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
  23 LibShuffCommand::LibShuffCommand(string option){
  24         try {
  25                 srand( (unsigned)time( NULL ) );
  26
  27                 globaldata = GlobalData::getInstance();
  28                 abort = false;
  29                 Groups.clear();
  30
  31
  32                 //allow user to run help
  33                 if(option == "help") { help(); abort = true; }
  34
  35                 else {
  36                         //valid paramters for this command
  37                         string Array[] =  {"iters","groups","step","form","cutoff"};
  38                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  39
  40                         OptionParser parser(option);
  41                         map<string, string> parameters = parser.getParameters();
  42
  43                         ValidParameters validParameter;
  44
  45                         //check to make sure all parameters are valid for command
  46                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  47                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  48                         }
  49
  50                         //make sure the user has already run the read.dist command
  51                         if ((globaldata->gMatrix == NULL) || (globaldata->gGroupmap == NULL)) {
  52                                 cout << "You must read in a matrix and groupfile using the read.dist command, before you use the libshuff command. " << endl; abort = true;;
  53                         }
  54
  55                         //check for optional parameter and set defaults
  56                         // ...at some point should added some additional type checking...
  57                         groups = validParameter.validFile(parameters, "groups", false);
  58                         if (groups == "not found") { groups = ""; savegroups = groups; }
  59                         else {
  60                                 savegroups = groups;
  61                                 splitAtDash(groups, Groups);
  62                                 globaldata->Groups = Groups;
  63                         }
  64
  65                         string temp;
  66                         temp = validParameter.validFile(parameters, "iters", false);                            if (temp == "not found") { temp = "10000"; }
  67                         convert(temp, iters);
  68
  69                         temp = validParameter.validFile(parameters, "cutoff", false);                           if (temp == "not found") { temp = "1.0"; }
  70                         convert(temp, cutOff);
  71
  72                         temp = validParameter.validFile(parameters, "step", false);                             if (temp == "not found") { temp = "0.01"; }
  73                         convert(temp, step);
  74
  75                         userform = validParameter.validFile(parameters, "form", false);                 if (userform == "not found") { userform = "integral"; }
  76
  77                         if (abort == false) {
  78
  79                                 matrix = globaldata->gMatrix;                           //get the distance matrix
  80                                 setGroups();                                                            //set the groups to be analyzed
  81
  82                                 if(userform == "discrete"){
  83                                         form = new DLibshuff(matrix, iters, step, cutOff);
  84                                 }
  85                                 else{
  86                                         form = new SLibshuff(matrix, iters, cutOff);
  87                                 }
  88                         }
  89
  90                 }
  91
  92         }
  93         catch(exception& e) {
  94                 cout << "Standard Error: " << e.what() << " has occurred in the LibShuffCommand class Function LibShuffCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  95                 exit(1);
  96         }
  97         catch(...) {
  98                 cout << "An unknown error has occurred in the LibShuffCommand class function LibShuffCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  99                 exit(1);
 100         }
 101
 102 }
 103 //**********************************************************************************************************************
 104
 105 void LibShuffCommand::help(){
 106         try {
 107                 cout << "The libshuff command can only be executed after a successful read.dist command including a groupfile." << "\n";
 108                 cout << "The libshuff command parameters are groups, iters, step, form and cutoff.  No parameters are required." << "\n";
 109                 cout << "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";
 110                 cout << "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";
 111                 cout << "The step parameter allows you to specify change in distance you would like between each output if you are using the discrete form." << "\n";
 112                 cout << "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";
 113                 cout << "The libshuff command should be in the following format: libshuff(groups=yourGroups, iters=yourIters, cutOff=yourCutOff, form=yourForm, step=yourStep)." << "\n";
 114                 cout << "Example libshuff(groups=A-B-C, iters=500, form=discrete, step=0.01, cutOff=2.0)." << "\n";
 115                 cout << "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";
 116                 cout << "The libshuff command output two files: .coverage and .slsummary their descriptions are in the manual." << "\n";
 117                 cout << "Note: No spaces between parameter labels (i.e. iters), '=' and parameters (i.e.yourIters)." << "\n" << "\n";
 118         }
 119         catch(exception& e) {
 120                 cout << "Standard Error: " << e.what() << " has occurred in the LibShuffCommand class Function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 121                 exit(1);
 122         }
 123         catch(...) {
 124                 cout << "An unknown error has occurred in the LibShuffCommand class function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 125                 exit(1);
 126         }
 127 }
 128
 129 //**********************************************************************************************************************
 130
 131 int LibShuffCommand::execute(){
 132         try {
 133
 134                 if (abort == true) {    return 0;       }
 135
 136                 savedDXYValues = form->evaluateAll();
 137                 savedMinValues = form->getSavedMins();
 138
 139                 pValueCounts.resize(numGroups);
 140                 for(int i=0;i<numGroups;i++){
 141                         pValueCounts[i].assign(numGroups, 0);
 142                 }
 143
 144                 Progress* reading = new Progress();
 145
 146                 for(int i=0;i<numGroups-1;i++) {
 147                         for(int j=i+1;j<numGroups;j++) {
 148                                 reading->newLine(groupNames[i]+'-'+groupNames[j], iters);
 149                                 for(int p=0;p<iters;p++) {
 150                                         form->randomizeGroups(i,j);
 151                                         if(form->evaluatePair(i,j) >= savedDXYValues[i][j])     {       pValueCounts[i][j]++;   }
 152                                         if(form->evaluatePair(j,i) >= savedDXYValues[j][i])     {       pValueCounts[j][i]++;   }
 153                                         reading->update(p);
 154                                 }
 155                                 form->resetGroup(i);
 156                                 form->resetGroup(j);
 157                         }
 158                 }
 159                 reading->finish();
 160                 delete reading;
 161
 162                 cout << endl;
 163                 printSummaryFile();
 164                 printCoverageFile();
 165
 166                 //clear out users groups
 167                 globaldata->Groups.clear();
 168                 delete form;
 169
 170                 //delete globaldata's copy of the gmatrix to free up memory
 171                 delete globaldata->gMatrix;  globaldata->gMatrix = NULL;
 172
 173                 return 0;
 174         }
 175         catch(exception& e) {
 176                 cout << "Standard Error: " << e.what() << " has occurred in the LibShuffCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 177                 exit(1);
 178         }
 179         catch(...) {
 180                 cout << "An unknown error has occurred in the LibShuffCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 181                 exit(1);
 182         }
 183 }
 184
 185 //**********************************************************************************************************************
 186
 187 void LibShuffCommand::printCoverageFile() {
 188         try {
 189
 190                 ofstream outCov;
 191                 summaryFile = getRootName(globaldata->getPhylipFile()) + "libshuff.coverage";
 192                 openOutputFile(summaryFile, outCov);
 193                 outCov.setf(ios::fixed, ios::floatfield); outCov.setf(ios::showpoint);
 194                 cout.setf(ios::fixed, ios::floatfield); cout.setf(ios::showpoint);
 195
 196                 map<double,vector<int> > allDistances;
 197                 map<double,vector<int> >::iterator it;
 198
 199                 vector<vector<int> > indices(numGroups);
 200                 int numIndices = numGroups * numGroups;
 201
 202                 int index = 0;
 203                 for(int i=0;i<numGroups;i++){
 204                         indices[i].assign(numGroups,0);
 205                         for(int j=0;j<numGroups;j++){
 206                                 indices[i][j] = index++;
 207                                 for(int k=0;k<savedMinValues[i][j].size();k++){
 208                                         if(allDistances[savedMinValues[i][j][k]].size() != 0){
 209                                                 allDistances[savedMinValues[i][j][k]][indices[i][j]]++;
 210                                         }
 211                                         else{
 212                                                 allDistances[savedMinValues[i][j][k]].assign(numIndices, 0);
 213                                                 allDistances[savedMinValues[i][j][k]][indices[i][j]] = 1;
 214                                         }
 215                                 }
 216                         }
 217                 }
 218                 it=allDistances.begin();
 219
 220                 cout << setprecision(8);
 221
 222                 vector<int> prevRow = it->second;
 223                 it++;
 224
 225                 for(it;it!=allDistances.end();it++){
 226                         for(int i=0;i<it->second.size();i++){
 227                                 it->second[i] += prevRow[i];
 228                         }
 229                         prevRow = it->second;
 230                 }
 231
 232                 vector<int> lastRow = allDistances.rbegin()->second;
 233                 outCov << setprecision(8);
 234
 235                 outCov << "dist";
 236                 for (int i = 0; i < numGroups; i++){
 237                         outCov << '\t' << groupNames[i];
 238                 }
 239                 for (int i=0;i<numGroups;i++){
 240                         for(int j=i+1;j<numGroups;j++){
 241                                 outCov << '\t' << groupNames[i] << '-' << groupNames[j] << '\t';
 242                                 outCov << groupNames[j] << '-' << groupNames[i];
 243                         }
 244                 }
 245                 outCov << endl;
 246
 247                 for(it=allDistances.begin();it!=allDistances.end();it++){
 248                         outCov << it->first << '\t';
 249                         for(int i=0;i<numGroups;i++){
 250                                 outCov << it->second[indices[i][i]]/(float)lastRow[indices[i][i]] << '\t';
 251                         }
 252                         for(int i=0;i<numGroups;i++){
 253                                 for(int j=i+1;j<numGroups;j++){
 254                                         outCov << it->second[indices[i][j]]/(float)lastRow[indices[i][j]] << '\t';
 255                                         outCov << it->second[indices[j][i]]/(float)lastRow[indices[j][i]] << '\t';
 256                                 }
 257                         }
 258                         outCov << endl;
 259                 }
 260
 261         }
 262         catch(exception& e) {
 263                 cout << "Standard Error: " << e.what() << " has occurred in the LibShuffCommand class Function printCoverageFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 264                 exit(1);
 265         }
 266         catch(...) {
 267                 cout << "An unknown error has occurred in the LibShuffCommand class function printCoverageFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 268                 exit(1);
 269         }
 270 }
 271
 272 //**********************************************************************************************************************
 273
 274 void LibShuffCommand::printSummaryFile() {
 275         try {
 276
 277                 ofstream outSum;
 278                 summaryFile = getRootName(globaldata->getPhylipFile()) + "libshuff.summary";
 279                 openOutputFile(summaryFile, outSum);
 280
 281                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 282                 cout.setf(ios::fixed, ios::floatfield); cout.setf(ios::showpoint);
 283
 284                 cout << setw(20) << left << "Comparison" << '\t' << setprecision(8) << "dCXYScore" << '\t' << "Significance" << endl;
 285                 outSum << setw(20) << left << "Comparison" << '\t' << setprecision(8) << "dCXYScore" << '\t' << "Significance" << endl;
 286
 287                 int precision = (int)log10(iters);
 288                 for(int i=0;i<numGroups;i++){
 289                         for(int j=i+1;j<numGroups;j++){
 290                                 if(pValueCounts[i][j]){
 291                                         cout << setw(20) << left << groupNames[i]+'-'+groupNames[j] << '\t' << setprecision(8) << savedDXYValues[i][j] << '\t' << setprecision(precision) << pValueCounts[i][j]/(float)iters << endl;
 292                                         outSum << setw(20) << left << groupNames[i]+'-'+groupNames[j] << '\t' << setprecision(8) << savedDXYValues[i][j] << '\t' << setprecision(precision) << pValueCounts[i][j]/(float)iters << endl;
 293                                 }
 294                                 else{
 295                                         cout << setw(20) << left << groupNames[i]+'-'+groupNames[j] << '\t' << setprecision(8) << savedDXYValues[i][j] << '\t' << '<' <<setprecision(precision) << 1/(float)iters << endl;
 296                                         outSum << setw(20) << left << groupNames[i]+'-'+groupNames[j] << '\t' << setprecision(8) << savedDXYValues[i][j] << '\t' << '<' <<setprecision(precision) << 1/(float)iters << endl;
 297                                 }
 298                                 if(pValueCounts[j][i]){
 299                                         cout << setw(20) << left << groupNames[j]+'-'+groupNames[i] << '\t' << setprecision(8) << savedDXYValues[j][i] << '\t' << setprecision (precision) << pValueCounts[j][i]/(float)iters << endl;
 300                                         outSum << setw(20) << left << groupNames[j]+'-'+groupNames[i] << '\t' << setprecision(8) << savedDXYValues[j][i] << '\t' << setprecision (precision) << pValueCounts[j][i]/(float)iters << endl;
 301                                 }
 302                                 else{
 303                                         cout << setw(20) << left << groupNames[j]+'-'+groupNames[i] << '\t' << setprecision(8) << savedDXYValues[j][i] << '\t' << '<' <<setprecision (precision) << 1/(float)iters << endl;
 304                                         outSum << setw(20) << left << groupNames[j]+'-'+groupNames[i] << '\t' << setprecision(8) << savedDXYValues[j][i] << '\t' << '<' <<setprecision (precision) << 1/(float)iters << endl;
 305                                 }
 306                         }
 307                 }
 308
 309
 310         }
 311         catch(exception& e) {
 312                 cout << "Standard Error: " << e.what() << " has occurred in the LibShuffCommand class Function printSummaryFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 313                 exit(1);
 314         }
 315         catch(...) {
 316                 cout << "An unknown error has occurred in the LibShuffCommand class function printSummaryFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 317                 exit(1);
 318         }
 319 }
 320
 321 //**********************************************************************************************************************
 322
 323 void LibShuffCommand::setGroups() {
 324         try {
 325                 //if the user has not entered specific groups to analyze then do them all
 326                 if (globaldata->Groups.size() == 0) {
 327                         numGroups = globaldata->gGroupmap->getNumGroups();
 328                         for (int i=0; i < numGroups; i++) {
 329                                 globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
 330                         }
 331                 } else {
 332                         if (savegroups != "all") {
 333                                 //check that groups are valid
 334                                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 335                                         if (globaldata->gGroupmap->isValidGroup(globaldata->Groups[i]) != true) {
 336                                                 cout << globaldata->Groups[i] << " is not a valid group, and will be disregarded." << endl;
 337                                                 // erase the invalid group from globaldata->Groups
 338                                                 globaldata->Groups.erase(globaldata->Groups.begin()+i);
 339                                         }
 340                                 }
 341
 342                                 //if the user only entered invalid groups
 343                                 if ((globaldata->Groups.size() == 0) || (globaldata->Groups.size() == 1)) {
 344                                         numGroups = globaldata->gGroupmap->getNumGroups();
 345                                         for (int i=0; i < numGroups; i++) {
 346                                                 globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
 347                                         }
 348                                         cout << "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." << endl;
 349                                 } else { numGroups = globaldata->Groups.size(); }
 350                         } else { //users wants all groups
 351                                 numGroups = globaldata->gGroupmap->getNumGroups();
 352                                 globaldata->Groups.clear();
 353                                 for (int i=0; i < numGroups; i++) {
 354                                         globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
 355                                 }
 356                         }
 357                 }
 358
 359                 //sort so labels match
 360                 sort(globaldata->Groups.begin(), globaldata->Groups.end());
 361
 362                 //sort
 363                 sort(globaldata->gGroupmap->namesOfGroups.begin(), globaldata->gGroupmap->namesOfGroups.end());
 364
 365                 groupNames = globaldata->Groups;
 366
 367                 // number of comparisons i.e. with groups A,B,C = AA, AB, AC, BA, BB, BC...;
 368 //              for (int i=0; i<numGroups; i++) {
 369 //                      for (int l = 0; l < numGroups; l++) {
 370 //                              //set group comparison labels
 371 //                              groupComb.push_back(globaldata->Groups[i] + "-" + globaldata->Groups[l]);
 372 //                      }
 373 //              }
 374         }
 375         catch(exception& e) {
 376                 cout << "Standard Error: " << e.what() << " has occurred in the LibShuffCommand class Function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 377                 exit(1);
 378         }
 379         catch(...) {
 380                 cout << "An unknown error has occurred in the LibShuffCommand class function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 381                 exit(1);
 382         }
 383 }
 384
 385 /***********************************************************/