parsimony.cpp

   1 /*
   2  *  parsimony.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 1/26/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "parsimony.h"
  11
  12 /**************************************************************************************************/
  13
  14 EstOutput Parsimony::getValues(Tree* t, int p, string o) {
  15         try {
  16                 globaldata = GlobalData::getInstance();
  17                 processors = p;
  18                 outputDir = o;
  19
  20                 //if the users enters no groups then give them the score of all groups
  21                 int numGroups = globaldata->Groups.size();
  22
  23                 //calculate number of comparsions
  24                 int numComp = 0;
  25                 vector< vector<string> > namesOfGroupCombos;
  26                 for (int r=0; r<numGroups; r++) {
  27                         for (int l = r+1; l < numGroups; l++) {
  28                                 numComp++;
  29                                 vector<string> groups; groups.push_back(globaldata->Groups[r]); groups.push_back(globaldata->Groups[l]);
  30                                 //cout << globaldata->Groups[r] << '\t' << globaldata->Groups[l] << endl;
  31                                 namesOfGroupCombos.push_back(groups);
  32                         }
  33                 }
  34
  35                 //numComp+1 for AB, AC, BC, ABC
  36                 if (numComp != 1) {
  37                         vector<string> groups;
  38                         if (numGroups == 0) {
  39                                 //get score for all users groups
  40                                 for (int i = 0; i < tmap->namesOfGroups.size(); i++) {
  41                                         if (tmap->namesOfGroups[i] != "xxx") {
  42                                                 groups.push_back(tmap->namesOfGroups[i]);
  43                                                 //cout << tmap->namesOfGroups[i] << endl;
  44                                         }
  45                                 }
  46                                 namesOfGroupCombos.push_back(groups);
  47                         }else {
  48                                 for (int i = 0; i < globaldata->Groups.size(); i++) {
  49                                         groups.push_back(globaldata->Groups[i]);
  50                                         //cout << globaldata->Groups[i] << endl;
  51                                 }
  52                                 namesOfGroupCombos.push_back(groups);
  53                         }
  54                 }
  55
  56         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
  57                 if(processors == 1){
  58                         data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size());
  59                 }else{
  60                         lines.clear();
  61                         int numPairs = namesOfGroupCombos.size();
  62
  63                         int numPairsPerProcessor = numPairs / processors;
  64
  65                         for (int i = 0; i < processors; i++) {
  66                                 int startPos = i * numPairsPerProcessor;
  67
  68                                 if(i == processors - 1){
  69                                         numPairsPerProcessor = numPairs - i * numPairsPerProcessor;
  70                                 }
  71
  72                                 lines.push_back(linePair(startPos, numPairsPerProcessor));
  73                         }
  74
  75                         data = createProcesses(t, namesOfGroupCombos);
  76                 }
  77         #else
  78                 data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size());
  79         #endif
  80
  81                 return data;
  82
  83         }
  84         catch(exception& e) {
  85                 m->errorOut(e, "Parsimony", "getValues");
  86                 exit(1);
  87         }
  88 }
  89 /**************************************************************************************************/
  90
  91 EstOutput Parsimony::createProcesses(Tree* t, vector< vector<string> > namesOfGroupCombos) {
  92         try {
  93 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
  94                 int process = 1;
  95                 vector<int> processIDS;
  96
  97                 EstOutput results;
  98
  99                 //loop through and create all the processes you want
 100                 while (process != processors) {
 101                         int pid = fork();
 102
 103                         if (pid > 0) {
 104                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 105                                 process++;
 106                         }else if (pid == 0){
 107                                 EstOutput myresults;
 108                                 myresults = driver(t, namesOfGroupCombos, lines[process].start, lines[process].num);
 109
 110                                 if (m->control_pressed) { exit(0); }
 111
 112                                 //pass numSeqs to parent
 113                                 ofstream out;
 114                                 string tempFile = outputDir + toString(getpid()) + ".pars.results.temp";
 115                                 m->openOutputFile(tempFile, out);
 116                                 out << myresults.size() << endl;
 117                                 for (int i = 0; i < myresults.size(); i++) {  out << myresults[i] << '\t';  } out << endl;
 118                                 out.close();
 119
 120                                 exit(0);
 121                         }else {
 122                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 123                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 124                                 exit(0);
 125                         }
 126                 }
 127
 128                 results = driver(t, namesOfGroupCombos, lines[0].start, lines[0].num);
 129
 130                 //force parent to wait until all the processes are done
 131                 for (int i=0;i<processIDS.size();i++) {
 132                         int temp = processIDS[i];
 133                         wait(&temp);
 134                 }
 135
 136                 if (m->control_pressed) { return results; }
 137
 138                 //get data created by processes
 139                 for (int i=0;i<processIDS.size();i++) {
 140                         ifstream in;
 141                         string s = outputDir + toString(processIDS[i]) + ".pars.results.temp";
 142                         m->openInputFile(s, in);
 143
 144                         //get scores
 145                         if (!in.eof()) {
 146                                 int num;
 147                                 in >> num; m->gobble(in);
 148
 149                                 if (m->control_pressed) { break; }
 150
 151                                 double w;
 152                                 for (int j = 0; j < num; j++) {
 153                                         in >> w;
 154                                         results.push_back(w);
 155                                 }
 156                                 m->gobble(in);
 157                         }
 158                         in.close();
 159                         remove(s.c_str());
 160                 }
 161
 162                 return results;
 163 #endif
 164         }
 165         catch(exception& e) {
 166                 m->errorOut(e, "Parsimony", "createProcesses");
 167                 exit(1);
 168         }
 169 }
 170 /**************************************************************************************************/
 171 EstOutput Parsimony::driver(Tree* t, vector< vector<string> > namesOfGroupCombos, int start, int num) {
 172         try {
 173
 174                 EstOutput results; results.resize(num);
 175
 176                 Tree* copyTree = new Tree();
 177                 int count = 0;
 178
 179                 for (int h = start; h < (start+num); h++) {
 180
 181                         if (m->control_pressed) { delete copyTree; return results; }
 182
 183                         int score = 0;
 184
 185                         //groups in this combo
 186                         vector<string> groups = namesOfGroupCombos[h];
 187
 188                         //copy users tree so that you can redo pgroups
 189                         copyTree->getCopy(t);
 190
 191                         //create pgroups that reflect the groups the user want to use
 192                         for(int i=copyTree->getNumLeaves();i<copyTree->getNumNodes();i++){
 193                                 copyTree->tree[i].pGroups = (copyTree->mergeUserGroups(i, groups));
 194                         }
 195
 196                         for(int i=copyTree->getNumLeaves();i<copyTree->getNumNodes();i++){
 197
 198                                 if (m->control_pressed) { return data; }
 199
 200                                 int lc = copyTree->tree[i].getLChild();
 201                                 int rc = copyTree->tree[i].getRChild();
 202
 203                                 int iSize = copyTree->tree[i].pGroups.size();
 204                                 int rcSize = copyTree->tree[rc].pGroups.size();
 205                                 int lcSize = copyTree->tree[lc].pGroups.size();
 206
 207                                 //if isize are 0 then that branch is to be ignored
 208                                 if (iSize == 0) { }
 209                                 else if ((rcSize == 0) || (lcSize == 0)) { }
 210                                 //if you have more groups than either of your kids then theres been a change.
 211                                 else if(iSize > rcSize || iSize > lcSize){
 212                                         score++;
 213                                 }
 214                         }
 215
 216                         results[count] = score;
 217                         count++;
 218                 }
 219
 220                 delete copyTree;
 221
 222                 return results;
 223         }
 224         catch(exception& e) {
 225                 m->errorOut(e, "Parsimony", "driver");
 226                 exit(1);
 227         }
 228 }
 229
 230 /**************************************************************************************************/
 231