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
  39         string getHelpString();
  40     string getOutputPattern(string);
  41         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"; }
  42         string getDescription()         { return "splits your sequences by distance or taxonomy then clusters into OTUs"; }
  43
  44         int execute();
  45         void help() { m->mothurOut(getHelpString()); }
  46
  47 private:
  48         vector<int> processIDS;   //processid
  49         vector<string> outputNames;
  50
  51         string method, fileroot, tag, outputDir, phylipfile, columnfile, namefile, countfile, distfile, format, showabund, timing, splitmethod, taxFile, fastafile;
  52         double cutoff, splitcutoff;
  53         int precision, length, processors, taxLevelCutoff;
  54         bool print_start, abort, hard, large, classic, runCluster;
  55         time_t start;
  56         ofstream outList, outRabund, outSabund;
  57
  58         void printData(ListVector*);
  59         vector<string> createProcesses(vector< map<string, string> >, set<string>&);
  60         vector<string> cluster(vector< map<string, string> >, set<string>&);
  61     string clusterFile(string, string, set<string>&, double&);
  62     string clusterClassicFile(string, string, set<string>&, double&);
  63         int mergeLists(vector<string>, map<float, int>, ListVector*);
  64         map<float, int> completeListFile(vector<string>, string, set<string>&, ListVector*&);
  65         int createMergedDistanceFile(vector< map<string, string> >);
  66     int createRabund(CountTable*& ct, ListVector*& list, RAbundVector*& rabund);
  67 };
  68
  69 /////////////////not working for Windows////////////////////////////////////////////////////////////
  70 // getting an access violation error.  This is most likely caused by the
  71 // threads stepping on eachother's structures, as I can run the thread function and the cluster fuction
  72 // in separately without errors occuring.  I suspect it may be in the use of the
  73 // static class mothurOut, but I can't pinpoint the problem.  All other objects are made new
  74 // within the thread.  MothurOut is used by almost all the classes in mothur, so if this was
  75 // really the cause I would expect to see all the windows threaded commands to have issues, but not
  76 // all do. So far, shhh.flows and trim.flows have similiar problems. Other thoughts, could it have
  77 // anything to do with mothur's use of copy constructors in many of our data structures. ie. listvector
  78 // is copied by nameassignment and passed to read which passes to the thread?  -westcott 2-8-12
  79 ////////////////////////////////////////////////////////////////////////////////////////////////////
  80 /**************************************************************************************************
  81 //custom data structure for threads to use.
  82 // This is passed by void pointer so it can be any data type
  83 // that can be passed using a single void pointer (LPVOID).
  84 struct clusterData {
  85         set<string> labels;
  86         vector < map<string, string> > distNames;
  87         string method;
  88     MothurOut* m;
  89         double cutoff, precision;
  90     string tag, outputDir;
  91     vector<string> listFiles;
  92     bool hard;
  93     int length, threadID;
  94
  95
  96         clusterData(){}
  97         clusterData(vector < map<string, string> > dv, MothurOut* mout, double cu, string me, string ou, bool hd, double pre, int len, int th) {
  98                 distNames = dv;
  99                 m = mout;
 100                 cutoff = cu;
 101         method = me;
 102                 outputDir = ou;
 103         hard = hd;
 104         precision = pre;
 105         length = len;
 106         threadID = th;
 107         }
 108 };
 109
 110 /**************************************************************************************************
 111 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 112 #else
 113 static DWORD WINAPI MyClusterThreadFunction(LPVOID lpParam){
 114         clusterData* pDataArray;
 115         pDataArray = (clusterData*)lpParam;
 116
 117         try {
 118                 cout << "starting " << endl;
 119
 120                 double smallestCutoff = pDataArray->cutoff;
 121
 122                 //cluster each distance file
 123                 for (int i = 0; i < pDataArray->distNames.size(); i++) {
 124
 125             Cluster* mycluster = NULL;
 126             SparseMatrix* mymatrix = NULL;
 127             ListVector* mylist = NULL;
 128             ListVector myoldList;
 129             RAbundVector* myrabund = NULL;
 130
 131                         if (pDataArray->m->control_pressed) { break; }
 132
 133                         string thisNamefile = pDataArray->distNames[i].begin()->second;
 134                         string thisDistFile = pDataArray->distNames[i].begin()->first;
 135             cout << thisNamefile << '\t' << thisDistFile << endl;
 136                         pDataArray->m->mothurOutEndLine(); pDataArray->m->mothurOut("Reading " + thisDistFile); pDataArray->m->mothurOutEndLine();
 137
 138                         ReadMatrix* myread = new ReadColumnMatrix(thisDistFile);
 139                         myread->setCutoff(pDataArray->cutoff);
 140                         NameAssignment* mynameMap = new NameAssignment(thisNamefile);
 141                         mynameMap->readMap();
 142             cout << "done reading " << thisNamefile << endl;
 143                         myread->read(mynameMap);
 144                         cout << "done reading " << thisDistFile << endl;
 145                         if (pDataArray->m->control_pressed) {  delete myread; delete mynameMap; break; }
 146
 147                         mylist = myread->getListVector();
 148                         myoldList = *mylist;
 149                         mymatrix = myread->getMatrix();
 150             cout << "here" << endl;
 151                         delete myread; myread = NULL;
 152                         delete mynameMap; mynameMap = NULL;
 153
 154             pDataArray->m->mothurOutEndLine(); pDataArray->m->mothurOut("Clustering " + thisDistFile); pDataArray->m->mothurOutEndLine();
 155
 156                         myrabund = new RAbundVector(mylist->getRAbundVector());
 157                          cout << "here" << endl;
 158                         //create cluster
 159                         if (pDataArray->method == "furthest")   {       mycluster = new CompleteLinkage(myrabund, mylist, mymatrix, pDataArray->cutoff, pDataArray->method); }
 160                         else if(pDataArray->method == "nearest"){       mycluster = new SingleLinkage(myrabund, mylist, mymatrix, pDataArray->cutoff, pDataArray->method); }
 161                         else if(pDataArray->method == "average"){       mycluster = new AverageLinkage(myrabund, mylist, mymatrix, pDataArray->cutoff, pDataArray->method);     }
 162                         pDataArray->tag = mycluster->getTag();
 163              cout << "here" << endl;
 164                         if (pDataArray->outputDir == "") { pDataArray->outputDir += pDataArray->m->hasPath(thisDistFile); }
 165                         string fileroot = pDataArray->outputDir + pDataArray->m->getRootName(pDataArray->m->getSimpleName(thisDistFile));
 166                          cout << "here" << endl;
 167                         ofstream listFile;
 168                         pDataArray->m->openOutputFile(fileroot+ pDataArray->tag + ".list",      listFile);
 169              cout << "here" << endl;
 170                         pDataArray->listFiles.push_back(fileroot+ pDataArray->tag + ".list");
 171
 172                         float previousDist = 0.00000;
 173                         float rndPreviousDist = 0.00000;
 174
 175                         myoldList = *mylist;
 176
 177                         bool print_start = true;
 178                         int start = time(NULL);
 179                         double saveCutoff = pDataArray->cutoff;
 180
 181                         while (mymatrix->getSmallDist() < pDataArray->cutoff && mymatrix->getNNodes() > 0){
 182
 183                                 if (pDataArray->m->control_pressed) { //clean up
 184                                         delete mymatrix; delete mylist; delete mycluster; delete myrabund;
 185                                         listFile.close();
 186                                         for (int i = 0; i < pDataArray->listFiles.size(); i++) {        pDataArray->m->mothurRemove(pDataArray->listFiles[i]);  }
 187                                         pDataArray->listFiles.clear(); break;
 188                                 }
 189
 190                                 mycluster->update(saveCutoff);
 191
 192                                 float dist = mymatrix->getSmallDist();
 193                                 float rndDist;
 194                                 if (pDataArray->hard) {
 195                                         rndDist = pDataArray->m->ceilDist(dist, pDataArray->precision);
 196                                 }else{
 197                                         rndDist = pDataArray->m->roundDist(dist, pDataArray->precision);
 198                                 }
 199
 200                                 if(previousDist <= 0.0000 && dist != previousDist){
 201                                         myoldList.setLabel("unique");
 202                                         myoldList.print(listFile);
 203                                         if (pDataArray->labels.count("unique") == 0) {  pDataArray->labels.insert("unique");  }
 204                                 }
 205                                 else if(rndDist != rndPreviousDist){
 206                                         myoldList.setLabel(toString(rndPreviousDist,  pDataArray->length-1));
 207                                         myoldList.print(listFile);
 208                                         if (pDataArray->labels.count(toString(rndPreviousDist,  pDataArray->length-1)) == 0) { pDataArray->labels.insert(toString(rndPreviousDist,  pDataArray->length-1)); }
 209                                 }
 210
 211                                 previousDist = dist;
 212                                 rndPreviousDist = rndDist;
 213                                 myoldList = *mylist;
 214                         }
 215
 216              cout << "here2" << endl;
 217                         if(previousDist <= 0.0000){
 218                                 myoldList.setLabel("unique");
 219                                 myoldList.print(listFile);
 220                                 if (pDataArray->labels.count("unique") == 0) { pDataArray->labels.insert("unique"); }
 221                         }
 222                         else if(rndPreviousDist<pDataArray->cutoff){
 223                                 myoldList.setLabel(toString(rndPreviousDist,  pDataArray->length-1));
 224                                 myoldList.print(listFile);
 225                                 if (pDataArray->labels.count(toString(rndPreviousDist,  pDataArray->length-1)) == 0) { pDataArray->labels.insert(toString(rndPreviousDist,  pDataArray->length-1)); }
 226                         }
 227
 228                         delete mymatrix; delete mylist; delete mycluster; delete myrabund;
 229             mymatrix = NULL; mylist = NULL; mycluster = NULL; myrabund = NULL;
 230                         listFile.close();
 231
 232                         if (pDataArray->m->control_pressed) { //clean up
 233                                 for (int i = 0; i < pDataArray->listFiles.size(); i++) {        pDataArray->m->mothurRemove(pDataArray->listFiles[i]);  }
 234                                 pDataArray->listFiles.clear(); break;
 235                         }
 236                          cout << "here3" << endl;
 237                         pDataArray->m->mothurRemove(thisDistFile);
 238                         pDataArray->m->mothurRemove(thisNamefile);
 239                          cout << "here4" << endl;
 240                         if (saveCutoff != pDataArray->cutoff) {
 241                                 if (pDataArray->hard)   {  saveCutoff = pDataArray->m->ceilDist(saveCutoff, pDataArray->precision);     }
 242                                 else            {       saveCutoff = pDataArray->m->roundDist(saveCutoff, pDataArray->precision);  }
 243
 244                                 pDataArray->m->mothurOut("Cutoff was " + toString(pDataArray->cutoff) + " changed cutoff to " + toString(saveCutoff)); pDataArray->m->mothurOutEndLine();
 245                         }
 246                          cout << "here5" << endl;
 247                         if (saveCutoff < smallestCutoff) { smallestCutoff = saveCutoff;  }
 248                 }
 249
 250                 pDataArray->cutoff = smallestCutoff;
 251
 252                 return 0;
 253
 254         }
 255         catch(exception& e) {
 256                 pDataArray->m->errorOut(e, "ClusterSplitCommand", "MyClusterThreadFunction");
 257                 exit(1);
 258         }
 259 }
 260 #endif
 261
 262 */
 263
 264
 265 #endif
 266