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