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
  80
  81                                 if(userform == "discrete"){
  82                                         form = new DLibshuff(matrix, iters, step, cutOff);
  83                                 }
  84                                 else{
  85                                         form = new SLibshuff(matrix, iters, cutOff);
  86                                 }
  87                         }
  88
  89                 }
  90
  91         }
  92         catch(exception& e) {
  93                 errorOut(e, "LibShuffCommand", "LibShuffCommand");
  94                 exit(1);
  95         }
  96 }
  97 //**********************************************************************************************************************
  98
  99 void LibShuffCommand::help(){
 100         try {
 101                 mothurOut("The libshuff command can only be executed after a successful read.dist command including a groupfile.\n");
 102                 mothurOut("The libshuff command parameters are groups, iters, step, form and cutoff.  No parameters are required.\n");
 103                 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");
 104                 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");
 105                 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");
 106                 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");
 107                 mothurOut("The libshuff command should be in the following format: libshuff(groups=yourGroups, iters=yourIters, cutOff=yourCutOff, form=yourForm, step=yourStep).\n");
 108                 mothurOut("Example libshuff(groups=A-B-C, iters=500, form=discrete, step=0.01, cutOff=2.0).\n");
 109                 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");
 110                 mothurOut("The libshuff command output two files: .coverage and .slsummary their descriptions are in the manual.\n");
 111                 mothurOut("Note: No spaces between parameter labels (i.e. iters), '=' and parameters (i.e.yourIters).\n\n");
 112         }
 113         catch(exception& e) {
 114                 errorOut(e, "LibShuffCommand", "help");
 115                 exit(1);
 116         }
 117 }
 118
 119 //**********************************************************************************************************************
 120
 121 int LibShuffCommand::execute(){
 122         try {
 123
 124                 if (abort == true) {    return 0;       }
 125
 126                 savedDXYValues = form->evaluateAll();
 127                 savedMinValues = form->getSavedMins();
 128
 129                 pValueCounts.resize(numGroups);
 130                 for(int i=0;i<numGroups;i++){
 131                         pValueCounts[i].assign(numGroups, 0);
 132                 }
 133
 134                 Progress* reading = new Progress();
 135
 136                 for(int i=0;i<numGroups-1;i++) {
 137                         for(int j=i+1;j<numGroups;j++) {
 138                                 reading->newLine(groupNames[i]+'-'+groupNames[j], iters);
 139                                 for(int p=0;p<iters;p++) {
 140                                         form->randomizeGroups(i,j);
 141                                         if(form->evaluatePair(i,j) >= savedDXYValues[i][j])     {       pValueCounts[i][j]++;   }
 142                                         if(form->evaluatePair(j,i) >= savedDXYValues[j][i])     {       pValueCounts[j][i]++;   }
 143                                         reading->update(p);
 144                                 }
 145                                 form->resetGroup(i);
 146                                 form->resetGroup(j);
 147                         }
 148                 }
 149                 reading->finish();
 150                 delete reading;
 151
 152                 mothurOutEndLine();
 153                 printSummaryFile();
 154                 printCoverageFile();
 155
 156                 //clear out users groups
 157                 globaldata->Groups.clear();
 158                 delete form;
 159
 160                 //delete globaldata's copy of the gmatrix to free up memory
 161                 delete globaldata->gMatrix;  globaldata->gMatrix = NULL;
 162
 163                 return 0;
 164         }
 165         catch(exception& e) {
 166                 errorOut(e, "LibShuffCommand", "execute");
 167                 exit(1);
 168         }
 169 }
 170
 171 //**********************************************************************************************************************
 172
 173 void LibShuffCommand::printCoverageFile() {
 174         try {
 175
 176                 ofstream outCov;
 177                 summaryFile = getRootName(globaldata->getPhylipFile()) + "libshuff.coverage";
 178                 openOutputFile(summaryFile, outCov);
 179                 outCov.setf(ios::fixed, ios::floatfield); outCov.setf(ios::showpoint);
 180                 //cout.setf(ios::fixed, ios::floatfield); cout.setf(ios::showpoint);
 181
 182                 map<double,vector<int> > allDistances;
 183                 map<double,vector<int> >::iterator it;
 184
 185                 vector<vector<int> > indices(numGroups);
 186                 int numIndices = numGroups * numGroups;
 187
 188                 int index = 0;
 189                 for(int i=0;i<numGroups;i++){
 190                         indices[i].assign(numGroups,0);
 191                         for(int j=0;j<numGroups;j++){
 192                                 indices[i][j] = index++;
 193                                 for(int k=0;k<savedMinValues[i][j].size();k++){
 194                                         if(allDistances[savedMinValues[i][j][k]].size() != 0){
 195                                                 allDistances[savedMinValues[i][j][k]][indices[i][j]]++;
 196                                         }
 197                                         else{
 198                                                 allDistances[savedMinValues[i][j][k]].assign(numIndices, 0);
 199                                                 allDistances[savedMinValues[i][j][k]][indices[i][j]] = 1;
 200                                         }
 201                                 }
 202                         }
 203                 }
 204                 it=allDistances.begin();
 205
 206                 //cout << setprecision(8);
 207
 208                 vector<int> prevRow = it->second;
 209                 it++;
 210
 211                 for(it;it!=allDistances.end();it++){
 212                         for(int i=0;i<it->second.size();i++){
 213                                 it->second[i] += prevRow[i];
 214                         }
 215                         prevRow = it->second;
 216                 }
 217
 218                 vector<int> lastRow = allDistances.rbegin()->second;
 219                 outCov << setprecision(8);
 220
 221                 outCov << "dist";
 222                 for (int i = 0; i < numGroups; i++){
 223                         outCov << '\t' << groupNames[i];
 224                 }
 225                 for (int i=0;i<numGroups;i++){
 226                         for(int j=i+1;j<numGroups;j++){
 227                                 outCov << '\t' << groupNames[i] << '-' << groupNames[j] << '\t';
 228                                 outCov << groupNames[j] << '-' << groupNames[i];
 229                         }
 230                 }
 231                 outCov << endl;
 232
 233                 for(it=allDistances.begin();it!=allDistances.end();it++){
 234                         outCov << it->first << '\t';
 235                         for(int i=0;i<numGroups;i++){
 236                                 outCov << it->second[indices[i][i]]/(float)lastRow[indices[i][i]] << '\t';
 237                         }
 238                         for(int i=0;i<numGroups;i++){
 239                                 for(int j=i+1;j<numGroups;j++){
 240                                         outCov << it->second[indices[i][j]]/(float)lastRow[indices[i][j]] << '\t';
 241                                         outCov << it->second[indices[j][i]]/(float)lastRow[indices[j][i]] << '\t';
 242                                 }
 243                         }
 244                         outCov << endl;
 245                 }
 246                 outCov.close();
 247         }
 248         catch(exception& e) {
 249                 errorOut(e, "LibShuffCommand", "printCoverageFile");
 250                 exit(1);
 251         }
 252 }
 253
 254 //**********************************************************************************************************************
 255
 256 void LibShuffCommand::printSummaryFile() {
 257         try {
 258
 259                 ofstream outSum;
 260                 summaryFile = getRootName(globaldata->getPhylipFile()) + "libshuff.summary";
 261                 openOutputFile(summaryFile, outSum);
 262
 263                 outSum.setf(ios::fixed, ios::floatfield); outSum.setf(ios::showpoint);
 264                 cout.setf(ios::fixed, ios::floatfield); cout.setf(ios::showpoint);
 265
 266                 cout << setw(20) << left << "Comparison" << '\t' << setprecision(8) << "dCXYScore" << '\t' << "Significance" << endl;
 267                 mothurOutJustToLog("Comparison\tdCXYScore\tSignificance"); mothurOutEndLine();
 268                 outSum << setw(20) << left << "Comparison" << '\t' << setprecision(8) << "dCXYScore" << '\t' << "Significance" << endl;
 269
 270                 int precision = (int)log10(iters);
 271                 for(int i=0;i<numGroups;i++){
 272                         for(int j=i+1;j<numGroups;j++){
 273                                 if(pValueCounts[i][j]){
 274                                         cout << setw(20) << left << groupNames[i]+'-'+groupNames[j] << '\t' << setprecision(8) << savedDXYValues[i][j] << '\t' << setprecision(precision) << pValueCounts[i][j]/(float)iters << endl;
 275                                         mothurOutJustToLog(groupNames[i]+"-"+groupNames[j] + "\t" + toString(savedDXYValues[i][j]) + "\t" + toString((pValueCounts[i][j]/(float)iters))); mothurOutEndLine();
 276                                         outSum << setw(20) << left << groupNames[i]+'-'+groupNames[j] << '\t' << setprecision(8) << savedDXYValues[i][j] << '\t' << setprecision(precision) << pValueCounts[i][j]/(float)iters << endl;
 277                                 }
 278                                 else{
 279                                         cout << setw(20) << left << groupNames[i]+'-'+groupNames[j] << '\t' << setprecision(8) << savedDXYValues[i][j] << '\t' << '<' <<setprecision(precision) << 1/(float)iters << endl;
 280                                         mothurOutJustToLog(groupNames[i]+"-"+groupNames[j] + "\t" + toString(savedDXYValues[i][j]) + "\t" + toString((1/(float)iters))); mothurOutEndLine();
 281                                         outSum << setw(20) << left << groupNames[i]+'-'+groupNames[j] << '\t' << setprecision(8) << savedDXYValues[i][j] << '\t' << '<' <<setprecision(precision) << 1/(float)iters << endl;
 282                                 }
 283                                 if(pValueCounts[j][i]){
 284                                         cout << setw(20) << left << groupNames[j]+'-'+groupNames[i] << '\t' << setprecision(8) << savedDXYValues[j][i] << '\t' << setprecision (precision) << pValueCounts[j][i]/(float)iters << endl;
 285                                         mothurOutJustToLog(groupNames[j]+"-"+groupNames[i] + "\t" + toString(savedDXYValues[j][i]) + "\t" + toString((pValueCounts[j][i]/(float)iters))); mothurOutEndLine();
 286                                         outSum << setw(20) << left << groupNames[j]+'-'+groupNames[i] << '\t' << setprecision(8) << savedDXYValues[j][i] << '\t' << setprecision (precision) << pValueCounts[j][i]/(float)iters << endl;
 287                                 }
 288                                 else{
 289                                         cout << setw(20) << left << groupNames[j]+'-'+groupNames[i] << '\t' << setprecision(8) << savedDXYValues[j][i] << '\t' << '<' <<setprecision (precision) << 1/(float)iters << endl;
 290                                         mothurOutJustToLog(groupNames[j]+"-"+groupNames[i] + "\t" + toString(savedDXYValues[j][i]) + "\t" + toString((1/(float)iters))); mothurOutEndLine();
 291                                         outSum << setw(20) << left << groupNames[j]+'-'+groupNames[i] << '\t' << setprecision(8) << savedDXYValues[j][i] << '\t' << '<' <<setprecision (precision) << 1/(float)iters << endl;
 292                                 }
 293                         }
 294                 }
 295
 296                 outSum.close();
 297         }
 298         catch(exception& e) {
 299                 errorOut(e, "LibShuffCommand", "printSummaryFile");
 300                 exit(1);
 301         }
 302 }
 303
 304 //**********************************************************************************************************************
 305
 306 void LibShuffCommand::setGroups() {
 307         try {
 308                 //if the user has not entered specific groups to analyze then do them all
 309                 if (globaldata->Groups.size() == 0) {
 310                         numGroups = globaldata->gGroupmap->getNumGroups();
 311                         for (int i=0; i < numGroups; i++) {
 312                                 globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
 313                         }
 314                 } else {
 315                         if (savegroups != "all") {
 316                                 //check that groups are valid
 317                                 for (int i = 0; i < globaldata->Groups.size(); i++) {
 318                                         if (globaldata->gGroupmap->isValidGroup(globaldata->Groups[i]) != true) {
 319                                                 mothurOut(globaldata->Groups[i] + " is not a valid group, and will be disregarded."); mothurOutEndLine();
 320                                                 // erase the invalid group from globaldata->Groups
 321                                                 globaldata->Groups.erase(globaldata->Groups.begin()+i);
 322                                         }
 323                                 }
 324
 325                                 //if the user only entered invalid groups
 326                                 if ((globaldata->Groups.size() == 0) || (globaldata->Groups.size() == 1)) {
 327                                         numGroups = globaldata->gGroupmap->getNumGroups();
 328                                         for (int i=0; i < numGroups; i++) {
 329                                                 globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
 330                                         }
 331                                         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();
 332                                 } else { numGroups = globaldata->Groups.size(); }
 333                         } else { //users wants all groups
 334                                 numGroups = globaldata->gGroupmap->getNumGroups();
 335                                 globaldata->Groups.clear();
 336                                 for (int i=0; i < numGroups; i++) {
 337                                         globaldata->Groups.push_back(globaldata->gGroupmap->namesOfGroups[i]);
 338                                 }
 339                         }
 340                 }
 341
 342                 //sort so labels match
 343                 sort(globaldata->Groups.begin(), globaldata->Groups.end());
 344
 345                 //sort
 346                 sort(globaldata->gGroupmap->namesOfGroups.begin(), globaldata->gGroupmap->namesOfGroups.end());
 347
 348                 groupNames = globaldata->Groups;
 349
 350         }
 351         catch(exception& e) {
 352                 errorOut(e, "LibShuffCommand", "setGroups");
 353                 exit(1);
 354         }
 355 }
 356
 357 /***********************************************************/