clustersplitcommand.h

   1 #ifndef CLUSTERSPLITCOMMAND_H
   2 #define CLUSTERSPLITCOMMAND_H
   3
   4 /*
   5  *  clustersplitcommand.h
   6  *  Mothur
   7  *
   8  *  Created by westcott on 5/19/10.
   9  *  Copyright 2010 Schloss Lab. All rights reserved.
  10  *
  11  */
  12
  13 #include "command.hpp"
  14 #include "rabundvector.hpp"
  15 #include "sabundvector.hpp"
  16 #include "listvector.hpp"
  17 #include "cluster.hpp"
  18 #include "sparsedistancematrix.h"
  19 #include "readcluster.h"
  20 #include "splitmatrix.h"
  21 #include "readphylip.h"
  22 #include "readcolumn.h"
  23 #include "readmatrix.hpp"
  24 #include "inputdata.h"
  25 #include "clustercommand.h"
  26 #include "clusterclassic.h"
  27
  28 class ClusterSplitCommand : public Command {
  29
  30 public:
  31         ClusterSplitCommand(string);
  32         ClusterSplitCommand();
  33         ~ClusterSplitCommand() {}
  34
  35         vector<string> setParameters();
  36         string getCommandName()                 { return "cluster.split";               }
  37         string getCommandCategory()             { return "Clustering";                  }
  38         string getOutputFileNameTag(string, string);
  39         string getHelpString();
  40         string getCitation() { return "Schloss PD, Westcott SL (2011). Assessing and improving methods used in OTU-based approaches for 16S rRNA gene sequence analysis. Appl Environ Microbiol. \nhttp://www.mothur.org/wiki/Cluster.split"; }
  41         string getDescription()         { return "splits your sequences by distance or taxonomy then clusters into OTUs"; }
  42
  43         int execute();
  44         void help() { m->mothurOut(getHelpString()); }
  45
  46 private:
  47         vector<int> processIDS;   //processid
  48         vector<string> outputNames;
  49
  50         string method, fileroot, tag, outputDir, phylipfile, columnfile, namefile, distfile, format, showabund, timing, splitmethod, taxFile, fastafile;
  51         double cutoff, splitcutoff;
  52         int precision, length, processors, taxLevelCutoff;
  53         bool print_start, abort, hard, large, classic;
  54         time_t start;
  55         ofstream outList, outRabund, outSabund;
  56
  57         void printData(ListVector*);
  58         vector<string> createProcesses(vector< map<string, string> >, set<string>&);
  59         vector<string> cluster(vector< map<string, string> >, set<string>&);
  60     string clusterFile(string, string, set<string>&, double&);
  61     string clusterClassicFile(string, string, set<string>&, double&);
  62         int mergeLists(vector<string>, map<float, int>, ListVector*);
  63         map<float, int> completeListFile(vector<string>, string, set<string>&, ListVector*&);
  64         int createMergedDistanceFile(vector< map<string, string> >);
  65 };
  66
  67 /////////////////not working for Windows////////////////////////////////////////////////////////////
  68 // getting an access violation error.  This is most likely caused by the
  69 // threads stepping on eachother's structures, as I can run the thread function and the cluster fuction
  70 // in separately without errors occuring.  I suspect it may be in the use of the
  71 // static class mothurOut, but I can't pinpoint the problem.  All other objects are made new
  72 // within the thread.  MothurOut is used by almost all the classes in mothur, so if this was
  73 // really the cause I would expect to see all the windows threaded commands to have issues, but not
  74 // all do. So far, shhh.flows and trim.flows have similiar problems. Other thoughts, could it have
  75 // anything to do with mothur's use of copy constructors in many of our data structures. ie. listvector
  76 // is copied by nameassignment and passed to read which passes to the thread?  -westcott 2-8-12
  77 ////////////////////////////////////////////////////////////////////////////////////////////////////
  78 /**************************************************************************************************/
  79 //custom data structure for threads to use.
  80 // This is passed by void pointer so it can be any data type
  81 // that can be passed using a single void pointer (LPVOID).
  82 struct clusterData {
  83         set<string> labels;
  84         vector < map<string, string> > distNames;
  85         string method;
  86     MothurOut* m;
  87         double cutoff, precision;
  88     string tag, outputDir;
  89     vector<string> listFiles;
  90     bool hard;
  91     int length, threadID;
  92
  93
  94         clusterData(){}
  95         clusterData(vector < map<string, string> > dv, MothurOut* mout, double cu, string me, string ou, bool hd, double pre, int len, int th) {
  96                 distNames = dv;
  97                 m = mout;
  98                 cutoff = cu;
  99         method = me;
 100                 outputDir = ou;
 101         hard = hd;
 102         precision = pre;
 103         length = len;
 104         threadID = th;
 105         }
 106 };
 107
 108 /**************************************************************************************************/
 109 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 110 #else
 111 static DWORD WINAPI MyClusterThreadFunction(LPVOID lpParam){
 112         clusterData* pDataArray;
 113         pDataArray = (clusterData*)lpParam;
 114
 115         try {
 116                 cout << "starting " << endl;
 117
 118                 double smallestCutoff = pDataArray->cutoff;
 119
 120                 //cluster each distance file
 121                 for (int i = 0; i < pDataArray->distNames.size(); i++) {
 122
 123             Cluster* mycluster = NULL;
 124             SparseMatrix* mymatrix = NULL;
 125             ListVector* mylist = NULL;
 126             ListVector myoldList;
 127             RAbundVector* myrabund = NULL;
 128
 129                         if (pDataArray->m->control_pressed) { break; }
 130
 131                         string thisNamefile = pDataArray->distNames[i].begin()->second;
 132                         string thisDistFile = pDataArray->distNames[i].begin()->first;
 133             cout << thisNamefile << '\t' << thisDistFile << endl;
 134                         pDataArray->m->mothurOutEndLine(); pDataArray->m->mothurOut("Reading " + thisDistFile); pDataArray->m->mothurOutEndLine();
 135
 136                         ReadMatrix* myread = new ReadColumnMatrix(thisDistFile);
 137                         myread->setCutoff(pDataArray->cutoff);
 138                         NameAssignment* mynameMap = new NameAssignment(thisNamefile);
 139                         mynameMap->readMap();
 140             cout << "done reading " << thisNamefile << endl;
 141                         myread->read(mynameMap);
 142                         cout << "done reading " << thisDistFile << endl;
 143                         if (pDataArray->m->control_pressed) {  delete myread; delete mynameMap; break; }
 144
 145                         mylist = myread->getListVector();
 146                         myoldList = *mylist;
 147                         mymatrix = myread->getMatrix();
 148             cout << "here" << endl;
 149                         delete myread; myread = NULL;
 150                         delete mynameMap; mynameMap = NULL;
 151
 152             pDataArray->m->mothurOutEndLine(); pDataArray->m->mothurOut("Clustering " + thisDistFile); pDataArray->m->mothurOutEndLine();
 153
 154                         myrabund = new RAbundVector(mylist->getRAbundVector());
 155                          cout << "here" << endl;
 156                         //create cluster
 157                         if (pDataArray->method == "furthest")   {       mycluster = new CompleteLinkage(myrabund, mylist, mymatrix, pDataArray->cutoff, pDataArray->method); }
 158                         else if(pDataArray->method == "nearest"){       mycluster = new SingleLinkage(myrabund, mylist, mymatrix, pDataArray->cutoff, pDataArray->method); }
 159                         else if(pDataArray->method == "average"){       mycluster = new AverageLinkage(myrabund, mylist, mymatrix, pDataArray->cutoff, pDataArray->method);     }
 160                         pDataArray->tag = mycluster->getTag();
 161              cout << "here" << endl;
 162                         if (pDataArray->outputDir == "") { pDataArray->outputDir += pDataArray->m->hasPath(thisDistFile); }
 163                         string fileroot = pDataArray->outputDir + pDataArray->m->getRootName(pDataArray->m->getSimpleName(thisDistFile));
 164                          cout << "here" << endl;
 165                         ofstream listFile;
 166                         pDataArray->m->openOutputFile(fileroot+ pDataArray->tag + ".list",      listFile);
 167              cout << "here" << endl;
 168                         pDataArray->listFiles.push_back(fileroot+ pDataArray->tag + ".list");
 169
 170                         float previousDist = 0.00000;
 171                         float rndPreviousDist = 0.00000;
 172
 173                         myoldList = *mylist;
 174
 175                         bool print_start = true;
 176                         int start = time(NULL);
 177                         double saveCutoff = pDataArray->cutoff;
 178
 179                         while (mymatrix->getSmallDist() < pDataArray->cutoff && mymatrix->getNNodes() > 0){
 180
 181                                 if (pDataArray->m->control_pressed) { //clean up
 182                                         delete mymatrix; delete mylist; delete mycluster; delete myrabund;
 183                                         listFile.close();
 184                                         for (int i = 0; i < pDataArray->listFiles.size(); i++) {        pDataArray->m->mothurRemove(pDataArray->listFiles[i]);  }
 185                                         pDataArray->listFiles.clear(); break;
 186                                 }
 187
 188                                 mycluster->update(saveCutoff);
 189
 190                                 float dist = mymatrix->getSmallDist();
 191                                 float rndDist;
 192                                 if (pDataArray->hard) {
 193                                         rndDist = pDataArray->m->ceilDist(dist, pDataArray->precision);
 194                                 }else{
 195                                         rndDist = pDataArray->m->roundDist(dist, pDataArray->precision);
 196                                 }
 197
 198                                 if(previousDist <= 0.0000 && dist != previousDist){
 199                                         myoldList.setLabel("unique");
 200                                         myoldList.print(listFile);
 201                                         if (pDataArray->labels.count("unique") == 0) {  pDataArray->labels.insert("unique");  }
 202                                 }
 203                                 else if(rndDist != rndPreviousDist){
 204                                         myoldList.setLabel(toString(rndPreviousDist,  pDataArray->length-1));
 205                                         myoldList.print(listFile);
 206                                         if (pDataArray->labels.count(toString(rndPreviousDist,  pDataArray->length-1)) == 0) { pDataArray->labels.insert(toString(rndPreviousDist,  pDataArray->length-1)); }
 207                                 }
 208
 209                                 previousDist = dist;
 210                                 rndPreviousDist = rndDist;
 211                                 myoldList = *mylist;
 212                         }
 213
 214              cout << "here2" << endl;
 215                         if(previousDist <= 0.0000){
 216                                 myoldList.setLabel("unique");
 217                                 myoldList.print(listFile);
 218                                 if (pDataArray->labels.count("unique") == 0) { pDataArray->labels.insert("unique"); }
 219                         }
 220                         else if(rndPreviousDist<pDataArray->cutoff){
 221                                 myoldList.setLabel(toString(rndPreviousDist,  pDataArray->length-1));
 222                                 myoldList.print(listFile);
 223                                 if (pDataArray->labels.count(toString(rndPreviousDist,  pDataArray->length-1)) == 0) { pDataArray->labels.insert(toString(rndPreviousDist,  pDataArray->length-1)); }
 224                         }
 225
 226                         delete mymatrix; delete mylist; delete mycluster; delete myrabund;
 227             mymatrix = NULL; mylist = NULL; mycluster = NULL; myrabund = NULL;
 228                         listFile.close();
 229
 230                         if (pDataArray->m->control_pressed) { //clean up
 231                                 for (int i = 0; i < pDataArray->listFiles.size(); i++) {        pDataArray->m->mothurRemove(pDataArray->listFiles[i]);  }
 232                                 pDataArray->listFiles.clear(); break;
 233                         }
 234                          cout << "here3" << endl;
 235                         pDataArray->m->mothurRemove(thisDistFile);
 236                         pDataArray->m->mothurRemove(thisNamefile);
 237                          cout << "here4" << endl;
 238                         if (saveCutoff != pDataArray->cutoff) {
 239                                 if (pDataArray->hard)   {  saveCutoff = pDataArray->m->ceilDist(saveCutoff, pDataArray->precision);     }
 240                                 else            {       saveCutoff = pDataArray->m->roundDist(saveCutoff, pDataArray->precision);  }
 241
 242                                 pDataArray->m->mothurOut("Cutoff was " + toString(pDataArray->cutoff) + " changed cutoff to " + toString(saveCutoff)); pDataArray->m->mothurOutEndLine();
 243                         }
 244                          cout << "here5" << endl;
 245                         if (saveCutoff < smallestCutoff) { smallestCutoff = saveCutoff;  }
 246                 }
 247
 248                 pDataArray->cutoff = smallestCutoff;
 249
 250                 return 0;
 251
 252         }
 253         catch(exception& e) {
 254                 pDataArray->m->errorOut(e, "ClusterSplitCommand", "MyClusterThreadFunction");
 255                 exit(1);
 256         }
 257 }
 258 #endif
 259
 260
 261
 262
 263 #endif
 264