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