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 77:3219. \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, countfile, 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     int createRabund(CountTable*& ct, ListVector*& list, RAbundVector*& rabund);
  66 };
  67
  68 /////////////////not working for Windows////////////////////////////////////////////////////////////
  69 // getting an access violation error.  This is most likely caused by the
  70 // threads stepping on eachother's structures, as I can run the thread function and the cluster fuction
  71 // in separately without errors occuring.  I suspect it may be in the use of the
  72 // static class mothurOut, but I can't pinpoint the problem.  All other objects are made new
  73 // within the thread.  MothurOut is used by almost all the classes in mothur, so if this was
  74 // really the cause I would expect to see all the windows threaded commands to have issues, but not
  75 // all do. So far, shhh.flows and trim.flows have similiar problems. Other thoughts, could it have
  76 // anything to do with mothur's use of copy constructors in many of our data structures. ie. listvector
  77 // is copied by nameassignment and passed to read which passes to the thread?  -westcott 2-8-12
  78 ////////////////////////////////////////////////////////////////////////////////////////////////////
  79 /**************************************************************************************************
  80 //custom data structure for threads to use.
  81 // This is passed by void pointer so it can be any data type
  82 // that can be passed using a single void pointer (LPVOID).
  83 struct clusterData {
  84         set<string> labels;
  85         vector < map<string, string> > distNames;
  86         string method;
  87     MothurOut* m;
  88         double cutoff, precision;
  89     string tag, outputDir;
  90     vector<string> listFiles;
  91     bool hard;
  92     int length, threadID;
  93
  94
  95         clusterData(){}
  96         clusterData(vector < map<string, string> > dv, MothurOut* mout, double cu, string me, string ou, bool hd, double pre, int len, int th) {
  97                 distNames = dv;
  98                 m = mout;
  99                 cutoff = cu;
 100         method = me;
 101                 outputDir = ou;
 102         hard = hd;
 103         precision = pre;
 104         length = len;
 105         threadID = th;
 106         }
 107 };
 108
 109 /**************************************************************************************************
 110 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 111 #else
 112 static DWORD WINAPI MyClusterThreadFunction(LPVOID lpParam){
 113         clusterData* pDataArray;
 114         pDataArray = (clusterData*)lpParam;
 115
 116         try {
 117                 cout << "starting " << endl;
 118
 119                 double smallestCutoff = pDataArray->cutoff;
 120
 121                 //cluster each distance file
 122                 for (int i = 0; i < pDataArray->distNames.size(); i++) {
 123
 124             Cluster* mycluster = NULL;
 125             SparseMatrix* mymatrix = NULL;
 126             ListVector* mylist = NULL;
 127             ListVector myoldList;
 128             RAbundVector* myrabund = NULL;
 129
 130                         if (pDataArray->m->control_pressed) { break; }
 131
 132                         string thisNamefile = pDataArray->distNames[i].begin()->second;
 133                         string thisDistFile = pDataArray->distNames[i].begin()->first;
 134             cout << thisNamefile << '\t' << thisDistFile << endl;
 135                         pDataArray->m->mothurOutEndLine(); pDataArray->m->mothurOut("Reading " + thisDistFile); pDataArray->m->mothurOutEndLine();
 136
 137                         ReadMatrix* myread = new ReadColumnMatrix(thisDistFile);
 138                         myread->setCutoff(pDataArray->cutoff);
 139                         NameAssignment* mynameMap = new NameAssignment(thisNamefile);
 140                         mynameMap->readMap();
 141             cout << "done reading " << thisNamefile << endl;
 142                         myread->read(mynameMap);
 143                         cout << "done reading " << thisDistFile << endl;
 144                         if (pDataArray->m->control_pressed) {  delete myread; delete mynameMap; break; }
 145
 146                         mylist = myread->getListVector();
 147                         myoldList = *mylist;
 148                         mymatrix = myread->getMatrix();
 149             cout << "here" << endl;
 150                         delete myread; myread = NULL;
 151                         delete mynameMap; mynameMap = NULL;
 152
 153             pDataArray->m->mothurOutEndLine(); pDataArray->m->mothurOut("Clustering " + thisDistFile); pDataArray->m->mothurOutEndLine();
 154
 155                         myrabund = new RAbundVector(mylist->getRAbundVector());
 156                          cout << "here" << endl;
 157                         //create cluster
 158                         if (pDataArray->method == "furthest")   {       mycluster = new CompleteLinkage(myrabund, mylist, mymatrix, pDataArray->cutoff, pDataArray->method); }
 159                         else if(pDataArray->method == "nearest"){       mycluster = new SingleLinkage(myrabund, mylist, mymatrix, pDataArray->cutoff, pDataArray->method); }
 160                         else if(pDataArray->method == "average"){       mycluster = new AverageLinkage(myrabund, mylist, mymatrix, pDataArray->cutoff, pDataArray->method);     }
 161                         pDataArray->tag = mycluster->getTag();
 162              cout << "here" << endl;
 163                         if (pDataArray->outputDir == "") { pDataArray->outputDir += pDataArray->m->hasPath(thisDistFile); }
 164                         string fileroot = pDataArray->outputDir + pDataArray->m->getRootName(pDataArray->m->getSimpleName(thisDistFile));
 165                          cout << "here" << endl;
 166                         ofstream listFile;
 167                         pDataArray->m->openOutputFile(fileroot+ pDataArray->tag + ".list",      listFile);
 168              cout << "here" << endl;
 169                         pDataArray->listFiles.push_back(fileroot+ pDataArray->tag + ".list");
 170
 171                         float previousDist = 0.00000;
 172                         float rndPreviousDist = 0.00000;
 173
 174                         myoldList = *mylist;
 175
 176                         bool print_start = true;
 177                         int start = time(NULL);
 178                         double saveCutoff = pDataArray->cutoff;
 179
 180                         while (mymatrix->getSmallDist() < pDataArray->cutoff && mymatrix->getNNodes() > 0){
 181
 182                                 if (pDataArray->m->control_pressed) { //clean up
 183                                         delete mymatrix; delete mylist; delete mycluster; delete myrabund;
 184                                         listFile.close();
 185                                         for (int i = 0; i < pDataArray->listFiles.size(); i++) {        pDataArray->m->mothurRemove(pDataArray->listFiles[i]);  }
 186                                         pDataArray->listFiles.clear(); break;
 187                                 }
 188
 189                                 mycluster->update(saveCutoff);
 190
 191                                 float dist = mymatrix->getSmallDist();
 192                                 float rndDist;
 193                                 if (pDataArray->hard) {
 194                                         rndDist = pDataArray->m->ceilDist(dist, pDataArray->precision);
 195                                 }else{
 196                                         rndDist = pDataArray->m->roundDist(dist, pDataArray->precision);
 197                                 }
 198
 199                                 if(previousDist <= 0.0000 && dist != previousDist){
 200                                         myoldList.setLabel("unique");
 201                                         myoldList.print(listFile);
 202                                         if (pDataArray->labels.count("unique") == 0) {  pDataArray->labels.insert("unique");  }
 203                                 }
 204                                 else if(rndDist != rndPreviousDist){
 205                                         myoldList.setLabel(toString(rndPreviousDist,  pDataArray->length-1));
 206                                         myoldList.print(listFile);
 207                                         if (pDataArray->labels.count(toString(rndPreviousDist,  pDataArray->length-1)) == 0) { pDataArray->labels.insert(toString(rndPreviousDist,  pDataArray->length-1)); }
 208                                 }
 209
 210                                 previousDist = dist;
 211                                 rndPreviousDist = rndDist;
 212                                 myoldList = *mylist;
 213                         }
 214
 215              cout << "here2" << endl;
 216                         if(previousDist <= 0.0000){
 217                                 myoldList.setLabel("unique");
 218                                 myoldList.print(listFile);
 219                                 if (pDataArray->labels.count("unique") == 0) { pDataArray->labels.insert("unique"); }
 220                         }
 221                         else if(rndPreviousDist<pDataArray->cutoff){
 222                                 myoldList.setLabel(toString(rndPreviousDist,  pDataArray->length-1));
 223                                 myoldList.print(listFile);
 224                                 if (pDataArray->labels.count(toString(rndPreviousDist,  pDataArray->length-1)) == 0) { pDataArray->labels.insert(toString(rndPreviousDist,  pDataArray->length-1)); }
 225                         }
 226
 227                         delete mymatrix; delete mylist; delete mycluster; delete myrabund;
 228             mymatrix = NULL; mylist = NULL; mycluster = NULL; myrabund = NULL;
 229                         listFile.close();
 230
 231                         if (pDataArray->m->control_pressed) { //clean up
 232                                 for (int i = 0; i < pDataArray->listFiles.size(); i++) {        pDataArray->m->mothurRemove(pDataArray->listFiles[i]);  }
 233                                 pDataArray->listFiles.clear(); break;
 234                         }
 235                          cout << "here3" << endl;
 236                         pDataArray->m->mothurRemove(thisDistFile);
 237                         pDataArray->m->mothurRemove(thisNamefile);
 238                          cout << "here4" << endl;
 239                         if (saveCutoff != pDataArray->cutoff) {
 240                                 if (pDataArray->hard)   {  saveCutoff = pDataArray->m->ceilDist(saveCutoff, pDataArray->precision);     }
 241                                 else            {       saveCutoff = pDataArray->m->roundDist(saveCutoff, pDataArray->precision);  }
 242
 243                                 pDataArray->m->mothurOut("Cutoff was " + toString(pDataArray->cutoff) + " changed cutoff to " + toString(saveCutoff)); pDataArray->m->mothurOutEndLine();
 244                         }
 245                          cout << "here5" << endl;
 246                         if (saveCutoff < smallestCutoff) { smallestCutoff = saveCutoff;  }
 247                 }
 248
 249                 pDataArray->cutoff = smallestCutoff;
 250
 251                 return 0;
 252
 253         }
 254         catch(exception& e) {
 255                 pDataArray->m->errorOut(e, "ClusterSplitCommand", "MyClusterThreadFunction");
 256                 exit(1);
 257         }
 258 }
 259 #endif
 260
 261 */
 262
 263
 264 #endif
 265