126                         convert(temp, cutOff);
 127
 128                         temp = validParameter.validFile(parameters, "step", false);                             if (temp == "not found") { temp = "0.01"; }
 129                         convert(temp, step);
 130
 131                         userform = validParameter.validFile(parameters, "form", false);                 if (userform == "not found") { userform = "integral"; }
 132
 133                         if (abort == false) {
 134
 135                                 matrix = globaldata->gMatrix;                           //get the distance matrix
 136                                 setGroups();                                                            //set the groups to be analyzed and sorts them
 137
 138                                 /********************************************************************************************/
 139                                 //this is needed because when we read the matrix we sort it into groups in alphabetical order
 140                                 //the rest of the command and the classes used in this command assume specific order
 141                                 /********************************************************************************************/
 142                                 matrix->setGroups(globaldata->gGroupmap->namesOfGroups);
 143                                 vector<int> sizes;
 144                                 for (int i = 0; i < globaldata->gGroupmap->namesOfGroups.size(); i++) {   sizes.push_back(globaldata->gGroupmap->getNumSeqs(globaldata->gGroupmap->namesOfGroups[i]));  }
 145                                 matrix->setSizes(sizes);
 146
 147
 148                                 if(userform == "discrete"){
 149                                         form = new DLibshuff(matrix, iters, step, cutOff);
 150                                 }
 151                                 else{
 152                                         form = new SLibshuff(matrix, iters, cutOff);
 153                                 }
 154                         }
 155
 156                 }
 157
 158         }
 159         catch(exception& e) {
 160                 m->errorOut(e, "LibShuffCommand", "LibShuffCommand");
 161                 exit(1);
 162         }
 163 }
 164 //**********************************************************************************************************************
 165
 166 void LibShuffCommand::help(){
 167         try {
 168                 m->mothurOut("The libshuff command can only be executed after a successful read.dist command including a groupfile.\n");
 169                 m->mothurOut("The libshuff command parameters are groups, iters, step, form and cutoff.  No parameters are required.\n");
 170                 m->mothurOut("The groups parameter allows you to specify which of the groups in your groupfile you would like analyzed.  You must enter at least 2 valid groups.\n");
 171                 m->mothurOut("The group names are separated by dashes.  The iters parameter allows you to specify how many random matrices you would like compared to your matrix.\n");
 172                 m->mothurOut("The step parameter allows you to specify change in distance you would like between each output if you are using the discrete form.\n");
 173                 m->mothurOut("The form parameter allows you to specify if you would like to analyze your matrix using the discrete or integral form. Your options are integral or discrete.\n");
 174                 m->mothurOut("The libshuff command should be in the following format: libshuff(groups=yourGroups, iters=yourIters, cutOff=yourCutOff, form=yourForm, step=yourStep).\n");
 175                 m->mothurOut("Example libshuff(groups=A-B-C, iters=500, form=discrete, step=0.01, cutOff=2.0).\n");
 176                 m->mothurOut("The default value for groups is all the groups in your groupfile, iters is 10000, cutoff is 1.0, form is integral and step is 0.01.\n");
 177                 m->mothurOut("The libshuff command output two files: .coverage and .slsummary their descriptions are in the manual.\n");
 178                 m->mothurOut("Note: No spaces between parameter labels (i.e. iters), '=' and parameters (i.e.yourIters).\n\n");
 179         }
 180         catch(exception& e) {
 181                 m->errorOut(e, "LibShuffCommand", "help");
 182                 exit(1);
 183         }
 184 }
 185
 186 //**********************************************************************************************************************
 187
 188 int LibShuffCommand::execute(){
 189         try {
 190
 191                 if (abort == true) {    return 0;       }
 192
 193                 savedDXYValues = form->evaluateAll();
 194                 savedMinValues = form->getSavedMins();
 195
 196                 if (m->control_pressed) {  delete form; globaldata->Groups.clear(); delete globaldata->gMatrix;  globaldata->gMatrix = NULL; return 0; }
 197
 198                 pValueCounts.resize(numGroups);
 199                 for(int i=0;i<numGroups;i++){
 200                         pValueCounts[i].assign(numGroups, 0);
 201                 }
 202
 203                 if (m->control_pressed) {  outputTypes.clear(); delete form; globaldata->Groups.clear(); delete globaldata->gMatrix;  globaldata->gMatrix = NULL; return 0; }
 204
 205                 Progress* reading = new Progress();
 206
 207                 for(int i=0;i<numGroups-1;i++) {
 208                         for(int j=i+1;j<numGroups;j++) {
 209
 210                                 if (m->control_pressed) {  outputTypes.clear();  delete form; globaldata->Groups.clear(); delete globaldata->gMatrix;  globaldata->gMatrix = NULL; delete reading; return 0; }
 211
 212                                 reading->newLine(groupNames[i]+'-'+groupNames[j], iters);
 213                                 int spoti = globaldata->gGroupmap->groupIndex[groupNames[i]]; //neccessary in case user selects groups so you know where they are in the matrix
 214                                 int spotj = globaldata->gGroupmap->groupIndex[groupNames[j]];
 215
 216                                 for(int p=0;p<iters;p++) {
 217
 218                                         if (m->control_pressed) {  outputTypes.clear(); delete form; globaldata->Groups.clear(); delete globaldata->gMatrix;  globaldata->gMatrix = NULL; delete reading; return 0; }
 219
 220                                         form->randomizeGroups(spoti,spotj);
 221                                         if(form->evaluatePair(spoti,spotj) >= savedDXYValues[spoti][spotj])     {       pValueCounts[i][j]++;   }
 222                                         if(form->evaluatePair(spotj,spoti) >= savedDXYValues[spotj][spoti])     {       pValueCounts[j][i]++;   }
 223
 224                                         if (m->control_pressed) {  outputTypes.clear(); delete form; globaldata->Groups.clear(); delete globaldata->gMatrix;  globaldata->gMatrix = NULL; delete reading; return 0; }
 225
 226                                         reading->update(p);
 227                                 }
 228                                 form->resetGroup(spoti);
 229                                 form->resetGroup(spotj);
 230                         }
 231                 }
 232
 233                 if (m->control_pressed) { outputTypes.clear();  delete form; globaldata->Groups.clear(); delete globaldata->gMatrix;  globaldata->gMatrix = NULL; delete reading; return 0; }
 234
 235                 reading->finish();
 236                 delete reading;
 237
 238                 m->mothurOutEndLine();
 239                 printSummaryFile();
 240                 printCoverageFile();
 241
 242                 //clear out users groups
 243                 globaldata->Groups.clear();
 244                 delete form;
 245
 246                 //delete globaldata's copy of the gmatrix to free up memory
 247                 delete globaldata->gMatrix;  globaldata->gMatrix = NULL;
 248
 249                 if (m->control_pressed) {  outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); } return 0; }
 250
 251
 252                 m->mothurOutEndLine();
 253                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 254                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 255                 m->mothurOutEndLine();
 256
 257                 return 0;
 258         }
 259         catch(exception& e) {
 260                 m->errorOut(e, "LibShuffCommand", "execute");
 261                 exit(1);
 262         }
 263 }
 264
 265 //**********************************************************************************************************************
 266
 267 int LibShuffCommand::printCoverageFile() {
 268         try {
 269
 270                 ofstream outCov;
 271                 summaryFile = outputDir + m->getRootName(m->getSimpleName(globaldata->getPhylipFile())) + "libshuff.coverage";
 272                 m->openOutputFile(summaryFile, outCov);
 273                 outputNames.push_back(summaryFile); outputTypes["coverage"].push_back(summaryFile);
 274                 outCov.setf(ios::fixed, ios::floatfield); outCov.setf(ios::showpoint);
 275                 //cout.setf(ios::fixed, ios::floatfield); cout.setf(ios::showpoint);
 276
 277                 map<double,vector<int> > allDistances;
 278                 map<double,vector<int> >::iterator it;
 279
 280                 vector<vector<int> > indices(numGroups);
 281                 int numIndices = numGroups * numGroups;
 282
 283                 int index = 0;
 284                 for(int i=0;i<numGroups;i++){
 285                         indices[i].assign(numGroups,0);
 286                         for(int j=0;j<numGroups;j++){
 287                                 indices[i][j] = index++;
 288
 289                                 int spoti = globaldata->gGroupmap->groupIndex[groupNames[i]]; //neccessary in case user selects groups so you know where they are in the matrix
 290                                 int spotj = globaldata->gGroupmap->groupIndex[groupNames[j]];
 291
 292                                 for(int k=0;k<savedMinValues[spoti][spotj].size();k++){
 293
 294                                         if(m->control_pressed)  { outCov.close(); return 0; }
 295
 296                                         if(allDistances[savedMinValues[spoti][spotj][k]].size() != 0){
 297                                                 allDistances[savedMinValues[spoti][spotj][k]][indices[i][j]]++;
 298                                         }
 299                                         else{
 300                                                 allDistances[savedMinValues[spoti][spotj][k]].assign(numIndices, 0);
 301                                                 allDistances[savedMinValues[spoti][spotj][k]][indices[i][j]] = 1;
 302                                         }
 303                                 }
 304                         }
 305                 }
 306                 it=allDistances.begin();
 307
 308                 //cout << setprecision(8);
 309
 310                 vector<int> prevRow = it->second;
 311                 it++;
 312
 313                 for(;it!=allDistances.end();it++){
 314                         for(int i=0;i<it->second.size();i++){
 315                                 it->second[i] += prevRow[i];
 316                         }
 317                         prevRow = it->second;
 318                 }
 319
 320                 vector<int> lastRow = allDistances.rbegin()->second;
 321                 outCov << setprecision(8);
 322
 323                 outCov << "dist";
 324                 for (int i = 0; i < numGroups; i++){
 325                         outCov << '\t' << groupNames[i];
 326                 }
 327                 for (int i=0;i<numGroups;i++){
 328                         for(int j=i+1;j<numGroups;j++){
 329                                 if(m->control_pressed)  { outCov.close(); return 0; }
 330                                 outCov << '\t' << groupNames[i] << '-' << groupNames[j] << '\t';
 331                                 outCov << groupNames[j] << '-' << groupNames[i];
 332                         }
 333                 }
 334                 outCov << endl;
 335
 336                 for(it=allDistances.begin();it!=allDistances.end();it++){
 337                         outCov << it->first << '\t';
 338                         for(int i=0;i<numGroups;i++){
 339                                 outCov << it->second[indices[i][i]]/(float)lastRow[indices[i][i]] << '\t';
 340                         }
 341                         for(int i=0;i<numGroups;i++){
 342                                 for(int j=i+1;j<numGroups;j++){
 343                                         if(m->control_pressed)  { outCov.close(); return 0; }
 344
 345                                         outCov << it->second[indices[i][j]]/(float)lastRow[indices[i][j]] << '\t';
 346                                         outCov << it->second[indices[j][i]]/(float)lastRow[indices[j][i]] << '\t';
 347                                 }
 348                         }
 349                         outCov << endl;
 350                 }
 351                 outCov.close();
 352
 353                 return 0;
 354         }
 355         catch(exception& e) {
 356                 m->errorOut(e, "LibShuffCommand", "printCoverageFile");
 357                 exit(1);
 358         }
 359 }
 360
 361 //**********************************************************************************************************************
 362
 363 int LibShuffCommand::printSummaryFile() {
 364         try {
 365
 366                 ofstream outSum;
 367                 summaryFile = outputDir + m->getRootName(m->getSimpleName(globaldata->getPhylipFile())) + "libshuff.summary";
 368                 m->openOutputFile(summaryFile, outSum);
 369                 outputNames.push_back(summaryFile); outputTypes["libshuffsummary"].push_back(summaryFile);
 370
 371                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 372                 cout.setf(ios::fixed, ios::floatfield); cout.setf(ios::showpoint);
 373
 374                 cout << setw(20) << left << "Comparison" << '\t' << setprecision(8) << "dCXYScore" << '\t' << "Significance" << endl;
 375                 m->mothurOutJustToLog("Comparison\tdCXYScore\tSignificance"); m->mothurOutEndLine();
 376                 outSum << setw(20) << left << "Comparison" << '\t' << setprecision(8) << "dCXYScore" << '\t' << "Significance" << endl;
 377
 378                 int precision = (int)log10(iters);
 379                 for(int i=0;i<numGroups;i++){
 380                         for(int j=i+1;j<numGroups;j++){
 381                                 if(m->control_pressed)  { outSum.close(); return 0; }
 382
 383                                 int spoti = globaldata->gGroupmap->groupIndex[groupNames[i]]; //neccessary in case user selects groups so you know where they are in the matrix
 384                                 int spotj = globaldata->gGroupmap->groupIndex[groupNames[j]];
 385
 386                                 if(pValueCounts[i][j]){
 387                                         cout << setw(20) << left << groupNames[i]+'-'+groupNames[j] << '\t' << setprecision(8) << savedDXYValues[spoti][spotj] << '\t' << setprecision(precision) << pValueCounts[i][j]/(float)iters << endl;
 388                                         m->mothurOutJustToLog(groupNames[i]+"-"+groupNames[j] + "\t" + toString(savedDXYValues[spoti][spotj]) + "\t" + toString((pValueCounts[i][j]/(float)iters))); m->mothurOutEndLine();
 389                                         outSum << setw(20) << left << groupNames[i]+'-'+groupNames[j] << '\t' << setprecision(8) << savedDXYValues[spoti][spotj] << '\t' << setprecision(precision) << pValueCounts[i][j]/(float)iters << endl;
 390                                 }
 391                                 else{
 392                                         cout << setw(20) << left << groupNames[i]+'-'+groupNames[j] << '\t' << setprecision(8) << savedDXYValues[spoti][spotj] << '\t' << '<' <<setprecision(precision) << 1/(float)iters << endl;
 393                                         m->mothurOutJustToLog(groupNames[i]+"-"+groupNames[j] + "\t" + toString(savedDXYValues[spoti][spotj]) + "\t" + toString((1/(float)iters))); m->mothurOutEndLine();
 394                                         outSum << setw(20) << left << groupNames[i]+'-'+groupNames[j] << '\t' << setprecision(8) << savedDXYValues[spoti][spotj] << '\t' << '<' <<setprecision(precision) << 1/(float)iters << endl;
 395                                 }
 396                                 if(pValueCounts[j][i]){
 397                                         cout << setw(20) << left << groupNames[j]+'-'+groupNames[i] << '\t' << setprecision(8) << savedDXYValues[spotj][spoti] << '\t' << setprecision (precision) << pValueCounts[j][i]/(float)iters << endl;
 398                                         m->mothurOutJustToLog(groupNames[j]+"-"+groupNames[i] + "\t" + toString(savedDXYValues[spotj][spoti]) + "\t" + toString((pValueCounts[j][i]/(float)iters))); m->mothurOutEndLine();
 399                                         outSum << setw(20) << left << groupNames[j]+'-'+groupNames[i] << '\t' << setprecision(8) << savedDXYValues[spotj][spoti] << '\t' << setprecision (precision) << pValueCounts[j][i]/(float)iters << endl;
 400                                 }
 401                                 else{
 402                                         cout << setw(20) << left << groupNames[j]+'-'+groupNames[i] << '\t' << setprecision(8) << savedDXYValues[spotj][spoti] << '\t' << '<' <<setprecision (precision) << 1/(float)iters << endl;
 403                                         m->mothurOutJustToLog(groupNames[j]+"-"+groupNames[i] + "\t" + toString(savedDXYValues[spotj][spoti]) + "\t" + toString((1/(float)iters))); m->mothurOutEndLine();
 404                                         outSum << setw(20) << left << groupNames[j]+'-'+groupNames[i] << '\t' << setprecision(8) << savedDXYValues[spotj][spoti] << '\t' << '<' <<setprecision (precision) << 1/(float)iters << endl;
 405                                 }
 406                         }
 407                 }
 408
 409                 outSum.close();
 410                 return 0;
 411         }
 412         catch(exception& e) {
 413                 m->errorOut(e, "LibShuffCommand", "printSummaryFile");
 414                 exit(1);
 415         }
 416 }
 417
 418 //**********************************************************************************************************************
 419
 420 void LibShuffCommand::setGroups() {
 421         try {
 422                 //if the user has not entered specific groups to analyze then do them all
 423                 if (globaldata->Groups.size() == 0) {
 424                         numGroups = globaldata->gGroupmap->getNumGroups();
 425                         for (int i=0; i < numGroups; i++) {
 426                                 globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
 427                         }
 428                 } else {
 429                         if (savegroups != "all") {
 430                                 //check that groups are valid
 431                                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 432                                         if (globaldata->gGroupmap->isValidGroup(globaldata->Groups[i]) != true) {
 433                                                 m->mothurOut(globaldata->Groups[i] + " is not a valid group, and will be disregarded."); m->mothurOutEndLine();
 434                                                 // erase the invalid group from globaldata->Groups
 435                                                 globaldata->Groups.erase(globaldata->Groups.begin()+i);
 436                                         }
 437                                 }
 438
 439                                 //if the user only entered invalid groups
 440                                 if ((globaldata->Groups.size() == 0) || (globaldata->Groups.size() == 1)) {
 441                                         numGroups = globaldata->gGroupmap->getNumGroups();
 442                                         for (int i=0; i < numGroups; i++) {
 443                                                 globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
 444                                         }
 445                                         m->mothurOut("When using the groups parameter you must have at least 2 valid groups. I will run the command using all the groups in your groupfile."); m->mothurOutEndLine();
 446                                 } else { numGroups = globaldata->Groups.size(); }
 447                         } else { //users wants all groups
 448                                 numGroups = globaldata->gGroupmap->getNumGroups();
 449                                 globaldata->Groups.clear();
 450                                 for (int i=0; i < numGroups; i++) {
 451                                         globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
 452                                 }
 453                         }
 454                 }
 455
 456                 //sort so labels match
 457                 sort(globaldata->Groups.begin(), globaldata->Groups.end());
 458
 459                 //sort
 460                 sort(globaldata->gGroupmap->namesOfGroups.begin(), globaldata->gGroupmap->namesOfGroups.end());
 461
 462                 for (int i = 0; i < globaldata->gGroupmap->namesOfGroups.size(); i++) {  globaldata->gGroupmap->groupIndex[globaldata->gGroupmap->namesOfGroups[i]] = i;  }
 463
 464                 groupNames = globaldata->Groups;
 465
 466         }
 467         catch(exception& e) {
 468                 m->errorOut(e, "LibShuffCommand", "setGroups");
 469                 exit(1);
 470         }
 471 }
 472
 473 /***********************************************************/