preclustercommand.h

   1 #ifndef PRECLUSTERCOMMAND_H
   2 #define PRECLUSTERCOMMAND_H
   3
   4
   5 /*
   6  *  preclustercommand.h
   7  *  Mothur
   8  *
   9  *  Created by westcott on 12/21/09.
  10  *  Copyright 2009 Schloss Lab. All rights reserved.
  11  *
  12  */
  13
  14
  15 #include "command.hpp"
  16 #include "sequence.hpp"
  17 #include "sequenceparser.h"
  18
  19 /************************************************************/
  20 struct seqPNode {
  21         int numIdentical;
  22         Sequence seq;
  23         string names;
  24         bool active;
  25         int diffs;
  26         seqPNode() {}
  27         seqPNode(int n, Sequence s, string nm) : numIdentical(n), seq(s), names(nm), active(1) { diffs = 0; }
  28         ~seqPNode() {}
  29 };
  30 /************************************************************/
  31 inline bool comparePriority(seqPNode first, seqPNode second) {  return (first.numIdentical > second.numIdentical); }
  32 //************************************************************/
  33
  34 class PreClusterCommand : public Command {
  35
  36 public:
  37         PreClusterCommand(string);
  38         PreClusterCommand();
  39         ~PreClusterCommand(){}
  40
  41         vector<string> setParameters();
  42         string getCommandName()                 { return "pre.cluster";                         }
  43         string getCommandCategory()             { return "Sequence Processing";         }
  44         string getOutputFileNameTag(string, string);
  45         string getHelpString();
  46         string getCitation() { return "Schloss PD, Gevers D, Westcott SL (2011).  Reducing the effects of PCR amplification and sequencing artifacts on 16S rRNA-based studies.  PLoS ONE.  6:e27310.\nhttp://www.mothur.org/wiki/Pre.cluster"; }
  47         string getDescription()         { return "implements a pseudo-single linkage algorithm with the goal of removing sequences that are likely due to pyrosequencing errors"; }
  48
  49
  50         int execute();
  51         void help() { m->mothurOut(getHelpString()); }
  52
  53 private:
  54
  55         struct linePair {
  56                 int start;
  57                 int end;
  58                 linePair(int i, int j) : start(i), end(j) {}
  59         };
  60
  61         int diffs, length, processors;
  62         bool abort, bygroup;
  63         string fastafile, namefile, outputDir, groupfile;
  64         vector<seqPNode> alignSeqs; //maps the number of identical seqs to a sequence
  65         map<string, string> names; //represents the names file first column maps to second column
  66         map<string, int> sizes;  //this map a seq name to the number of identical seqs in the names file
  67         map<string, int>::iterator itSize;
  68 //      map<string, bool> active; //maps sequence name to whether it has already been merged or not.
  69         vector<string> outputNames;
  70         map<string, vector<string> > outputTypes;
  71
  72         int readFASTA();
  73         void readNameFile();
  74         //int readNamesFASTA();
  75         int calcMisMatches(string, string);
  76         void printData(string, string); //fasta filename, names file name
  77         int process(string);
  78         int loadSeqs(map<string, string>&, vector<Sequence>&);
  79         int driverGroups(SequenceParser*, string, string, string, int, int, vector<string> groups);
  80         int createProcessesGroups(SequenceParser*, string, string, string, vector<string>);
  81 };
  82
  83 /**************************************************************************************************/
  84 //custom data structure for threads to use.
  85 // This is passed by void pointer so it can be any data type
  86 // that can be passed using a single void pointer (LPVOID).
  87 struct preClusterData {
  88         string fastafile;
  89         string namefile;
  90         string groupfile;
  91         string newFName, newNName, newMName;
  92         MothurOut* m;
  93         int start;
  94         int end;
  95         int diffs, threadID;
  96         vector<string> groups;
  97         vector<string> mapFileNames;
  98
  99         preClusterData(){}
 100         preClusterData(string f, string n, string g, string nff,  string nnf, string nmf, vector<string> gr, MothurOut* mout, int st, int en, int d, int tid) {
 101                 fastafile = f;
 102                 namefile = n;
 103                 groupfile = g;
 104                 newFName = nff;
 105                 newNName = nnf;
 106                 newMName = nmf;
 107                 m = mout;
 108                 start = st;
 109                 end = en;
 110                 diffs = d;
 111                 threadID = tid;
 112                 groups = gr;
 113         }
 114 };
 115
 116 /**************************************************************************************************/
 117 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 118 #else
 119 static DWORD WINAPI MyPreclusterThreadFunction(LPVOID lpParam){
 120         preClusterData* pDataArray;
 121         pDataArray = (preClusterData*)lpParam;
 122
 123         try {
 124
 125                 //parse fasta and name file by group
 126                 SequenceParser* parser;
 127                 if (pDataArray->namefile != "") { parser = new SequenceParser(pDataArray->groupfile, pDataArray->fastafile, pDataArray->namefile);      }
 128                 else                                                    { parser = new SequenceParser(pDataArray->groupfile, pDataArray->fastafile);                                            }
 129
 130                 int numSeqs = 0;
 131                 vector<seqPNode> alignSeqs;
 132                 //clear out old files
 133                 ofstream outF; pDataArray->m->openOutputFile(pDataArray->newFName, outF); outF.close();
 134                 ofstream outN; pDataArray->m->openOutputFile(pDataArray->newNName, outN);  outN.close();
 135
 136                 //precluster each group
 137                 for (int k = pDataArray->start; k < pDataArray->end; k++) {
 138
 139                         int start = time(NULL);
 140
 141                         if (pDataArray->m->control_pressed) {  delete parser; return 0; }
 142
 143                         pDataArray->m->mothurOutEndLine(); pDataArray->m->mothurOut("Processing group " + pDataArray->groups[k] + ":"); pDataArray->m->mothurOutEndLine();
 144
 145                         map<string, string> thisNameMap;
 146                         if (pDataArray->namefile != "") { thisNameMap = parser->getNameMap(pDataArray->groups[k]); }
 147                         vector<Sequence> thisSeqs = parser->getSeqs(pDataArray->groups[k]);
 148
 149                         //fill alignSeqs with this groups info.
 150                         ////////////////////////////////////////////////////
 151                         //numSeqs = loadSeqs(thisNameMap, thisSeqs); same function below
 152
 153                         int length = 0;
 154                         alignSeqs.clear();
 155                         map<string, string>::iterator it;
 156                         bool error = false;
 157
 158                         for (int i = 0; i < thisSeqs.size(); i++) {
 159
 160                                 if (pDataArray->m->control_pressed) { delete parser; return 0; }
 161
 162                                 if (pDataArray->namefile != "") {
 163                                         it = thisNameMap.find(thisSeqs[i].getName());
 164
 165                                         //should never be true since parser checks for this
 166                                         if (it == thisNameMap.end()) { pDataArray->m->mothurOut(thisSeqs[i].getName() + " is not in your names file, please correct."); pDataArray->m->mothurOutEndLine(); error = true; }
 167                                         else{
 168                                                 //get number of reps
 169                                                 int numReps = 1;
 170                                                 for(int j=0;j<(it->second).length();j++){
 171                                                         if((it->second)[j] == ','){     numReps++;      }
 172                                                 }
 173
 174                                                 seqPNode tempNode(numReps, thisSeqs[i], it->second);
 175                                                 alignSeqs.push_back(tempNode);
 176                                                 if (thisSeqs[i].getAligned().length() > length) {  length = thisSeqs[i].getAligned().length();  }
 177                                         }
 178                                 }else { //no names file, you are identical to yourself
 179                                         seqPNode tempNode(1, thisSeqs[i], thisSeqs[i].getName());
 180                                         alignSeqs.push_back(tempNode);
 181                                         if (thisSeqs[i].getAligned().length() > length) {  length = thisSeqs[i].getAligned().length();  }
 182                                 }
 183                         }
 184
 185                         //sanity check
 186                         if (error) { pDataArray->m->control_pressed = true; }
 187
 188                         thisSeqs.clear();
 189                         numSeqs = alignSeqs.size();
 190
 191                         ////////////////////////////////////////////////////
 192
 193                         if (pDataArray->m->control_pressed) {   delete parser; return 0; }
 194
 195                         if (pDataArray->diffs > length) { pDataArray->m->mothurOut("Error: diffs is greater than your sequence length."); pDataArray->m->mothurOutEndLine(); pDataArray->m->control_pressed = true; return 0;  }
 196
 197                         ////////////////////////////////////////////////////
 198                         //int count = process(); - same function below
 199
 200                         ofstream out;
 201                         pDataArray->m->openOutputFile(pDataArray->newMName+pDataArray->groups[k]+".map", out);
 202                         pDataArray->mapFileNames.push_back(pDataArray->newMName+pDataArray->groups[k]+".map");
 203
 204                         //sort seqs by number of identical seqs
 205                         sort(alignSeqs.begin(), alignSeqs.end(), comparePriority);
 206
 207                         int count = 0;
 208
 209                         //think about running through twice...
 210                         for (int i = 0; i < numSeqs; i++) {
 211
 212                                 //are you active
 213                                 //                      itActive = active.find(alignSeqs[i].seq.getName());
 214
 215                                 if (alignSeqs[i].active) {  //this sequence has not been merged yet
 216
 217                                         string chunk = alignSeqs[i].seq.getName() + "\t" + toString(alignSeqs[i].numIdentical) + "\t" + toString(0) + "\t" + alignSeqs[i].seq.getAligned() + "\n";
 218
 219                                         //try to merge it with all smaller seqs
 220                                         for (int j = i+1; j < numSeqs; j++) {
 221
 222                                                 if (pDataArray->m->control_pressed) { delete parser; return 0; }
 223
 224                                                 if (alignSeqs[j].active) {  //this sequence has not been merged yet
 225                                                         //are you within "diff" bases
 226                                                         //int mismatch = calcMisMatches(alignSeqs[i].seq.getAligned(), alignSeqs[j].seq.getAligned());
 227                                                         int mismatch = 0;
 228
 229                                                         for (int k = 0; k < alignSeqs[i].seq.getAligned().length(); k++) {
 230                                                                 //do they match
 231                                                                 if (alignSeqs[i].seq.getAligned()[k] != alignSeqs[j].seq.getAligned()[k]) { mismatch++; }
 232                                                                 if (mismatch > pDataArray->diffs) { mismatch = length; break; } //to far to cluster
 233                                                         }
 234
 235                                                         if (mismatch <= pDataArray->diffs) {
 236                                                                 //merge
 237                                                                 alignSeqs[i].names += ',' + alignSeqs[j].names;
 238                                                                 alignSeqs[i].numIdentical += alignSeqs[j].numIdentical;
 239
 240                                                                 alignSeqs[j].active = 0;
 241                                                                 alignSeqs[j].numIdentical = 0;
 242                                                                 alignSeqs[j].diffs = mismatch;
 243                                                                 count++;
 244                                                                 chunk += alignSeqs[j].seq.getName() + "\t" + toString(alignSeqs[j].numIdentical) + "\t" + toString(mismatch) + "\t" + alignSeqs[j].seq.getAligned() + "\n";
 245                                                         }
 246                                                 }//end if j active
 247                                         }//end for loop j
 248
 249                                         //remove from active list
 250                                         alignSeqs[i].active = 0;
 251
 252                                         out << "ideal_seq_" << (i+1) << '\t' << alignSeqs[i].numIdentical << endl << chunk << endl;
 253
 254                                 }//end if active i
 255                                 if(i % 100 == 0)        { pDataArray->m->mothurOut(toString(i) + "\t" + toString(numSeqs - count) + "\t" + toString(count)); pDataArray->m->mothurOutEndLine(); }
 256                         }
 257                         out.close();
 258                         if(numSeqs % 100 != 0)  { pDataArray->m->mothurOut(toString(numSeqs) + "\t" + toString(numSeqs - count) + "\t" + toString(count)); pDataArray->m->mothurOutEndLine();   }
 259                         ////////////////////////////////////////////////////
 260
 261                         if (pDataArray->m->control_pressed) {  delete parser; return 0; }
 262
 263                         pDataArray->m->mothurOut("Total number of sequences before pre.cluster was " + toString(alignSeqs.size()) + ".");pDataArray-> m->mothurOutEndLine();
 264                         pDataArray->m->mothurOut("pre.cluster removed " + toString(count) + " sequences."); pDataArray->m->mothurOutEndLine(); pDataArray->m->mothurOutEndLine();
 265
 266                         ////////////////////////////////////////////////////
 267                         //printData(pDataArray->newFFile, pDataArray->newNFile); - same as below
 268                         ofstream outFasta;
 269                         ofstream outNames;
 270
 271                         pDataArray->m->openOutputFileAppend(pDataArray->newFName, outFasta);
 272                         pDataArray->m->openOutputFileAppend(pDataArray->newNName, outNames);
 273
 274                         for (int i = 0; i < alignSeqs.size(); i++) {
 275                                 if (alignSeqs[i].numIdentical != 0) {
 276                                         alignSeqs[i].seq.printSequence(outFasta);
 277                                         outNames << alignSeqs[i].seq.getName() << '\t' << alignSeqs[i].names << endl;
 278                                 }
 279                         }
 280
 281                         outFasta.close();
 282                         outNames.close();
 283                         ////////////////////////////////////////////////////
 284
 285                         pDataArray->m->mothurOut("It took " + toString(time(NULL) - start) + " secs to cluster " + toString(numSeqs) + " sequences."); pDataArray->m->mothurOutEndLine();
 286
 287                 }
 288
 289                 return numSeqs;
 290
 291
 292         }
 293         catch(exception& e) {
 294                 pDataArray->m->errorOut(e, "PreClusterCommand", "MyPreclusterThreadFunction");
 295                 exit(1);
 296         }
 297 }
 298 #endif
 299
 300 /**************************************************************************************************/
 301
 302
 303 #endif
 304
 305