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