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