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                                 mothurOut("You must read in a matrix and groupfile using the read.dist command, before you use the libshuff command. "); mothurOutEndLine(); 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 and sorts them
  80
  81                                 /********************************************************************************************/
  82                                 //this is needed because when we read the matrix we sort it into groups in alphabetical order
  83                                 //the rest of the command and the classes used in this command assume specific order
  84                                 /********************************************************************************************/
  85                                 matrix->setGroups(globaldata->gGroupmap->namesOfGroups);
  86                                 vector<int> sizes;
  87                                 for (int i = 0; i < globaldata->gGroupmap->namesOfGroups.size(); i++) {  sizes.push_back(globaldata->gGroupmap->getNumSeqs(globaldata->gGroupmap->namesOfGroups[i]));  }
  88                                 matrix->setSizes(sizes);
  89
  90
  91                                 if(userform == "discrete"){
  92                                         form = new DLibshuff(matrix, iters, step, cutOff);
  93                                 }
  94                                 else{
  95                                         form = new SLibshuff(matrix, iters, cutOff);
  96                                 }
  97                         }
  98
  99                 }
 100
 101         }
 102         catch(exception& e) {
 103                 errorOut(e, "LibShuffCommand", "LibShuffCommand");
 104                 exit(1);
 105         }
 106 }
 107 //**********************************************************************************************************************
 108
 109 void LibShuffCommand::help(){
 110         try {
 111                 mothurOut("The libshuff command can only be executed after a successful read.dist command including a groupfile.\n");
 112                 mothurOut("The libshuff command parameters are groups, iters, step, form and cutoff.  No parameters are required.\n");
 113                 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");
 114                 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");
 115                 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");
 116                 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");
 117                 mothurOut("The libshuff command should be in the following format: libshuff(groups=yourGroups, iters=yourIters, cutOff=yourCutOff, form=yourForm, step=yourStep).\n");
 118                 mothurOut("Example libshuff(groups=A-B-C, iters=500, form=discrete, step=0.01, cutOff=2.0).\n");
 119                 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");
 120                 mothurOut("The libshuff command output two files: .coverage and .slsummary their descriptions are in the manual.\n");
 121                 mothurOut("Note: No spaces between parameter labels (i.e. iters), '=' and parameters (i.e.yourIters).\n\n");
 122         }
 123         catch(exception& e) {
 124                 errorOut(e, "LibShuffCommand", "help");
 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                                 int spoti = globaldata->gGroupmap->groupIndex[groupNames[i]]; //neccessary in case user selects groups so you know where they are in the matrix
 150                                 int spotj = globaldata->gGroupmap->groupIndex[groupNames[j]];
 151
 152                                 for(int p=0;p<iters;p++) {
 153                                         form->randomizeGroups(spoti,spotj);
 154                                         if(form->evaluatePair(spoti,spotj) >= savedDXYValues[spoti][spotj])     {       pValueCounts[i][j]++;   }
 155                                         if(form->evaluatePair(spotj,spoti) >= savedDXYValues[spotj][spoti])     {       pValueCounts[j][i]++;   }
 156                                         reading->update(p);
 157                                 }
 158                                 form->resetGroup(spoti);
 159                                 form->resetGroup(spotj);
 160                         }
 161                 }
 162                 reading->finish();
 163                 delete reading;
 164
 165                 mothurOutEndLine();
 166                 printSummaryFile();
 167                 printCoverageFile();
 168
 169                 //clear out users groups
 170                 globaldata->Groups.clear();
 171                 delete form;
 172
 173                 //delete globaldata's copy of the gmatrix to free up memory
 174                 delete globaldata->gMatrix;  globaldata->gMatrix = NULL;
 175
 176                 return 0;
 177         }
 178         catch(exception& e) {
 179                 errorOut(e, "LibShuffCommand", "execute");
 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
 207                                 int spoti = globaldata->gGroupmap->groupIndex[groupNames[i]]; //neccessary in case user selects groups so you know where they are in the matrix
 208                                 int spotj = globaldata->gGroupmap->groupIndex[groupNames[j]];
 209
 210                                 for(int k=0;k<savedMinValues[spoti][spotj].size();k++){
 211                                         if(allDistances[savedMinValues[spoti][spotj][k]].size() != 0){
 212                                                 allDistances[savedMinValues[spoti][spotj][k]][indices[i][j]]++;
 213                                         }
 214                                         else{
 215                                                 allDistances[savedMinValues[spoti][spotj][k]].assign(numIndices, 0);
 216                                                 allDistances[savedMinValues[spoti][spotj][k]][indices[i][j]] = 1;
 217                                         }
 218                                 }
 219                         }
 220                 }
 221                 it=allDistances.begin();
 222
 223                 //cout << setprecision(8);
 224
 225                 vector<int> prevRow = it->second;
 226                 it++;
 227
 228                 for(;it!=allDistances.end();it++){
 229                         for(int i=0;i<it->second.size();i++){
 230                                 it->second[i] += prevRow[i];
 231                         }
 232                         prevRow = it->second;
 233                 }
 234
 235                 vector<int> lastRow = allDistances.rbegin()->second;
 236                 outCov << setprecision(8);
 237
 238                 outCov << "dist";
 239                 for (int i = 0; i < numGroups; i++){
 240                         outCov << '\t' << groupNames[i];
 241                 }
 242                 for (int i=0;i<numGroups;i++){
 243                         for(int j=i+1;j<numGroups;j++){
 244                                 outCov << '\t' << groupNames[i] << '-' << groupNames[j] << '\t';
 245                                 outCov << groupNames[j] << '-' << groupNames[i];
 246                         }
 247                 }
 248                 outCov << endl;
 249
 250                 for(it=allDistances.begin();it!=allDistances.end();it++){
 251                         outCov << it->first << '\t';
 252                         for(int i=0;i<numGroups;i++){
 253                                 outCov << it->second[indices[i][i]]/(float)lastRow[indices[i][i]] << '\t';
 254                         }
 255                         for(int i=0;i<numGroups;i++){
 256                                 for(int j=i+1;j<numGroups;j++){
 257                                         outCov << it->second[indices[i][j]]/(float)lastRow[indices[i][j]] << '\t';
 258                                         outCov << it->second[indices[j][i]]/(float)lastRow[indices[j][i]] << '\t';
 259                                 }
 260                         }
 261                         outCov << endl;
 262                 }
 263                 outCov.close();
 264         }
 265         catch(exception& e) {
 266                 errorOut(e, "LibShuffCommand", "printCoverageFile");
 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                 mothurOutJustToLog("Comparison\tdCXYScore\tSignificance"); mothurOutEndLine();
 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                                 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                                 if(pValueCounts[i][j]){
 294                                         cout << setw(20) << left << groupNames[i]+'-'+groupNames[j] << '\t' << setprecision(8) << savedDXYValues[spoti][spotj] << '\t' << setprecision(precision) << pValueCounts[i][j]/(float)iters << endl;
 295                                         mothurOutJustToLog(groupNames[i]+"-"+groupNames[j] + "\t" + toString(savedDXYValues[spoti][spotj]) + "\t" + toString((pValueCounts[i][j]/(float)iters))); mothurOutEndLine();
 296                                         outSum << setw(20) << left << groupNames[i]+'-'+groupNames[j] << '\t' << setprecision(8) << savedDXYValues[spoti][spotj] << '\t' << setprecision(precision) << pValueCounts[i][j]/(float)iters << endl;
 297                                 }
 298                                 else{
 299                                         cout << setw(20) << left << groupNames[i]+'-'+groupNames[j] << '\t' << setprecision(8) << savedDXYValues[spoti][spotj] << '\t' << '<' <<setprecision(precision) << 1/(float)iters << endl;
 300                                         mothurOutJustToLog(groupNames[i]+"-"+groupNames[j] + "\t" + toString(savedDXYValues[spoti][spotj]) + "\t" + toString((1/(float)iters))); mothurOutEndLine();
 301                                         outSum << setw(20) << left << groupNames[i]+'-'+groupNames[j] << '\t' << setprecision(8) << savedDXYValues[spoti][spotj] << '\t' << '<' <<setprecision(precision) << 1/(float)iters << endl;
 302                                 }
 303                                 if(pValueCounts[j][i]){
 304                                         cout << setw(20) << left << groupNames[j]+'-'+groupNames[i] << '\t' << setprecision(8) << savedDXYValues[spotj][spoti] << '\t' << setprecision (precision) << pValueCounts[j][i]/(float)iters << endl;
 305                                         mothurOutJustToLog(groupNames[j]+"-"+groupNames[i] + "\t" + toString(savedDXYValues[spotj][spoti]) + "\t" + toString((pValueCounts[j][i]/(float)iters))); mothurOutEndLine();
 306                                         outSum << setw(20) << left << groupNames[j]+'-'+groupNames[i] << '\t' << setprecision(8) << savedDXYValues[spotj][spoti] << '\t' << setprecision (precision) << pValueCounts[j][i]/(float)iters << endl;
 307                                 }
 308                                 else{
 309                                         cout << setw(20) << left << groupNames[j]+'-'+groupNames[i] << '\t' << setprecision(8) << savedDXYValues[spotj][spoti] << '\t' << '<' <<setprecision (precision) << 1/(float)iters << endl;
 310                                         mothurOutJustToLog(groupNames[j]+"-"+groupNames[i] + "\t" + toString(savedDXYValues[spotj][spoti]) + "\t" + toString((1/(float)iters))); mothurOutEndLine();
 311                                         outSum << setw(20) << left << groupNames[j]+'-'+groupNames[i] << '\t' << setprecision(8) << savedDXYValues[spotj][spoti] << '\t' << '<' <<setprecision (precision) << 1/(float)iters << endl;
 312                                 }
 313                         }
 314                 }
 315
 316                 outSum.close();
 317         }
 318         catch(exception& e) {
 319                 errorOut(e, "LibShuffCommand", "printSummaryFile");
 320                 exit(1);
 321         }
 322 }
 323
 324 //**********************************************************************************************************************
 325
 326 void LibShuffCommand::setGroups() {
 327         try {
 328                 //if the user has not entered specific groups to analyze then do them all
 329                 if (globaldata->Groups.size() == 0) {
 330                         numGroups = globaldata->gGroupmap->getNumGroups();
 331                         for (int i=0; i < numGroups; i++) {
 332                                 globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
 333                         }
 334                 } else {
 335                         if (savegroups != "all") {
 336                                 //check that groups are valid
 337                                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 338                                         if (globaldata->gGroupmap->isValidGroup(globaldata->Groups[i]) != true) {
 339                                                 mothurOut(globaldata->Groups[i] + " is not a valid group, and will be disregarded."); mothurOutEndLine();
 340                                                 // erase the invalid group from globaldata->Groups
 341                                                 globaldata->Groups.erase(globaldata->Groups.begin()+i);
 342                                         }
 343                                 }
 344
 345                                 //if the user only entered invalid groups
 346                                 if ((globaldata->Groups.size() == 0) || (globaldata->Groups.size() == 1)) {
 347                                         numGroups = globaldata->gGroupmap->getNumGroups();
 348                                         for (int i=0; i < numGroups; i++) {
 349                                                 globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
 350                                         }
 351                                         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."); mothurOutEndLine();
 352                                 } else { numGroups = globaldata->Groups.size(); }
 353                         } else { //users wants all groups
 354                                 numGroups = globaldata->gGroupmap->getNumGroups();
 355                                 globaldata->Groups.clear();
 356                                 for (int i=0; i < numGroups; i++) {
 357                                         globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
 358                                 }
 359                         }
 360                 }
 361
 362                 //sort so labels match
 363                 sort(globaldata->Groups.begin(), globaldata->Groups.end());
 364
 365                 //sort
 366                 sort(globaldata->gGroupmap->namesOfGroups.begin(), globaldata->gGroupmap->namesOfGroups.end());
 367
 368                 for (int i = 0; i < globaldata->gGroupmap->namesOfGroups.size(); i++) {  globaldata->gGroupmap->groupIndex[globaldata->gGroupmap->namesOfGroups[i]] = i;  }
 369
 370                 groupNames = globaldata->Groups;
 371
 372         }
 373         catch(exception& e) {
 374                 errorOut(e, "LibShuffCommand", "setGroups");
 375                 exit(1);
 376         }
 377 }
 378
 379 /***********************************************************/