distancecommand.h

   1 #ifndef DISTANCECOMMAND_H
   2 #define DISTANCECOMMAND_H
   3
   4 /*
   5  *  distancecommand.h
   6  *  Mothur
   7  *
   8  *  Created by Sarah Westcott on 5/7/09.
   9  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
  10  *
  11  */
  12
  13 #include "mothur.h"
  14 #include "command.hpp"
  15 #include "validcalculator.h"
  16 #include "dist.h"
  17 #include "sequencedb.h"
  18 #include "ignoregaps.h"
  19 #include "eachgapdist.h"
  20 #include "eachgapignore.h"
  21 #include "onegapdist.h"
  22 #include "onegapignore.h"
  23
  24 //custom data structure for threads to use.
  25 // This is passed by void pointer so it can be any data type
  26 // that can be passed using a single void pointer (LPVOID).
  27 struct distanceData {
  28         int startLine;
  29         int endLine;
  30         string dFileName;
  31         float cutoff;
  32         SequenceDB alignDB;
  33         vector<string> Estimators;
  34         MothurOut* m;
  35         string output;
  36         int numNewFasta, count;
  37         string countends;
  38
  39         distanceData(){}
  40         distanceData(int s, int e, string dbname, float c, SequenceDB db, vector<string> Est, MothurOut* mout, string o, int num, string count) {
  41                 startLine = s;
  42                 endLine = e;
  43                 dFileName = dbname;
  44                 cutoff = c;
  45                 alignDB = db;
  46                 Estimators = Est;
  47                 m = mout;
  48                 output = o;
  49                 numNewFasta = num;
  50                 countends = count;
  51
  52         }
  53 };
  54
  55 /**************************************************************************************************/
  56 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
  57 #else
  58 static DWORD WINAPI MyDistThreadFunction(LPVOID lpParam){
  59         distanceData* pDataArray;
  60         pDataArray = (distanceData*)lpParam;
  61
  62         try {
  63                 ValidCalculators validCalculator;
  64                 Dist* distCalculator;
  65                 if (pDataArray->m->isTrue(pDataArray->countends) == true) {
  66                         for (int i=0; i<pDataArray->Estimators.size(); i++) {
  67                                 if (validCalculator.isValidCalculator("distance", pDataArray->Estimators[i]) == true) {
  68                                         if (pDataArray->Estimators[i] == "nogaps")                      {       distCalculator = new ignoreGaps();      }
  69                                         else if (pDataArray->Estimators[i] == "eachgap")        {       distCalculator = new eachGapDist();     }
  70                                         else if (pDataArray->Estimators[i] == "onegap")         {       distCalculator = new oneGapDist();      }
  71                                 }
  72                         }
  73                 }else {
  74                         for (int i=0; i<pDataArray->Estimators.size(); i++) {
  75                                 if (validCalculator.isValidCalculator("distance", pDataArray->Estimators[i]) == true) {
  76                                         if (pDataArray->Estimators[i] == "nogaps")              {       distCalculator = new ignoreGaps();                                      }
  77                                         else if (pDataArray->Estimators[i] == "eachgap"){       distCalculator = new eachGapIgnoreTermGapDist();        }
  78                                         else if (pDataArray->Estimators[i] == "onegap") {       distCalculator = new oneGapIgnoreTermGapDist();         }
  79                                 }
  80                         }
  81                 }
  82
  83                 int startTime = time(NULL);
  84
  85                 //column file
  86                 ofstream outFile(pDataArray->dFileName.c_str(), ios::trunc);
  87                 outFile.setf(ios::fixed, ios::showpoint);
  88                 outFile << setprecision(4);
  89                 pDataArray->count = 0;
  90
  91                 if (pDataArray->output != "square") {
  92                         if((pDataArray->output == "lt") && (pDataArray->startLine == 0)){       outFile << pDataArray->alignDB.getNumSeqs() << endl;    }
  93
  94                         for(int i=pDataArray->startLine;i<pDataArray->endLine;i++){
  95                                 if(pDataArray->output == "lt")  {
  96                                         string name = pDataArray->alignDB.get(i).getName();
  97                                         if (name.length() < 10) { //pad with spaces to make compatible
  98                                                 while (name.length() < 10) {  name += " ";  }
  99                                         }
 100                                         outFile << name << '\t';
 101                                 }
 102                                 for(int j=0;j<i;j++){
 103
 104                                         if (pDataArray->m->control_pressed) { delete distCalculator; outFile.close(); return 0;  }
 105
 106                                         //if there was a column file given and we are appending, we don't want to calculate the distances that are already in the column file
 107                                         //the alignDB contains the new sequences and then the old, so if i an oldsequence and j is an old sequence then break out of this loop
 108                                         if ((i >= pDataArray->numNewFasta) && (j >= pDataArray->numNewFasta)) { break; }
 109
 110                                         distCalculator->calcDist(pDataArray->alignDB.get(i), pDataArray->alignDB.get(j));
 111                                         double dist = distCalculator->getDist();
 112
 113                                         if(dist <= pDataArray->cutoff){
 114                                                 if (pDataArray->output == "column") { outFile << pDataArray->alignDB.get(i).getName() << ' ' << pDataArray->alignDB.get(j).getName() << ' ' << dist << endl; }
 115                                         }
 116                                         if (pDataArray->output == "lt") {  outFile << dist << '\t'; }
 117                                 }
 118
 119                                 if (pDataArray->output == "lt") { outFile << endl; }
 120
 121                                 if(i % 100 == 0){
 122                                         pDataArray->m->mothurOutJustToScreen(toString(i) + "\t" + toString(time(NULL) - startTime)+"\n");                               }
 123                                 pDataArray->count++;
 124                         }
 125                         pDataArray->m->mothurOutJustToScreen(toString(pDataArray->count) + "\t" + toString(time(NULL) - startTime)+"\n");
 126                 }else{
 127                         if(pDataArray->startLine == 0){ outFile << pDataArray->alignDB.getNumSeqs() << endl;    }
 128
 129                         for(int i=pDataArray->startLine;i<pDataArray->endLine;i++){
 130
 131                                 string name = pDataArray->alignDB.get(i).getName();
 132                                 //pad with spaces to make compatible
 133                                 if (name.length() < 10) { while (name.length() < 10) {  name += " ";  } }
 134
 135                                 outFile << name << '\t';
 136
 137                                 for(int j=0;j<pDataArray->alignDB.getNumSeqs();j++){
 138
 139                                         if (pDataArray->m->control_pressed) { delete distCalculator; outFile.close(); return 0;  }
 140
 141                                         distCalculator->calcDist(pDataArray->alignDB.get(i), pDataArray->alignDB.get(j));
 142                                         double dist = distCalculator->getDist();
 143
 144                                         outFile << dist << '\t';
 145                                 }
 146
 147                                 outFile << endl;
 148
 149                                 if(i % 100 == 0){
 150                                         pDataArray->m->mothurOutJustToScreen(toString(i) + "\t" + toString(time(NULL) - startTime)+"\n");
 151                                 }
 152                                 pDataArray->count++;
 153                         }
 154                         pDataArray->m->mothurOutJustToScreen(toString(pDataArray->count) + "\t" + toString(time(NULL) - startTime)+"\n");
 155                 }
 156
 157                 outFile.close();
 158                 delete distCalculator;
 159
 160                 return 0;
 161         }
 162         catch(exception& e) {
 163                 pDataArray->m->errorOut(e, "DistanceCommand", "MyDistThreadFunction");
 164                 exit(1);
 165         }
 166 }
 167 #endif
 168
 169 /**************************************************************************************************/
 170 class DistanceCommand : public Command {
 171
 172 public:
 173         DistanceCommand(string);
 174         DistanceCommand();
 175         ~DistanceCommand() {}
 176
 177         vector<string> setParameters();
 178         string getCommandName()                 { return "dist.seqs";                   }
 179         string getCommandCategory()             { return "Sequence Processing"; }
 180
 181         string getHelpString();
 182     string getOutputPattern(string);
 183         string getCitation() { return "Schloss PD (2010). The effects of alignment quality, distance calculation method, sequence filtering, and region on the analysis of 16S rRNA gene-based studies. PLoS Comput Biol 6: e1000844. \nhttp://www.mothur.org/wiki/Dist.seqs"; }
 184         string getDescription()         { return "calculate the pairwaise distances between aligned sequences"; }
 185
 186         int execute();
 187         void help() { m->mothurOut(getHelpString()); }
 188
 189
 190 private:
 191         struct distlinePair {
 192                 int start;
 193                 int end;
 194
 195         };
 196
 197         //Dist* distCalculator;
 198         SequenceDB alignDB;
 199
 200         string countends, output, fastafile, calc, outputDir, oldfastafile, column, compress;
 201
 202         int processors, numNewFasta;
 203         float cutoff;
 204         vector<int> processIDS;   //end line, processid
 205         vector<distlinePair> lines;
 206
 207         bool abort;
 208         vector<string>  Estimators, outputNames; //holds estimators to be used
 209
 210         //void m->appendFiles(string, string);
 211         void createProcesses(string);
 212         int driver(/*Dist*, SequenceDB, */int, int, string, float);
 213         int driver(int, int, string, string);
 214
 215         #ifdef USE_MPI
 216         int driverMPI(int, int, MPI_File&, float);
 217         int driverMPI(int, int, string, unsigned long long&);
 218         int driverMPI(int, int, string, unsigned long long&, string);
 219         #endif
 220
 221         //int convertMatrix(string);
 222         bool sanityCheck();
 223         //int convertToLowerTriangle(string);
 224
 225 };
 226
 227 #endif
 228
 229 /**************************************************************************************************/
 230
 231
 232