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 numLinkers, numSpacers;
  62
  63     int maxFlows, minFlows, minLength, maxLength, maxHomoP, tdiffs, bdiffs, pdiffs;
  64         int numFlows;
  65         float signal, noise;
  66         bool fasta;
  67         string flowOrder;
  68
  69         string flowFileName, oligoFileName, outputDir;
  70
  71         map<string, int> barcodes;
  72         map<string, int> primers;
  73         vector<string> revPrimer;
  74
  75         vector<string> primerNameVector;        //needed here?
  76         vector<string> barcodeNameVector;       //needed here?
  77
  78         map<string, int> combos;                        //needed here?
  79         map<string, int> groupToIndex;          //needed here?
  80
  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 trimFlowData {
  88         string flowFileName;
  89         string trimFlowFileName;
  90         string scrapFlowFileName;
  91         string fastaFileName;
  92         string flowOrder;
  93         vector<vector<string> > barcodePrimerComboFileNames;
  94         map<string, int> barcodes;
  95         map<string, int> primers;
  96         vector<string> revPrimer;
  97         bool fasta, allFiles;
  98         unsigned long long start;
  99         unsigned long long end;
 100         MothurOut* m;
 101         float signal, noise;
 102         int numFlows, maxFlows, minFlows, maxHomoP, tdiffs, bdiffs, pdiffs, threadID, count;
 103
 104         trimFlowData(){}
 105         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) {
 106                 flowFileName = ff;
 107                 trimFlowFileName = tf;
 108                 scrapFlowFileName = sf;
 109                 fastaFileName = f;
 110                 flowOrder = fo;
 111                 barcodePrimerComboFileNames = bfn;
 112                 barcodes = bar;
 113                 primers = pri;
 114                 revPrimer = rev;
 115                 fasta = fa;
 116                 allFiles = al;
 117                 start = st;
 118                 end = en;
 119                 m = mout;
 120                 signal = sig;
 121                 noise = n;
 122                 numFlows = numF;
 123                 maxFlows = maxF;
 124                 minFlows = minF;
 125                 maxHomoP = maxH;
 126                 tdiffs = td;
 127                 bdiffs = bd;
 128                 pdiffs = pd;
 129                 threadID = tid;
 130         }
 131 };
 132
 133 /**************************************************************************************************/
 134 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 135 #else
 136 static DWORD WINAPI MyTrimFlowThreadFunction(LPVOID lpParam){
 137         trimFlowData* pDataArray;
 138         pDataArray = (trimFlowData*)lpParam;
 139
 140         try {
 141                 ofstream trimFlowFile;
 142                 pDataArray->m->openOutputFile(pDataArray->trimFlowFileName, trimFlowFile);
 143                 trimFlowFile.setf(ios::fixed, ios::floatfield); trimFlowFile.setf(ios::showpoint);
 144
 145                 ofstream scrapFlowFile;
 146                 pDataArray->m->openOutputFile(pDataArray->scrapFlowFileName, scrapFlowFile);
 147                 scrapFlowFile.setf(ios::fixed, ios::floatfield); scrapFlowFile.setf(ios::showpoint);
 148
 149                 ofstream fastaFile;
 150                 if(pDataArray->fasta){  pDataArray->m->openOutputFile(pDataArray->fastaFileName, fastaFile);    }
 151
 152                 ifstream flowFile;
 153                 pDataArray->m->openInputFile(pDataArray->flowFileName, flowFile);
 154
 155                 flowFile.seekg(pDataArray->start);
 156
 157                 if(pDataArray->start == 0){
 158                         flowFile >> pDataArray->numFlows; pDataArray->m->gobble(flowFile);
 159                         scrapFlowFile << pDataArray->maxFlows << endl;
 160                         trimFlowFile << pDataArray->maxFlows << endl;
 161                         if(pDataArray->allFiles){
 162                                 for(int i=0;i<pDataArray->barcodePrimerComboFileNames.size();i++){
 163                                         for(int j=0;j<pDataArray->barcodePrimerComboFileNames[0].size();j++){
 164                                                 ofstream temp;
 165                                                 pDataArray->m->openOutputFile(pDataArray->barcodePrimerComboFileNames[i][j], temp);
 166                                                 temp << pDataArray->maxFlows << endl;
 167                                                 temp.close();
 168                                         }
 169                                 }
 170                         }
 171                 }
 172
 173                 FlowData flowData(pDataArray->numFlows, pDataArray->signal, pDataArray->noise, pDataArray->maxHomoP, pDataArray->flowOrder);
 174                 cout << " thread flowdata address " <<  &flowData  << '\t' << &flowFile << endl;
 175                 TrimOligos trimOligos(pDataArray->pdiffs, pDataArray->bdiffs, pDataArray->primers, pDataArray->barcodes, pDataArray->revPrimer);
 176
 177                 pDataArray->count = pDataArray->end;
 178                 cout << pDataArray->threadID << '\t' << pDataArray->count << endl;
 179                 int count = 0;
 180                 for(int i = 0; i < pDataArray->end; i++){ //end is the number of sequences to process
 181
 182                         if (pDataArray->m->control_pressed) {  break; }
 183                         cout << pDataArray->threadID << '\t' << count << endl;
 184                         int success = 1;
 185                         int currentSeqDiffs = 0;
 186                         string trashCode = "";
 187
 188                         flowData.getNext(flowFile);
 189                         cout << "thread good bit " << flowFile.good() << endl;
 190                         flowData.capFlows(pDataArray->maxFlows);
 191
 192                         Sequence currSeq = flowData.getSequence();
 193                         if(!flowData.hasMinFlows(pDataArray->minFlows)){        //screen to see if sequence is of a minimum number of flows
 194                                 success = 0;
 195                                 trashCode += 'l';
 196                         }
 197
 198                         int primerIndex = 0;
 199                         int barcodeIndex = 0;
 200
 201                         if(pDataArray->barcodes.size() != 0){
 202                                 success = trimOligos.stripBarcode(currSeq, barcodeIndex);
 203                                 if(success > pDataArray->bdiffs)                {       trashCode += 'b';       }
 204                                 else{ currentSeqDiffs += success;  }
 205                         }
 206
 207                         if(pDataArray->primers.size() != 0){
 208                                 success = trimOligos.stripForward(currSeq, primerIndex);
 209                                 if(success > pDataArray->pdiffs)                {       trashCode += 'f';       }
 210                                 else{ currentSeqDiffs += success;  }
 211                         }
 212
 213                         if (currentSeqDiffs > pDataArray->tdiffs)       {       trashCode += 't';   }
 214
 215                         if(pDataArray->revPrimer.size() != 0){
 216                                 success = trimOligos.stripReverse(currSeq);
 217                                 if(!success)                            {       trashCode += 'r';       }
 218                         }
 219
 220                         if(trashCode.length() == 0){
 221
 222                                 flowData.printFlows(trimFlowFile);
 223
 224                                 if(pDataArray->fasta)   {       currSeq.printSequence(fastaFile);       }
 225
 226                                 if(pDataArray->allFiles){
 227                                         ofstream output;
 228                                         pDataArray->m->openOutputFileAppend(pDataArray->barcodePrimerComboFileNames[barcodeIndex][primerIndex], output);
 229                                         output.setf(ios::fixed, ios::floatfield); trimFlowFile.setf(ios::showpoint);
 230
 231                                         flowData.printFlows(output);
 232                                         output.close();
 233                                 }
 234                         }
 235                         else{
 236                                 flowData.printFlows(scrapFlowFile, trashCode);
 237                         }
 238
 239                         count++;
 240                                 cout << pDataArray->threadID << '\t' << currSeq.getName() << endl;
 241                         //report progress
 242                         if((count) % 10000 == 0){       pDataArray->m->mothurOut(toString(count)); pDataArray->m->mothurOutEndLine();           }
 243
 244                 }
 245                 //report progress
 246                 if((count) % 10000 != 0){       pDataArray->m->mothurOut(toString(count)); pDataArray->m->mothurOutEndLine();           }
 247
 248                 trimFlowFile.close();
 249                 scrapFlowFile.close();
 250                 flowFile.close();
 251                 if(pDataArray->fasta){  fastaFile.close();      }
 252
 253         }
 254         catch(exception& e) {
 255                 pDataArray->m->errorOut(e, "TrimFlowsCommand", "MyTrimFlowsThreadFunction");
 256                 exit(1);
 257         }
 258 }
 259 #endif
 260
 261
 262 #endif