trimflowscommand.h

   1 #ifndef TRIMFLOWSCOMMAND_H
   2 #define TRIMFLOWSCOMMAND_H
   3
   4 /*
   5  *  trimflowscommand.h
   6  *  Mothur
   7  *
   8  *  Created by Pat Schloss on 12/22/10.
   9  *  Copyright 2010 Schloss Lab. All rights reserved.
  10  *
  11  */
  12
  13 #include "mothur.h"
  14 #include "command.hpp"
  15 #include "sequence.hpp"
  16 #include "flowdata.h"
  17 #include "groupmap.h"
  18 #include "trimoligos.h"
  19
  20 class TrimFlowsCommand : public Command {
  21 public:
  22         TrimFlowsCommand(string);
  23         TrimFlowsCommand();
  24         ~TrimFlowsCommand() {}
  25
  26         vector<string> setParameters();
  27         string getCommandName()                 { return "trim.flows";  }
  28         string getCommandCategory()             { return "Sequence Processing";         }
  29         string getHelpString();
  30         string getCitation() { return "http://www.mothur.org/wiki/Trim.flows"; }
  31         string getDescription()         { return "trim.flows"; }
  32
  33
  34         int execute();
  35         void help() { m->mothurOut(getHelpString()); }
  36
  37 private:
  38         bool abort;
  39
  40         struct linePair {
  41                 unsigned long long start;
  42                 unsigned long long end;
  43                 linePair(unsigned long long i, unsigned long long j) : start(i), end(j) {}
  44         };
  45         int comboStarts;
  46         vector<int> processIDS;   //processid
  47         vector<linePair*> lines;
  48
  49         vector<unsigned long long> getFlowFileBreaks();
  50         int createProcessesCreateTrim(string, string, string, string, vector<vector<string> >);
  51         int driverCreateTrim(string, string, string, string, vector<vector<string> >, linePair*);
  52
  53         vector<string> outputNames;
  54         set<string> filesToRemove;
  55
  56         void getOligos(vector<vector<string> >&);               //a rewrite of what is in trimseqscommand.h
  57
  58         bool allFiles;
  59         int processors;
  60         int numFPrimers, numRPrimers;
  61         int maxFlows, minFlows, minLength, maxLength, maxHomoP, tdiffs, bdiffs, pdiffs, sdiffs, ldiffs, numLinkers, numSpacers;
  62         int numFlows;
  63         float signal, noise;
  64         bool fasta;
  65         string flowOrder;
  66
  67         string flowFileName, oligoFileName, outputDir;
  68
  69
  70         map<string, int> barcodes;
  71         map<string, int> primers;
  72         vector<string> revPrimer;
  73     vector<string> linker;
  74     vector<string> spacer;
  75
  76         vector<string> primerNameVector;        //needed here?
  77         vector<string> barcodeNameVector;       //needed here?
  78
  79         map<string, int> combos;                        //needed here?
  80         map<string, int> groupToIndex;          //needed here?
  81
  82 };
  83
  84 /**************************************************************************************************/
  85 //custom data structure for threads to use.
  86 // This is passed by void pointer so it can be any data type
  87 // that can be passed using a single void pointer (LPVOID).
  88 struct trimFlowData {
  89         string flowFileName;
  90         string trimFlowFileName;
  91         string scrapFlowFileName;
  92         string fastaFileName;
  93         string flowOrder;
  94         vector<vector<string> > barcodePrimerComboFileNames;
  95         map<string, int> barcodes;
  96         map<string, int> primers;
  97         vector<string> revPrimer;
  98         bool fasta, allFiles;
  99         unsigned long long start;
 100         unsigned long long end;
 101         MothurOut* m;
 102         float signal, noise;
 103         int numFlows, maxFlows, minFlows, maxHomoP, tdiffs, bdiffs, pdiffs, threadID, count;
 104
 105         trimFlowData(){}
 106         trimFlowData(string ff, string tf, string sf, string f, string fo, vector<vector<string> > bfn, map<string, int> bar, map<string, int> pri, vector<string> rev, bool fa, bool al, unsigned long long st, unsigned long long en, MothurOut* mout, float sig, float n, int numF, int maxF, int minF, int maxH, int td, int bd, int pd, int tid) {
 107                 flowFileName = ff;
 108                 trimFlowFileName = tf;
 109                 scrapFlowFileName = sf;
 110                 fastaFileName = f;
 111                 flowOrder = fo;
 112                 barcodePrimerComboFileNames = bfn;
 113                 barcodes = bar;
 114                 primers = pri;
 115                 revPrimer = rev;
 116                 fasta = fa;
 117                 allFiles = al;
 118                 start = st;
 119                 end = en;
 120                 m = mout;
 121                 signal = sig;
 122                 noise = n;
 123                 numFlows = numF;
 124                 maxFlows = maxF;
 125                 minFlows = minF;
 126                 maxHomoP = maxH;
 127                 tdiffs = td;
 128                 bdiffs = bd;
 129                 pdiffs = pd;
 130                 threadID = tid;
 131         }
 132 };
 133
 134 /**************************************************************************************************/
 135 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 136 #else
 137 static DWORD WINAPI MyTrimFlowThreadFunction(LPVOID lpParam){
 138         trimFlowData* pDataArray;
 139         pDataArray = (trimFlowData*)lpParam;
 140
 141         try {
 142                 ofstream trimFlowFile;
 143                 pDataArray->m->openOutputFile(pDataArray->trimFlowFileName, trimFlowFile);
 144                 trimFlowFile.setf(ios::fixed, ios::floatfield); trimFlowFile.setf(ios::showpoint);
 145
 146                 ofstream scrapFlowFile;
 147                 pDataArray->m->openOutputFile(pDataArray->scrapFlowFileName, scrapFlowFile);
 148                 scrapFlowFile.setf(ios::fixed, ios::floatfield); scrapFlowFile.setf(ios::showpoint);
 149
 150                 ofstream fastaFile;
 151                 if(pDataArray->fasta){  pDataArray->m->openOutputFile(pDataArray->fastaFileName, fastaFile);    }
 152
 153                 ifstream flowFile;
 154                 pDataArray->m->openInputFile(pDataArray->flowFileName, flowFile);
 155
 156                 flowFile.seekg(pDataArray->start);
 157
 158                 if(pDataArray->start == 0){
 159                         flowFile >> pDataArray->numFlows; pDataArray->m->gobble(flowFile);
 160                         scrapFlowFile << pDataArray->maxFlows << endl;
 161                         trimFlowFile << pDataArray->maxFlows << endl;
 162                         if(pDataArray->allFiles){
 163                                 for(int i=0;i<pDataArray->barcodePrimerComboFileNames.size();i++){
 164                                         for(int j=0;j<pDataArray->barcodePrimerComboFileNames[0].size();j++){
 165                                                 ofstream temp;
 166                                                 pDataArray->m->openOutputFile(pDataArray->barcodePrimerComboFileNames[i][j], temp);
 167                                                 temp << pDataArray->maxFlows << endl;
 168                                                 temp.close();
 169                                         }
 170                                 }
 171                         }
 172                 }
 173
 174                 FlowData flowData(pDataArray->numFlows, pDataArray->signal, pDataArray->noise, pDataArray->maxHomoP, pDataArray->flowOrder);
 175                 cout << " thread flowdata address " <<  &flowData  << '\t' << &flowFile << endl;
 176                 TrimOligos trimOligos(pDataArray->pdiffs, pDataArray->bdiffs, pDataArray->primers, pDataArray->barcodes, pDataArray->revPrimer);
 177
 178                 pDataArray->count = pDataArray->end;
 179                 cout << pDataArray->threadID << '\t' << pDataArray->count << endl;
 180                 int count = 0;
 181                 for(int i = 0; i < pDataArray->end; i++){ //end is the number of sequences to process
 182
 183                         if (pDataArray->m->control_pressed) {  break; }
 184                         cout << pDataArray->threadID << '\t' << count << endl;
 185                         int success = 1;
 186                         int currentSeqDiffs = 0;
 187                         string trashCode = "";
 188
 189                         flowData.getNext(flowFile);
 190                         cout << "thread good bit " << flowFile.good() << endl;
 191                         flowData.capFlows(pDataArray->maxFlows);
 192
 193                         Sequence currSeq = flowData.getSequence();
 194                         if(!flowData.hasMinFlows(pDataArray->minFlows)){        //screen to see if sequence is of a minimum number of flows
 195                                 success = 0;
 196                                 trashCode += 'l';
 197                         }
 198
 199                         int primerIndex = 0;
 200                         int barcodeIndex = 0;
 201
 202                         if(pDataArray->barcodes.size() != 0){
 203                                 success = trimOligos.stripBarcode(currSeq, barcodeIndex);
 204                                 if(success > pDataArray->bdiffs)                {       trashCode += 'b';       }
 205                                 else{ currentSeqDiffs += success;  }
 206                         }
 207
 208                         if(pDataArray->primers.size() != 0){
 209                                 success = trimOligos.stripForward(currSeq, primerIndex);
 210                                 if(success > pDataArray->pdiffs)                {       trashCode += 'f';       }
 211                                 else{ currentSeqDiffs += success;  }
 212                         }
 213
 214                         if (currentSeqDiffs > pDataArray->tdiffs)       {       trashCode += 't';   }
 215
 216                         if(pDataArray->revPrimer.size() != 0){
 217                                 success = trimOligos.stripReverse(currSeq);
 218                                 if(!success)                            {       trashCode += 'r';       }
 219                         }
 220
 221                         if(trashCode.length() == 0){
 222
 223                                 flowData.printFlows(trimFlowFile);
 224
 225                                 if(pDataArray->fasta)   {       currSeq.printSequence(fastaFile);       }
 226
 227                                 if(pDataArray->allFiles){
 228                                         ofstream output;
 229                                         pDataArray->m->openOutputFileAppend(pDataArray->barcodePrimerComboFileNames[barcodeIndex][primerIndex], output);
 230                                         output.setf(ios::fixed, ios::floatfield); trimFlowFile.setf(ios::showpoint);
 231
 232                                         flowData.printFlows(output);
 233                                         output.close();
 234                                 }
 235                         }
 236                         else{
 237                                 flowData.printFlows(scrapFlowFile, trashCode);
 238                         }
 239
 240                         count++;
 241                                 cout << pDataArray->threadID << '\t' << currSeq.getName() << endl;
 242                         //report progress
 243                         if((count) % 10000 == 0){       pDataArray->m->mothurOut(toString(count)); pDataArray->m->mothurOutEndLine();           }
 244
 245                 }
 246                 //report progress
 247                 if((count) % 10000 != 0){       pDataArray->m->mothurOut(toString(count)); pDataArray->m->mothurOutEndLine();           }
 248
 249                 trimFlowFile.close();
 250                 scrapFlowFile.close();
 251                 flowFile.close();
 252                 if(pDataArray->fasta){  fastaFile.close();      }
 253
 254         }
 255         catch(exception& e) {
 256                 pDataArray->m->errorOut(e, "TrimFlowsCommand", "MyTrimFlowsThreadFunction");
 257                 exit(1);
 258         }
 259 }
 260 #endif
 261
 262
 263 #endif