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                 //allow user to run help
  30                 if(option == "help") { help(); abort = true; }
  31
  32                 else {
  33                         //valid paramters for this command
  34                         string Array[] =  {"iters","groups","step","form","cutoff","outputdir","inputdir"};
  35                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  36
  37                         OptionParser parser(option);
  38                         map<string, string> parameters = parser.getParameters();
  39
  40                         ValidParameters validParameter;
  41
  42                         //check to make sure all parameters are valid for command
  43                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  44                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  45                         }
  46
  47                         //if the user changes the output directory command factory will send this info to us in the output parameter
  48                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
  49                                 outputDir = "";
  50                                 outputDir += hasPath(globaldata->getPhylipFile()); //if user entered a file with a path then preserve it
  51                         }
  52
  53                         //make sure the user has already run the read.dist command
  54                         if ((globaldata->gMatrix == NULL) || (globaldata->gGroupmap == NULL)) {
  55                                 m->mothurOut("You must read in a matrix and groupfile using the read.dist command, before you use the libshuff command. "); m->mothurOutEndLine(); abort = true;;
  56                         }
  57
  58                         //check for optional parameter and set defaults
  59                         // ...at some point should added some additional type checking...
  60                         groups = validParameter.validFile(parameters, "groups", false);
  61                         if (groups == "not found") { groups = ""; savegroups = groups; }
  62                         else {
  63                                 savegroups = groups;
  64                                 splitAtDash(groups, Groups);
  65                                 globaldata->Groups = Groups;
  66                         }
  67
  68                         string temp;
  69                         temp = validParameter.validFile(parameters, "iters", false);                            if (temp == "not found") { temp = "10000"; }
  70                         convert(temp, iters);
  71
  72                         temp = validParameter.validFile(parameters, "cutoff", false);                           if (temp == "not found") { temp = "1.0"; }
  73                         convert(temp, cutOff);
  74
  75                         temp = validParameter.validFile(parameters, "step", false);                             if (temp == "not found") { temp = "0.01"; }
  76                         convert(temp, step);
  77
  78                         userform = validParameter.validFile(parameters, "form", false);                 if (userform == "not found") { userform = "integral"; }
  79
  80                         if (abort == false) {
  81
  82                                 matrix = globaldata->gMatrix;                           //get the distance matrix
  83                                 setGroups();                                                            //set the groups to be analyzed and sorts them
  84
  85                                 /********************************************************************************************/
  86                                 //this is needed because when we read the matrix we sort it into groups in alphabetical order
  87                                 //the rest of the command and the classes used in this command assume specific order
  88                                 /********************************************************************************************/
  89                                 matrix->setGroups(globaldata->gGroupmap->namesOfGroups);
  90                                 vector<int> sizes;
  91                                 for (int i = 0; i < globaldata->gGroupmap->namesOfGroups.size(); i++) {   sizes.push_back(globaldata->gGroupmap->getNumSeqs(globaldata->gGroupmap->namesOfGroups[i]));  }
  92                                 matrix->setSizes(sizes);
  93
  94
  95                                 if(userform == "discrete"){
  96                                         form = new DLibshuff(matrix, iters, step, cutOff);
  97                                 }
  98                                 else{
  99                                         form = new SLibshuff(matrix, iters, cutOff);
 100                                 }
 101                         }
 102
 103                 }
 104
 105         }
 106         catch(exception& e) {
 107                 m->errorOut(e, "LibShuffCommand", "LibShuffCommand");
 108                 exit(1);
 109         }
 110 }
 111 //**********************************************************************************************************************
 112
 113 void LibShuffCommand::help(){
 114         try {
 115                 m->mothurOut("The libshuff command can only be executed after a successful read.dist command including a groupfile.\n");
 116                 m->mothurOut("The libshuff command parameters are groups, iters, step, form and cutoff.  No parameters are required.\n");
 117                 m->mothurOut("The groups parameter allows you to specify which of the groups in your groupfile you would like analyzed.  You must enter at least 2 valid groups.\n");
 118                 m->mothurOut("The group names are separated by dashes.  The iters parameter allows you to specify how many random matrices you would like compared to your matrix.\n");
 119                 m->mothurOut("The step parameter allows you to specify change in distance you would like between each output if you are using the discrete form.\n");
 120                 m->mothurOut("The form parameter allows you to specify if you would like to analyze your matrix using the discrete or integral form. Your options are integral or discrete.\n");
 121                 m->mothurOut("The libshuff command should be in the following format: libshuff(groups=yourGroups, iters=yourIters, cutOff=yourCutOff, form=yourForm, step=yourStep).\n");
 122                 m->mothurOut("Example libshuff(groups=A-B-C, iters=500, form=discrete, step=0.01, cutOff=2.0).\n");
 123                 m->mothurOut("The default value for groups is all the groups in your groupfile, iters is 10000, cutoff is 1.0, form is integral and step is 0.01.\n");
 124                 m->mothurOut("The libshuff command output two files: .coverage and .slsummary their descriptions are in the manual.\n");
 125                 m->mothurOut("Note: No spaces between parameter labels (i.e. iters), '=' and parameters (i.e.yourIters).\n\n");
 126         }
 127         catch(exception& e) {
 128                 m->errorOut(e, "LibShuffCommand", "help");
 129                 exit(1);
 130         }
 131 }
 132
 133 //**********************************************************************************************************************
 134
 135 int LibShuffCommand::execute(){
 136         try {
 137
 138                 if (abort == true) {    return 0;       }
 139
 140                 savedDXYValues = form->evaluateAll();
 141                 savedMinValues = form->getSavedMins();
 142
 143                 pValueCounts.resize(numGroups);
 144                 for(int i=0;i<numGroups;i++){
 145                         pValueCounts[i].assign(numGroups, 0);
 146                 }
 147
 148                 Progress* reading = new Progress();
 149
 150                 for(int i=0;i<numGroups-1;i++) {
 151                         for(int j=i+1;j<numGroups;j++) {
 152                                 reading->newLine(groupNames[i]+'-'+groupNames[j], iters);
 153                                 int spoti = globaldata->gGroupmap->groupIndex[groupNames[i]]; //neccessary in case user selects groups so you know where they are in the matrix
 154                                 int spotj = globaldata->gGroupmap->groupIndex[groupNames[j]];
 155
 156                                 for(int p=0;p<iters;p++) {
 157                                         form->randomizeGroups(spoti,spotj);
 158                                         if(form->evaluatePair(spoti,spotj) >= savedDXYValues[spoti][spotj])     {       pValueCounts[i][j]++;   }
 159                                         if(form->evaluatePair(spotj,spoti) >= savedDXYValues[spotj][spoti])     {       pValueCounts[j][i]++;   }
 160                                         reading->update(p);
 161                                 }
 162                                 form->resetGroup(spoti);
 163                                 form->resetGroup(spotj);
 164                         }
 165                 }
 166                 reading->finish();
 167                 delete reading;
 168
 169                 m->mothurOutEndLine();
 170                 printSummaryFile();
 171                 printCoverageFile();
 172
 173                 //clear out users groups
 174                 globaldata->Groups.clear();
 175                 delete form;
 176
 177                 //delete globaldata's copy of the gmatrix to free up memory
 178                 delete globaldata->gMatrix;  globaldata->gMatrix = NULL;
 179
 180                 m->mothurOutEndLine();
 181                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 182                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 183                 m->mothurOutEndLine();
 184
 185                 return 0;
 186         }
 187         catch(exception& e) {
 188                 m->errorOut(e, "LibShuffCommand", "execute");
 189                 exit(1);
 190         }
 191 }
 192
 193 //**********************************************************************************************************************
 194
 195 void LibShuffCommand::printCoverageFile() {
 196         try {
 197
 198                 ofstream outCov;
 199                 summaryFile = outputDir + getRootName(getSimpleName(globaldata->getPhylipFile())) + "libshuff.coverage";
 200                 openOutputFile(summaryFile, outCov);
 201                 outputNames.push_back(summaryFile);
 202                 outCov.setf(ios::fixed, ios::floatfield); outCov.setf(ios::showpoint);
 203                 //cout.setf(ios::fixed, ios::floatfield); cout.setf(ios::showpoint);
 204
 205                 map<double,vector<int> > allDistances;
 206                 map<double,vector<int> >::iterator it;
 207
 208                 vector<vector<int> > indices(numGroups);
 209                 int numIndices = numGroups * numGroups;
 210
 211                 int index = 0;
 212                 for(int i=0;i<numGroups;i++){
 213                         indices[i].assign(numGroups,0);
 214                         for(int j=0;j<numGroups;j++){
 215                                 indices[i][j] = index++;
 216
 217                                 int spoti = globaldata->gGroupmap->groupIndex[groupNames[i]]; //neccessary in case user selects groups so you know where they are in the matrix
 218                                 int spotj = globaldata->gGroupmap->groupIndex[groupNames[j]];
 219
 220                                 for(int k=0;k<savedMinValues[spoti][spotj].size();k++){
 221                                         if(allDistances[savedMinValues[spoti][spotj][k]].size() != 0){
 222                                                 allDistances[savedMinValues[spoti][spotj][k]][indices[i][j]]++;
 223                                         }
 224                                         else{
 225                                                 allDistances[savedMinValues[spoti][spotj][k]].assign(numIndices, 0);
 226                                                 allDistances[savedMinValues[spoti][spotj][k]][indices[i][j]] = 1;
 227                                         }
 228                                 }
 229                         }
 230                 }
 231                 it=allDistances.begin();
 232
 233                 //cout << setprecision(8);
 234
 235                 vector<int> prevRow = it->second;
 236                 it++;
 237
 238                 for(;it!=allDistances.end();it++){
 239                         for(int i=0;i<it->second.size();i++){
 240                                 it->second[i] += prevRow[i];
 241                         }
 242                         prevRow = it->second;
 243                 }
 244
 245                 vector<int> lastRow = allDistances.rbegin()->second;
 246                 outCov << setprecision(8);
 247
 248                 outCov << "dist";
 249                 for (int i = 0; i < numGroups; i++){
 250                         outCov << '\t' << groupNames[i];
 251                 }
 252                 for (int i=0;i<numGroups;i++){
 253                         for(int j=i+1;j<numGroups;j++){
 254                                 outCov << '\t' << groupNames[i] << '-' << groupNames[j] << '\t';
 255                                 outCov << groupNames[j] << '-' << groupNames[i];
 256                         }
 257                 }
 258                 outCov << endl;
 259
 260                 for(it=allDistances.begin();it!=allDistances.end();it++){
 261                         outCov << it->first << '\t';
 262                         for(int i=0;i<numGroups;i++){
 263                                 outCov << it->second[indices[i][i]]/(float)lastRow[indices[i][i]] << '\t';
 264                         }
 265                         for(int i=0;i<numGroups;i++){
 266                                 for(int j=i+1;j<numGroups;j++){
 267                                         outCov << it->second[indices[i][j]]/(float)lastRow[indices[i][j]] << '\t';
 268                                         outCov << it->second[indices[j][i]]/(float)lastRow[indices[j][i]] << '\t';
 269                                 }
 270                         }
 271                         outCov << endl;
 272                 }
 273                 outCov.close();
 274         }
 275         catch(exception& e) {
 276                 m->errorOut(e, "LibShuffCommand", "printCoverageFile");
 277                 exit(1);
 278         }
 279 }
 280
 281 //**********************************************************************************************************************
 282
 283 void LibShuffCommand::printSummaryFile() {
 284         try {
 285
 286                 ofstream outSum;
 287                 summaryFile = outputDir + getRootName(getSimpleName(globaldata->getPhylipFile())) + "libshuff.summary";
 288                 openOutputFile(summaryFile, outSum);
 289                 outputNames.push_back(summaryFile);
 290
 291                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 292                 cout.setf(ios::fixed, ios::floatfield); cout.setf(ios::showpoint);
 293
 294                 cout << setw(20) << left << "Comparison" << '\t' << setprecision(8) << "dCXYScore" << '\t' << "Significance" << endl;
 295                 m->mothurOutJustToLog("Comparison\tdCXYScore\tSignificance"); m->mothurOutEndLine();
 296                 outSum << setw(20) << left << "Comparison" << '\t' << setprecision(8) << "dCXYScore" << '\t' << "Significance" << endl;
 297
 298                 int precision = (int)log10(iters);
 299                 for(int i=0;i<numGroups;i++){
 300                         for(int j=i+1;j<numGroups;j++){
 301                                 int spoti = globaldata->gGroupmap->groupIndex[groupNames[i]]; //neccessary in case user selects groups so you know where they are in the matrix
 302                                 int spotj = globaldata->gGroupmap->groupIndex[groupNames[j]];
 303
 304                                 if(pValueCounts[i][j]){
 305                                         cout << setw(20) << left << groupNames[i]+'-'+groupNames[j] << '\t' << setprecision(8) << savedDXYValues[spoti][spotj] << '\t' << setprecision(precision) << pValueCounts[i][j]/(float)iters << endl;
 306                                         m->mothurOutJustToLog(groupNames[i]+"-"+groupNames[j] + "\t" + toString(savedDXYValues[spoti][spotj]) + "\t" + toString((pValueCounts[i][j]/(float)iters))); m->mothurOutEndLine();
 307                                         outSum << setw(20) << left << groupNames[i]+'-'+groupNames[j] << '\t' << setprecision(8) << savedDXYValues[spoti][spotj] << '\t' << setprecision(precision) << pValueCounts[i][j]/(float)iters << endl;
 308                                 }
 309                                 else{
 310                                         cout << setw(20) << left << groupNames[i]+'-'+groupNames[j] << '\t' << setprecision(8) << savedDXYValues[spoti][spotj] << '\t' << '<' <<setprecision(precision) << 1/(float)iters << endl;
 311                                         m->mothurOutJustToLog(groupNames[i]+"-"+groupNames[j] + "\t" + toString(savedDXYValues[spoti][spotj]) + "\t" + toString((1/(float)iters))); m->mothurOutEndLine();
 312                                         outSum << setw(20) << left << groupNames[i]+'-'+groupNames[j] << '\t' << setprecision(8) << savedDXYValues[spoti][spotj] << '\t' << '<' <<setprecision(precision) << 1/(float)iters << endl;
 313                                 }
 314                                 if(pValueCounts[j][i]){
 315                                         cout << setw(20) << left << groupNames[j]+'-'+groupNames[i] << '\t' << setprecision(8) << savedDXYValues[spotj][spoti] << '\t' << setprecision (precision) << pValueCounts[j][i]/(float)iters << endl;
 316                                         m->mothurOutJustToLog(groupNames[j]+"-"+groupNames[i] + "\t" + toString(savedDXYValues[spotj][spoti]) + "\t" + toString((pValueCounts[j][i]/(float)iters))); m->mothurOutEndLine();
 317                                         outSum << setw(20) << left << groupNames[j]+'-'+groupNames[i] << '\t' << setprecision(8) << savedDXYValues[spotj][spoti] << '\t' << setprecision (precision) << pValueCounts[j][i]/(float)iters << endl;
 318                                 }
 319                                 else{
 320                                         cout << setw(20) << left << groupNames[j]+'-'+groupNames[i] << '\t' << setprecision(8) << savedDXYValues[spotj][spoti] << '\t' << '<' <<setprecision (precision) << 1/(float)iters << endl;
 321                                         m->mothurOutJustToLog(groupNames[j]+"-"+groupNames[i] + "\t" + toString(savedDXYValues[spotj][spoti]) + "\t" + toString((1/(float)iters))); m->mothurOutEndLine();
 322                                         outSum << setw(20) << left << groupNames[j]+'-'+groupNames[i] << '\t' << setprecision(8) << savedDXYValues[spotj][spoti] << '\t' << '<' <<setprecision (precision) << 1/(float)iters << endl;
 323                                 }
 324                         }
 325                 }
 326
 327                 outSum.close();
 328         }
 329         catch(exception& e) {
 330                 m->errorOut(e, "LibShuffCommand", "printSummaryFile");
 331                 exit(1);
 332         }
 333 }
 334
 335 //**********************************************************************************************************************
 336
 337 void LibShuffCommand::setGroups() {
 338         try {
 339                 //if the user has not entered specific groups to analyze then do them all
 340                 if (globaldata->Groups.size() == 0) {
 341                         numGroups = globaldata->gGroupmap->getNumGroups();
 342                         for (int i=0; i < numGroups; i++) {
 343                                 globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
 344                         }
 345                 } else {
 346                         if (savegroups != "all") {
 347                                 //check that groups are valid
 348                                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 349                                         if (globaldata->gGroupmap->isValidGroup(globaldata->Groups[i]) != true) {
 350                                                 m->mothurOut(globaldata->Groups[i] + " is not a valid group, and will be disregarded."); m->mothurOutEndLine();
 351                                                 // erase the invalid group from globaldata->Groups
 352                                                 globaldata->Groups.erase(globaldata->Groups.begin()+i);
 353                                         }
 354                                 }
 355
 356                                 //if the user only entered invalid groups
 357                                 if ((globaldata->Groups.size() == 0) || (globaldata->Groups.size() == 1)) {
 358                                         numGroups = globaldata->gGroupmap->getNumGroups();
 359                                         for (int i=0; i < numGroups; i++) {
 360                                                 globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
 361                                         }
 362                                         m->mothurOut("When using the groups parameter you must have at least 2 valid groups. I will run the command using all the groups in your groupfile."); m->mothurOutEndLine();
 363                                 } else { numGroups = globaldata->Groups.size(); }
 364                         } else { //users wants all groups
 365                                 numGroups = globaldata->gGroupmap->getNumGroups();
 366                                 globaldata->Groups.clear();
 367                                 for (int i=0; i < numGroups; i++) {
 368                                         globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
 369                                 }
 370                         }
 371                 }
 372
 373                 //sort so labels match
 374                 sort(globaldata->Groups.begin(), globaldata->Groups.end());
 375
 376                 //sort
 377                 sort(globaldata->gGroupmap->namesOfGroups.begin(), globaldata->gGroupmap->namesOfGroups.end());
 378
 379                 for (int i = 0; i < globaldata->gGroupmap->namesOfGroups.size(); i++) {  globaldata->gGroupmap->groupIndex[globaldata->gGroupmap->namesOfGroups[i]] = i;  }
 380
 381                 groupNames = globaldata->Groups;
 382
 383         }
 384         catch(exception& e) {
 385                 m->errorOut(e, "LibShuffCommand", "setGroups");
 386                 exit(1);
 387         }
 388 }
 389
 390 /***********************************************************/