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