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                 abort = true;
  37                 //initialize outputTypes
  38                 vector<string> tempOutNames;
  39                 outputTypes["coverage"] = tempOutNames;
  40                 outputTypes["libshuffsummary"] = tempOutNames;
  41         }
  42         catch(exception& e) {
  43                 m->errorOut(e, "LibShuffCommand", "LibShuffCommand");
  44                 exit(1);
  45         }
  46 }
  47 //**********************************************************************************************************************
  48 vector<string> LibShuffCommand::getRequiredParameters(){
  49         try {
  50                 vector<string> myArray;
  51                 return myArray;
  52         }
  53         catch(exception& e) {
  54                 m->errorOut(e, "LibShuffCommand", "getRequiredParameters");
  55                 exit(1);
  56         }
  57 }
  58 //**********************************************************************************************************************
  59 vector<string> LibShuffCommand::getRequiredFiles(){
  60         try {
  61                 string Array[] =  {"phylip","group"};
  62                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  63                 return myArray;
  64         }
  65         catch(exception& e) {
  66                 m->errorOut(e, "LibShuffCommand", "getRequiredFiles");
  67                 exit(1);
  68         }
  69 }
  70 //**********************************************************************************************************************
  71
  72 LibShuffCommand::LibShuffCommand(string option)  {
  73         try {
  74                 globaldata = GlobalData::getInstance();
  75                 abort = false;
  76                 Groups.clear();
  77
  78                 //allow user to run help
  79                 if(option == "help") { help(); abort = true; }
  80
  81                 else {
  82                         //valid paramters for this command
  83                         string Array[] =  {"iters","groups","step","form","cutoff","outputdir","inputdir"};
  84                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  85
  86                         OptionParser parser(option);
  87                         map<string, string> parameters = parser.getParameters();
  88
  89                         ValidParameters validParameter;
  90
  91                         //check to make sure all parameters are valid for command
  92                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  93                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  94                         }
  95
  96                         //initialize outputTypes
  97                         vector<string> tempOutNames;
  98                         outputTypes["coverage"] = tempOutNames;
  99                         outputTypes["libshuffsummary"] = tempOutNames;
 100
 101                         //if the user changes the output directory command factory will send this info to us in the output parameter
 102                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 103                                 outputDir = "";
 104                                 outputDir += m->hasPath(globaldata->getPhylipFile()); //if user entered a file with a path then preserve it
 105                         }
 106
 107                         //make sure the user has already run the read.dist command
 108                         if ((globaldata->gMatrix == NULL) || (globaldata->gGroupmap == NULL)) {
 109                                 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;;
 110                         }
 111
 112                         //check for optional parameter and set defaults
 113                         // ...at some point should added some additional type checking...
 114                         groups = validParameter.validFile(parameters, "groups", false);
 115                         if (groups == "not found") { groups = ""; savegroups = groups; }
 116                         else {
 117                                 savegroups = groups;
 118                                 m->splitAtDash(groups, Groups);
 119                                 globaldata->Groups = Groups;
 120                         }
 121
 122                         string temp;
 123                         temp = validParameter.validFile(parameters, "iters", false);                            if (temp == "not found") { temp = "10000"; }
 124                         convert(temp, iters);
 125
 126                         temp = validParameter.validFile(parameters, "cutoff", false);                           if (temp == "not found") { temp = "1.0"; }
 127                         convert(temp, cutOff);
 128
 129                         temp = validParameter.validFile(parameters, "step", false);                             if (temp == "not found") { temp = "0.01"; }
 130                         convert(temp, step);
 131
 132                         userform = validParameter.validFile(parameters, "form", false);                 if (userform == "not found") { userform = "integral"; }
 133
 134                         if (abort == false) {
 135
 136                                 matrix = globaldata->gMatrix;                           //get the distance matrix
 137                                 setGroups();                                                            //set the groups to be analyzed and sorts them
 138
 139                                 /********************************************************************************************/
 140                                 //this is needed because when we read the matrix we sort it into groups in alphabetical order
 141                                 //the rest of the command and the classes used in this command assume specific order
 142                                 /********************************************************************************************/
 143                                 matrix->setGroups(globaldata->gGroupmap->namesOfGroups);
 144                                 vector<int> sizes;
 145                                 for (int i = 0; i < globaldata->gGroupmap->namesOfGroups.size(); i++) {   sizes.push_back(globaldata->gGroupmap->getNumSeqs(globaldata->gGroupmap->namesOfGroups[i]));  }
 146                                 matrix->setSizes(sizes);
 147
 148
 149                                 if(userform == "discrete"){
 150                                         form = new DLibshuff(matrix, iters, step, cutOff);
 151                                 }
 152                                 else{
 153                                         form = new SLibshuff(matrix, iters, cutOff);
 154                                 }
 155                         }
 156
 157                 }
 158
 159         }
 160         catch(exception& e) {
 161                 m->errorOut(e, "LibShuffCommand", "LibShuffCommand");
 162                 exit(1);
 163         }
 164 }
 165 //**********************************************************************************************************************
 166
 167 void LibShuffCommand::help(){
 168         try {
 169                 m->mothurOut("The libshuff command can only be executed after a successful read.dist command including a groupfile.\n");
 170                 m->mothurOut("The libshuff command parameters are groups, iters, step, form and cutoff.  No parameters are required.\n");
 171                 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");
 172                 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");
 173                 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");
 174                 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");
 175                 m->mothurOut("The libshuff command should be in the following format: libshuff(groups=yourGroups, iters=yourIters, cutOff=yourCutOff, form=yourForm, step=yourStep).\n");
 176                 m->mothurOut("Example libshuff(groups=A-B-C, iters=500, form=discrete, step=0.01, cutOff=2.0).\n");
 177                 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");
 178                 m->mothurOut("The libshuff command output two files: .coverage and .slsummary their descriptions are in the manual.\n");
 179                 m->mothurOut("Note: No spaces between parameter labels (i.e. iters), '=' and parameters (i.e.yourIters).\n\n");
 180         }
 181         catch(exception& e) {
 182                 m->errorOut(e, "LibShuffCommand", "help");
 183                 exit(1);
 184         }
 185 }
 186
 187 //**********************************************************************************************************************
 188
 189 int LibShuffCommand::execute(){
 190         try {
 191
 192                 if (abort == true) {    return 0;       }
 193
 194                 savedDXYValues = form->evaluateAll();
 195                 savedMinValues = form->getSavedMins();
 196
 197                 if (m->control_pressed) {  delete form; globaldata->Groups.clear(); delete globaldata->gMatrix;  globaldata->gMatrix = NULL; return 0; }
 198
 199                 pValueCounts.resize(numGroups);
 200                 for(int i=0;i<numGroups;i++){
 201                         pValueCounts[i].assign(numGroups, 0);
 202                 }
 203
 204                 if (m->control_pressed) {  outputTypes.clear(); delete form; globaldata->Groups.clear(); delete globaldata->gMatrix;  globaldata->gMatrix = NULL; return 0; }
 205
 206                 Progress* reading = new Progress();
 207
 208                 for(int i=0;i<numGroups-1;i++) {
 209                         for(int j=i+1;j<numGroups;j++) {
 210
 211                                 if (m->control_pressed) {  outputTypes.clear();  delete form; globaldata->Groups.clear(); delete globaldata->gMatrix;  globaldata->gMatrix = NULL; delete reading; return 0; }
 212
 213                                 reading->newLine(groupNames[i]+'-'+groupNames[j], iters);
 214                                 int spoti = globaldata->gGroupmap->groupIndex[groupNames[i]]; //neccessary in case user selects groups so you know where they are in the matrix
 215                                 int spotj = globaldata->gGroupmap->groupIndex[groupNames[j]];
 216
 217                                 for(int p=0;p<iters;p++) {
 218
 219                                         if (m->control_pressed) {  outputTypes.clear(); delete form; globaldata->Groups.clear(); delete globaldata->gMatrix;  globaldata->gMatrix = NULL; delete reading; return 0; }
 220
 221                                         form->randomizeGroups(spoti,spotj);
 222                                         if(form->evaluatePair(spoti,spotj) >= savedDXYValues[spoti][spotj])     {       pValueCounts[i][j]++;   }
 223                                         if(form->evaluatePair(spotj,spoti) >= savedDXYValues[spotj][spoti])     {       pValueCounts[j][i]++;   }
 224
 225                                         if (m->control_pressed) {  outputTypes.clear(); delete form; globaldata->Groups.clear(); delete globaldata->gMatrix;  globaldata->gMatrix = NULL; delete reading; return 0; }
 226
 227                                         reading->update(p);
 228                                 }
 229                                 form->resetGroup(spoti);
 230                                 form->resetGroup(spotj);
 231                         }
 232                 }
 233
 234                 if (m->control_pressed) { outputTypes.clear();  delete form; globaldata->Groups.clear(); delete globaldata->gMatrix;  globaldata->gMatrix = NULL; delete reading; return 0; }
 235
 236                 reading->finish();
 237                 delete reading;
 238
 239                 m->mothurOutEndLine();
 240                 printSummaryFile();
 241                 printCoverageFile();
 242
 243                 //clear out users groups
 244                 globaldata->Groups.clear();
 245                 delete form;
 246
 247                 //delete globaldata's copy of the gmatrix to free up memory
 248                 delete globaldata->gMatrix;  globaldata->gMatrix = NULL;
 249
 250                 if (m->control_pressed) {  outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); } return 0; }
 251
 252
 253                 m->mothurOutEndLine();
 254                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 255                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 256                 m->mothurOutEndLine();
 257
 258                 return 0;
 259         }
 260         catch(exception& e) {
 261                 m->errorOut(e, "LibShuffCommand", "execute");
 262                 exit(1);
 263         }
 264 }
 265
 266 //**********************************************************************************************************************
 267
 268 int LibShuffCommand::printCoverageFile() {
 269         try {
 270
 271                 ofstream outCov;
 272                 summaryFile = outputDir + m->getRootName(m->getSimpleName(globaldata->getPhylipFile())) + "libshuff.coverage";
 273                 m->openOutputFile(summaryFile, outCov);
 274                 outputNames.push_back(summaryFile); outputTypes["coverage"].push_back(summaryFile);
 275                 outCov.setf(ios::fixed, ios::floatfield); outCov.setf(ios::showpoint);
 276                 //cout.setf(ios::fixed, ios::floatfield); cout.setf(ios::showpoint);
 277
 278                 map<double,vector<int> > allDistances;
 279                 map<double,vector<int> >::iterator it;
 280
 281                 vector<vector<int> > indices(numGroups);
 282                 int numIndices = numGroups * numGroups;
 283
 284                 int index = 0;
 285                 for(int i=0;i<numGroups;i++){
 286                         indices[i].assign(numGroups,0);
 287                         for(int j=0;j<numGroups;j++){
 288                                 indices[i][j] = index++;
 289
 290                                 int spoti = globaldata->gGroupmap->groupIndex[groupNames[i]]; //neccessary in case user selects groups so you know where they are in the matrix
 291                                 int spotj = globaldata->gGroupmap->groupIndex[groupNames[j]];
 292
 293                                 for(int k=0;k<savedMinValues[spoti][spotj].size();k++){
 294
 295                                         if(m->control_pressed)  { outCov.close(); return 0; }
 296
 297                                         if(allDistances[savedMinValues[spoti][spotj][k]].size() != 0){
 298                                                 allDistances[savedMinValues[spoti][spotj][k]][indices[i][j]]++;
 299                                         }
 300                                         else{
 301                                                 allDistances[savedMinValues[spoti][spotj][k]].assign(numIndices, 0);
 302                                                 allDistances[savedMinValues[spoti][spotj][k]][indices[i][j]] = 1;
 303                                         }
 304                                 }
 305                         }
 306                 }
 307                 it=allDistances.begin();
 308
 309                 //cout << setprecision(8);
 310
 311                 vector<int> prevRow = it->second;
 312                 it++;
 313
 314                 for(;it!=allDistances.end();it++){
 315                         for(int i=0;i<it->second.size();i++){
 316                                 it->second[i] += prevRow[i];
 317                         }
 318                         prevRow = it->second;
 319                 }
 320
 321                 vector<int> lastRow = allDistances.rbegin()->second;
 322                 outCov << setprecision(8);
 323
 324                 outCov << "dist";
 325                 for (int i = 0; i < numGroups; i++){
 326                         outCov << '\t' << groupNames[i];
 327                 }
 328                 for (int i=0;i<numGroups;i++){
 329                         for(int j=i+1;j<numGroups;j++){
 330                                 if(m->control_pressed)  { outCov.close(); return 0; }
 331                                 outCov << '\t' << groupNames[i] << '-' << groupNames[j] << '\t';
 332                                 outCov << groupNames[j] << '-' << groupNames[i];
 333                         }
 334                 }
 335                 outCov << endl;
 336
 337                 for(it=allDistances.begin();it!=allDistances.end();it++){
 338                         outCov << it->first << '\t';
 339                         for(int i=0;i<numGroups;i++){
 340                                 outCov << it->second[indices[i][i]]/(float)lastRow[indices[i][i]] << '\t';
 341                         }
 342                         for(int i=0;i<numGroups;i++){
 343                                 for(int j=i+1;j<numGroups;j++){
 344                                         if(m->control_pressed)  { outCov.close(); return 0; }
 345
 346                                         outCov << it->second[indices[i][j]]/(float)lastRow[indices[i][j]] << '\t';
 347                                         outCov << it->second[indices[j][i]]/(float)lastRow[indices[j][i]] << '\t';
 348                                 }
 349                         }
 350                         outCov << endl;
 351                 }
 352                 outCov.close();
 353
 354                 return 0;
 355         }
 356         catch(exception& e) {
 357                 m->errorOut(e, "LibShuffCommand", "printCoverageFile");
 358                 exit(1);
 359         }
 360 }
 361
 362 //**********************************************************************************************************************
 363
 364 int LibShuffCommand::printSummaryFile() {
 365         try {
 366
 367                 ofstream outSum;
 368                 summaryFile = outputDir + m->getRootName(m->getSimpleName(globaldata->getPhylipFile())) + "libshuff.summary";
 369                 m->openOutputFile(summaryFile, outSum);
 370                 outputNames.push_back(summaryFile); outputTypes["libshuffsummary"].push_back(summaryFile);
 371
 372                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 373                 cout.setf(ios::fixed, ios::floatfield); cout.setf(ios::showpoint);
 374
 375                 cout << setw(20) << left << "Comparison" << '\t' << setprecision(8) << "dCXYScore" << '\t' << "Significance" << endl;
 376                 m->mothurOutJustToLog("Comparison\tdCXYScore\tSignificance"); m->mothurOutEndLine();
 377                 outSum << setw(20) << left << "Comparison" << '\t' << setprecision(8) << "dCXYScore" << '\t' << "Significance" << endl;
 378
 379                 int precision = (int)log10(iters);
 380                 for(int i=0;i<numGroups;i++){
 381                         for(int j=i+1;j<numGroups;j++){
 382                                 if(m->control_pressed)  { outSum.close(); return 0; }
 383
 384                                 int spoti = globaldata->gGroupmap->groupIndex[groupNames[i]]; //neccessary in case user selects groups so you know where they are in the matrix
 385                                 int spotj = globaldata->gGroupmap->groupIndex[groupNames[j]];
 386
 387                                 if(pValueCounts[i][j]){
 388                                         cout << setw(20) << left << groupNames[i]+'-'+groupNames[j] << '\t' << setprecision(8) << savedDXYValues[spoti][spotj] << '\t' << setprecision(precision) << pValueCounts[i][j]/(float)iters << endl;
 389                                         m->mothurOutJustToLog(groupNames[i]+"-"+groupNames[j] + "\t" + toString(savedDXYValues[spoti][spotj]) + "\t" + toString((pValueCounts[i][j]/(float)iters))); m->mothurOutEndLine();
 390                                         outSum << setw(20) << left << groupNames[i]+'-'+groupNames[j] << '\t' << setprecision(8) << savedDXYValues[spoti][spotj] << '\t' << setprecision(precision) << pValueCounts[i][j]/(float)iters << endl;
 391                                 }
 392                                 else{
 393                                         cout << setw(20) << left << groupNames[i]+'-'+groupNames[j] << '\t' << setprecision(8) << savedDXYValues[spoti][spotj] << '\t' << '<' <<setprecision(precision) << 1/(float)iters << endl;
 394                                         m->mothurOutJustToLog(groupNames[i]+"-"+groupNames[j] + "\t" + toString(savedDXYValues[spoti][spotj]) + "\t" + toString((1/(float)iters))); m->mothurOutEndLine();
 395                                         outSum << setw(20) << left << groupNames[i]+'-'+groupNames[j] << '\t' << setprecision(8) << savedDXYValues[spoti][spotj] << '\t' << '<' <<setprecision(precision) << 1/(float)iters << endl;
 396                                 }
 397                                 if(pValueCounts[j][i]){
 398                                         cout << setw(20) << left << groupNames[j]+'-'+groupNames[i] << '\t' << setprecision(8) << savedDXYValues[spotj][spoti] << '\t' << setprecision (precision) << pValueCounts[j][i]/(float)iters << endl;
 399                                         m->mothurOutJustToLog(groupNames[j]+"-"+groupNames[i] + "\t" + toString(savedDXYValues[spotj][spoti]) + "\t" + toString((pValueCounts[j][i]/(float)iters))); m->mothurOutEndLine();
 400                                         outSum << setw(20) << left << groupNames[j]+'-'+groupNames[i] << '\t' << setprecision(8) << savedDXYValues[spotj][spoti] << '\t' << setprecision (precision) << pValueCounts[j][i]/(float)iters << endl;
 401                                 }
 402                                 else{
 403                                         cout << setw(20) << left << groupNames[j]+'-'+groupNames[i] << '\t' << setprecision(8) << savedDXYValues[spotj][spoti] << '\t' << '<' <<setprecision (precision) << 1/(float)iters << endl;
 404                                         m->mothurOutJustToLog(groupNames[j]+"-"+groupNames[i] + "\t" + toString(savedDXYValues[spotj][spoti]) + "\t" + toString((1/(float)iters))); m->mothurOutEndLine();
 405                                         outSum << setw(20) << left << groupNames[j]+'-'+groupNames[i] << '\t' << setprecision(8) << savedDXYValues[spotj][spoti] << '\t' << '<' <<setprecision (precision) << 1/(float)iters << endl;
 406                                 }
 407                         }
 408                 }
 409
 410                 outSum.close();
 411                 return 0;
 412         }
 413         catch(exception& e) {
 414                 m->errorOut(e, "LibShuffCommand", "printSummaryFile");
 415                 exit(1);
 416         }
 417 }
 418
 419 //**********************************************************************************************************************
 420
 421 void LibShuffCommand::setGroups() {
 422         try {
 423                 //if the user has not entered specific groups to analyze then do them all
 424                 if (globaldata->Groups.size() == 0) {
 425                         numGroups = globaldata->gGroupmap->getNumGroups();
 426                         for (int i=0; i < numGroups; i++) {
 427                                 globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
 428                         }
 429                 } else {
 430                         if (savegroups != "all") {
 431                                 //check that groups are valid
 432                                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 433                                         if (globaldata->gGroupmap->isValidGroup(globaldata->Groups[i]) != true) {
 434                                                 m->mothurOut(globaldata->Groups[i] + " is not a valid group, and will be disregarded."); m->mothurOutEndLine();
 435                                                 // erase the invalid group from globaldata->Groups
 436                                                 globaldata->Groups.erase(globaldata->Groups.begin()+i);
 437                                         }
 438                                 }
 439
 440                                 //if the user only entered invalid groups
 441                                 if ((globaldata->Groups.size() == 0) || (globaldata->Groups.size() == 1)) {
 442                                         numGroups = globaldata->gGroupmap->getNumGroups();
 443                                         for (int i=0; i < numGroups; i++) {
 444                                                 globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
 445                                         }
 446                                         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();
 447                                 } else { numGroups = globaldata->Groups.size(); }
 448                         } else { //users wants all groups
 449                                 numGroups = globaldata->gGroupmap->getNumGroups();
 450                                 globaldata->Groups.clear();
 451                                 for (int i=0; i < numGroups; i++) {
 452                                         globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
 453                                 }
 454                         }
 455                 }
 456
 457                 //sort so labels match
 458                 sort(globaldata->Groups.begin(), globaldata->Groups.end());
 459
 460                 //sort
 461                 sort(globaldata->gGroupmap->namesOfGroups.begin(), globaldata->gGroupmap->namesOfGroups.end());
 462
 463                 for (int i = 0; i < globaldata->gGroupmap->namesOfGroups.size(); i++) {  globaldata->gGroupmap->groupIndex[globaldata->gGroupmap->namesOfGroups[i]] = i;  }
 464
 465                 groupNames = globaldata->Groups;
 466
 467         }
 468         catch(exception& e) {
 469                 m->errorOut(e, "LibShuffCommand", "setGroups");
 470                 exit(1);
 471         }
 472 }
 473
 474 /***********************************************************/