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