sharedutilities.cpp

   1 /*
   2  *  sharedutilities.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 4/9/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "sharedutilities.h"
  11 #include "sharedrabundvector.h"
  12 #include "sharedordervector.h"
  13
  14 /**************************************************************************************************/
  15
  16 void SharedUtil::getSharedVectors(vector<string> Groups, vector<SharedRAbundVector*>& lookup, SharedOrderVector* order) {
  17         try {
  18
  19                 //delete each sharedrabundvector in lookup
  20                 for (int j = 0; j < lookup.size(); j++) {
  21                         delete lookup[j];
  22                 }
  23
  24                 lookup.clear();
  25
  26                 //create and initialize vector of sharedvectors, one for each group
  27                 for (int i = 0; i < Groups.size(); i++) {
  28                         SharedRAbundVector* temp = new SharedRAbundVector(order->getNumBins());
  29                         temp->setLabel(order->getLabel());
  30                         temp->setGroup(Groups[i]);
  31                         lookup.push_back(temp);
  32                 }
  33
  34                 int numSeqs = order->size();
  35                 //sample all the members
  36                 for(int i=0;i<numSeqs;i++){
  37                         //get first sample
  38                         individual chosen = order->get(i);
  39                         int abundance;
  40
  41                         //set info for sharedvector in chosens group
  42                         for (int j = 0; j < lookup.size(); j++) {
  43                                 if (chosen.group == lookup[j]->getGroup()) {
  44                                          abundance = lookup[j]->getAbundance(chosen.bin);
  45                                          lookup[j]->set(chosen.bin, (abundance + 1), chosen.group);
  46                                          break;
  47                                 }
  48                         }
  49                 }
  50         }
  51         catch(exception& e) {
  52                 cout << "Standard Error: " << e.what() << " has occurred in the SharedUtil class Function getSharedVectors. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  53                 exit(1);
  54         }
  55         catch(...) {
  56                 cout << "An unknown error has occurred in the SharedUtil class function getSharedVectors. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  57                 exit(1);
  58         }
  59 }
  60 /**************************************************************************************************/
  61
  62 void SharedUtil::getSharedVectorswithReplacement(vector<string> Groups, vector<SharedRAbundVector*>& lookup, SharedOrderVector* order) {
  63         try {
  64
  65                 //delete each sharedrabundvector in lookup
  66                 for (int j = 0; j < lookup.size(); j++) {
  67                         delete lookup[j];
  68                 }
  69
  70                 lookup.clear();
  71
  72                 //create and initialize vector of sharedvectors, one for each group
  73                 for (int i = 0; i < Groups.size(); i++) {
  74                         SharedRAbundVector* temp = new SharedRAbundVector(order->getNumBins());
  75                         temp->setLabel(order->getLabel());
  76                         temp->setGroup(Groups[i]);
  77                         lookup.push_back(temp);
  78                 }
  79
  80                 int numSeqs = order->size();
  81
  82                 //sample all the members
  83                 for(int i=0;i<numSeqs;i++){
  84                         //get random number
  85                         int random = int((float)(i+1) * (float)(rand()) / ((float)RAND_MAX+1.0));
  86                         individual chosen = order->get(random);
  87
  88                         int abundance;
  89                         //set info for sharedvector in chosens group
  90                         for (int j = 0; j < lookup.size(); j++) {
  91                                 if (chosen.group == lookup[j]->getGroup()) {
  92                                          abundance = lookup[j]->getAbundance(chosen.bin);
  93                                          lookup[j]->set(chosen.bin, (abundance + 1), chosen.group);
  94                                          break;
  95                                 }
  96                         }
  97                 }
  98
  99         }
 100         catch(exception& e) {
 101                 cout << "Standard Error: " << e.what() << " has occurred in the SharedUtil class Function getSharedVectorswithReplacement. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 102                 exit(1);
 103         }
 104         catch(...) {
 105                 cout << "An unknown error has occurred in the SharedUtil class function getSharedVectorswithReplacement. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 106                 exit(1);
 107         }
 108 }
 109
 110 /**************************************************************************************************/
 111 //need to have mode because different commands require different number of valid groups
 112 void SharedUtil::setGroups(vector<string>& userGroups, vector<string>& allGroups, string mode) {
 113         try {
 114                 if (userGroups.size() != 0) {
 115                         if (userGroups[0] != "all") {
 116                                 //check that groups are valid
 117                                 for (int i = 0; i < userGroups.size(); i++) {
 118                                         if (isValidGroup(userGroups[i], allGroups) != true) {
 119                                                 cout << userGroups[i] << " is not a valid group, and will be disregarded." << endl;
 120                                                 // erase the invalid group from userGroups
 121                                                 userGroups.erase(userGroups.begin()+i);
 122                                                 i--;
 123                                         }
 124                                 }
 125
 126                         }else{//user has enter "all" and wants the default groups
 127                                 userGroups.clear();
 128                                 for (int i = 0; i < allGroups.size(); i++) {
 129                                         userGroups.push_back(allGroups[i]);
 130                                 }
 131                         }
 132                 }else { //the user has not entered groups
 133                         for (int i = 0; i < allGroups.size(); i++) {
 134                                 userGroups.push_back(allGroups[i]);
 135                         }
 136                 }
 137
 138                 if (mode == "venn")  {
 139                                 //if the user only entered invalid groups
 140                                 if (userGroups.size() == 0) {
 141                                         if (allGroups.size() > 4) {
 142                                                 cout << "When using the groups parameter you must have at least 1 valid group. I will run the command using the first four groups in your groupfile." << endl;
 143                                                 for (int i = 0; i < 4; i++) {
 144                                                         userGroups.push_back(allGroups[i]);
 145                                                 }
 146                                         }else {
 147                                                 cout << "When using the groups parameter you must have at least 1 valid group. I will run the command using all the groups in your groupfile." << endl;
 148                                                 for (int i = 0; i < allGroups.size(); i++) {
 149                                                         userGroups.push_back(allGroups[i]);
 150                                                 }
 151                                         }
 152                                 }
 153
 154                                 //check to make sure their are only 4 groups
 155                                 if (userGroups.size() > 4) {
 156                                         cout << "You may only use up to 4 groups at a time with this command.  I will choose the first four and disregard the rest." << endl;
 157                                         for (int i = 4; i < userGroups.size(); i++) {
 158                                                 userGroups.erase(userGroups.begin()+i);
 159                                         }
 160                                 }
 161                 }else if (mode == "heat") {
 162                                 //if the user only entered invalid groups
 163                                 if (userGroups.size() == 0) {
 164                                         cout << "When using the groups parameter you must have at least 1 valid groups. I will run the command using all the groups in your groupfile." << endl;
 165                                         for (int i = 0; i < allGroups.size(); i++) {
 166                                                 userGroups.push_back(allGroups[i]);
 167                                         }
 168                                 }
 169                 }else if ((mode == "collect") || (mode == "rarefact") || (mode == "summary") || (mode == "treegroup")) {
 170                                 //if the user only entered invalid groups
 171                                 if ((userGroups.size() == 0) || (userGroups.size() == 1)) {
 172                                         cout << "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." << endl;
 173                                         for (int i = 0; i < allGroups.size(); i++) {
 174                                                 userGroups.push_back(allGroups[i]);
 175                                         }
 176                                 }
 177                 }
 178
 179         }
 180         catch(exception& e) {
 181                 cout << "Standard Error: " << e.what() << " has occurred in the SharedUtil class Function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 182                 exit(1);
 183         }
 184         catch(...) {
 185                 cout << "An unknown error has occurred in the SharedUtil class function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 186                 exit(1);
 187         }
 188
 189 }
 190
 191 /**************************************************************************************/
 192 //for parsimony and unifrac commands you set pairwise groups as well as an allgroups in calc
 193 void SharedUtil::setGroups(vector<string>& userGroups, vector<string>& allGroups, string& label, int& numGroups, string mode){  //globaldata->Groups, your tree or group map, allgroups, mode
 194         try {
 195                 numGroups = 0;
 196                 label = "";
 197                 //if the user has not entered specific groups to analyze then do them all
 198                 if (userGroups.size() != 0) {
 199                         if (userGroups[0] != "all") {
 200                                 //check that groups are valid
 201                                 for (int i = 0; i < userGroups.size(); i++) {
 202                                         if (isValidGroup(userGroups[i], allGroups) != true) {
 203                                                 cout << userGroups[i] << " is not a valid group, and will be disregarded." << endl;
 204                                                 // erase the invalid group from globaldata->Groups
 205                                                 userGroups.erase(userGroups.begin()+i);
 206                                                 i--;
 207                                         }
 208                                 }
 209                         }else { //users wants all groups
 210                                 userGroups.clear();
 211                                 for (int i=0; i < allGroups.size(); i++) {
 212                                         if (allGroups[i] != "xxx") {
 213                                                 userGroups.push_back(allGroups[i]);
 214                                                 label += allGroups[i] + "-";
 215                                         }
 216                                 }
 217                         }
 218                 }else { //the user has not entered groups
 219                         for (int i=0; i < allGroups.size(); i++) {
 220                                 if (allGroups[i] != "xxx") {
 221                                         label += allGroups[i] + "-";
 222                                         if (mode == "weighted") {
 223                                                 userGroups.push_back(allGroups[i]);
 224                                         }else {
 225                                                 numGroups = 1;
 226                                         }
 227                                 }
 228                         }
 229                 }
 230
 231
 232                 if (mode == "weighted") {
 233                         //if the user only entered invalid groups
 234                         if (userGroups.size() == 0) {
 235                                 for (int i=0; i < allGroups.size(); i++) {
 236                                         if (allGroups[i] != "xxx") {
 237                                                 userGroups.push_back(allGroups[i]);
 238                                         }
 239                                 }
 240                                 cout << "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." << endl;
 241                         }else if (userGroups.size() == 1) {
 242                                 cout << "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." << endl;
 243                                 userGroups.clear();
 244                                 for (int i=0; i < allGroups.size(); i++) {
 245                                         if (allGroups[i] != "xxx") {
 246                                                 userGroups.push_back(allGroups[i]);
 247                                         }
 248                                 }
 249                         }
 250                         numGroups = userGroups.size();
 251
 252                 }else if ((mode == "unweighted") || (mode == "parsimony")) {
 253                                 //if the user only entered invalid groups
 254                                 if ((userGroups.size() == 0) && (numGroups == 0)) {
 255                                         cout << "When using the groups parameter you must have at least 1 valid group. I will run the command using all the groups in your groupfile." << endl;
 256                                         for (int i = 0; i < allGroups.size(); i++) {
 257                                                 if (allGroups[i] != "xxx") {
 258                                                         userGroups.push_back(allGroups[i]);
 259                                                         label += allGroups[i] + "-";
 260                                                 }
 261                                         }
 262                                 }else {
 263                                         for (int i = 0; i < userGroups.size(); i++) {
 264                                                 label += userGroups[i] + "-";
 265                                         }
 266                                 }
 267
 268                                 //rip extra - off allgroups
 269                                 label = label.substr(0, label.length()-1);
 270
 271                                 if (numGroups != 1) { numGroups = userGroups.size(); }
 272
 273                 }
 274         }
 275         catch(exception& e) {
 276                 cout << "Standard Error: " << e.what() << " has occurred in the SharedUtil class Function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 277                 exit(1);
 278         }
 279         catch(...) {
 280                 cout << "An unknown error has occurred in the SharedUtil class function setGroups. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 281                 exit(1);
 282         }
 283 }
 284 /**************************************************************************************/
 285 void SharedUtil::getCombos(vector<string>& groupComb, vector<string> userGroups, int& numComp) { //groupcomb, globaldata->Groups, numcomb
 286         try {
 287                 //calculate number of comparisons i.e. with groups A,B,C = AB, AC, BC = 3;
 288                 numComp = 0;
 289                 for (int i=0; i< userGroups.size(); i++) {
 290                         numComp += i;
 291                         for (int l = i+1; l < userGroups.size(); l++) {
 292                                 //set group comparison labels
 293                                 groupComb.push_back(userGroups[i] + "-" + userGroups[l]);
 294                         }
 295                 }
 296         }
 297         catch(exception& e) {
 298                 cout << "Standard Error: " << e.what() << " has occurred in the SharedUtil class Function getCombos. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 299                 exit(1);
 300         }
 301         catch(...) {
 302                 cout << "An unknown error has occurred in the SharedUtil class function getCombos. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 303                 exit(1);
 304         }
 305 }
 306 /**************************************************************************************/
 307 bool SharedUtil::isValidGroup(string groupname, vector<string> groups) {
 308         try {
 309                 for (int i = 0; i < groups.size(); i++) {
 310                         if (groupname == groups[i]) { return true; }
 311                 }
 312
 313                 return false;
 314         }
 315         catch(exception& e) {
 316                 cout << "Standard Error: " << e.what() << " has occurred in the SharedUtil class Function isValidGroup. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 317                 exit(1);
 318         }
 319         catch(...) {
 320                 cout << "An unknown error has occurred in the SharedUtil class function isValidGroup. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 321                 exit(1);
 322         }
 323 }
 324
 325 /**************************************************************************************/
 326 void SharedUtil::updateGroupIndex(vector<string>& userGroups, map<string, int>& index) {
 327         try {
 328                 index.clear();
 329                 for (int i = 0; i < userGroups.size(); i++) {
 330                         index[userGroups[i]] = i;
 331                 }
 332         }
 333         catch(exception& e) {
 334                 cout << "Standard Error: " << e.what() << " has occurred in the SharedUtil class Function updateGroupIndex. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 335                 exit(1);
 336         }
 337         catch(...) {
 338                 cout << "An unknown error has occurred in the SharedUtil class function updateGroupIndex. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 339                 exit(1);
 340         }
 341 }