screenseqscommand.h

   1 #ifndef SCREENSEQSCOMMAND_H
   2 #define SCREENSEQSCOMMAND_H
   3
   4 /*
   5  *  screenseqscommand.h
   6  *  Mothur
   7  *
   8  *  Created by Pat Schloss on 6/3/09.
   9  *  Copyright 2009 Patrick D. Schloss. All rights reserved.
  10  *
  11  */
  12 #include "mothur.h"
  13 #include "command.hpp"
  14 #include "sequence.hpp"
  15
  16 class ScreenSeqsCommand : public Command {
  17
  18 public:
  19         ScreenSeqsCommand(string);
  20         ScreenSeqsCommand();
  21         ~ScreenSeqsCommand() {}
  22
  23         vector<string> setParameters();
  24         string getCommandName()                 { return "screen.seqs";                         }
  25         string getCommandCategory()             { return "Sequence Processing";         }
  26         string getHelpString();
  27         string getCitation() { return "http://www.mothur.org/wiki/Screen.seqs"; }
  28         string getDescription()         { return "enables you to keep sequences that fulfill certain user defined criteria"; }
  29
  30         int execute();
  31         void help() { m->mothurOut(getHelpString()); }
  32
  33
  34 private:
  35
  36         struct linePair {
  37                 unsigned long long start;
  38                 unsigned long long end;
  39                 linePair(unsigned long long i, unsigned long long j) : start(i), end(j) {}
  40         };
  41
  42         vector<linePair> lines;
  43
  44         int screenNameGroupFile(set<string>);
  45         int screenGroupFile(set<string>);
  46         int screenAlignReport(set<string>);
  47         int screenQual(set<string>);
  48         int screenTaxonomy(set<string>);
  49
  50         int driver(linePair, string, string, string, set<string>&);
  51         int createProcesses(string, string, string, set<string>&);
  52
  53         #ifdef USE_MPI
  54         int driverMPI(int, int, MPI_File&, MPI_File&, MPI_File&, vector<unsigned long long>&, set<string>&);
  55         #endif
  56
  57         bool abort;
  58         string fastafile, namefile, groupfile, alignreport, outputDir, qualfile, taxonomy;
  59         int startPos, endPos, maxAmbig, maxHomoP, minLength, maxLength, processors, criteria;
  60         vector<string> outputNames;
  61         vector<string> optimize;
  62         map<string, int> nameMap;
  63         int readNames();
  64
  65         int getSummary(vector<unsigned long long>&);
  66         int createProcessesCreateSummary(vector<int>&, vector<int>&, vector<int>&, vector<int>&, vector<int>&, string);
  67         int driverCreateSummary(vector<int>&, vector<int>&, vector<int>&, vector<int>&, vector<int>&, string, linePair);
  68 };
  69
  70 /**************************************************************************************************/
  71 //custom data structure for threads to use.
  72 // This is passed by void pointer so it can be any data type
  73 // that can be passed using a single void pointer (LPVOID).
  74 struct sumData {
  75         vector<int> startPosition;
  76         vector<int> endPosition;
  77         vector<int> seqLength;
  78         vector<int> ambigBases;
  79         vector<int> longHomoPolymer;
  80         string filename, namefile;
  81         unsigned long long start;
  82         unsigned long long end;
  83         int count;
  84         MothurOut* m;
  85         map<string, int> nameMap;
  86
  87
  88         sumData(){}
  89         sumData(string f, MothurOut* mout, unsigned long long st, unsigned long long en, string nf, map<string, int> nam) {
  90                 filename = f;
  91         namefile = nf;
  92                 m = mout;
  93                 start = st;
  94                 end = en;
  95                 nameMap = nam;
  96                 count = 0;
  97         }
  98 };
  99 /**************************************************************************************************/
 100 //custom data structure for threads to use.
 101 // This is passed by void pointer so it can be any data type
 102 // that can be passed using a single void pointer (LPVOID).
 103 struct sumScreenData {
 104     int startPos, endPos, maxAmbig, maxHomoP, minLength, maxLength;
 105         unsigned long long start;
 106         unsigned long long end;
 107         int count;
 108         MothurOut* m;
 109         string goodFName, badAccnosFName, filename;
 110     set<string> badSeqNames;
 111
 112
 113         sumScreenData(){}
 114         sumScreenData(int s, int e, int a, int h, int minl, int maxl, string f, MothurOut* mout, unsigned long long st, unsigned long long en, string gf, string bf) {
 115                 startPos = s;
 116                 endPos = e;
 117                 minLength = minl;
 118         maxLength = maxl;
 119                 maxAmbig = a;
 120                 maxHomoP = h;
 121                 filename = f;
 122         goodFName = gf;
 123         badAccnosFName = bf;
 124                 m = mout;
 125                 start = st;
 126                 end = en;
 127                 count = 0;
 128         }
 129 };
 130
 131
 132 /**************************************************************************************************/
 133 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 134 #else
 135 static DWORD WINAPI MySumThreadFunction(LPVOID lpParam){
 136         sumData* pDataArray;
 137         pDataArray = (sumData*)lpParam;
 138
 139         try {
 140                 ifstream in;
 141                 pDataArray->m->openInputFile(pDataArray->filename, in);
 142
 143                 //print header if you are process 0
 144                 if ((pDataArray->start == 0) || (pDataArray->start == 1)) {
 145                         in.seekg(0);
 146                 }else { //this accounts for the difference in line endings.
 147                         in.seekg(pDataArray->start-1); pDataArray->m->gobble(in);
 148                 }
 149
 150                 pDataArray->count = pDataArray->end;
 151                 for(int i = 0; i < pDataArray->end; i++){ //end is the number of sequences to process
 152
 153                         if (pDataArray->m->control_pressed) { in.close();  pDataArray->count = 1; return 1; }
 154
 155                         Sequence current(in); pDataArray->m->gobble(in);
 156
 157                         if (current.getName() != "") {
 158
 159                                 int num = 1;
 160                                 if (pDataArray->namefile != "") {
 161                                         //make sure this sequence is in the namefile, else error
 162                                         map<string, int>::iterator it = pDataArray->nameMap.find(current.getName());
 163
 164                                         if (it == pDataArray->nameMap.end()) { pDataArray->m->mothurOut("[ERROR]: " + current.getName() + " is not in your namefile, please correct."); pDataArray->m->mothurOutEndLine(); pDataArray->m->control_pressed = true; }
 165                                         else { num = it->second; }
 166                                 }
 167
 168                                 //for each sequence this sequence represents
 169                                 for (int i = 0; i < num; i++) {
 170                                         pDataArray->startPosition.push_back(current.getStartPos());
 171                                         pDataArray->endPosition.push_back(current.getEndPos());
 172                                         pDataArray->seqLength.push_back(current.getNumBases());
 173                                         pDataArray->ambigBases.push_back(current.getAmbigBases());
 174                                         pDataArray->longHomoPolymer.push_back(current.getLongHomoPolymer());
 175                                 }
 176             }
 177                 }
 178
 179                 in.close();
 180
 181                 return 0;
 182
 183         }
 184         catch(exception& e) {
 185                 pDataArray->m->errorOut(e, "ScreenSeqsCommand", "MySumThreadFunction");
 186                 exit(1);
 187         }
 188 }
 189
 190 /**************************************************************************************************/
 191
 192 static DWORD WINAPI MySumScreenThreadFunction(LPVOID lpParam){
 193         sumScreenData* pDataArray;
 194         pDataArray = (sumScreenData*)lpParam;
 195
 196         try {
 197
 198         ofstream goodFile;
 199                 pDataArray->m->openOutputFile(pDataArray->goodFName, goodFile);
 200
 201                 ofstream badAccnosFile;
 202                 pDataArray->m->openOutputFile(pDataArray->badAccnosFName, badAccnosFile);
 203
 204                 ifstream in;
 205                 pDataArray->m->openInputFile(pDataArray->filename, in);
 206
 207                 //print header if you are process 0
 208                 if ((pDataArray->start == 0) || (pDataArray->start == 1)) {
 209                         in.seekg(0);
 210                 }else { //this accounts for the difference in line endings.
 211                         in.seekg(pDataArray->start-1); pDataArray->m->gobble(in);
 212                 }
 213
 214                 pDataArray->count = pDataArray->end;
 215                 for(int i = 0; i < pDataArray->end; i++){ //end is the number of sequences to process
 216
 217                         if (pDataArray->m->control_pressed) { in.close(); badAccnosFile.close(); goodFile.close(); pDataArray->count = 1; return 1; }
 218
 219                         Sequence currSeq(in); pDataArray->m->gobble(in);
 220
 221                         if (currSeq.getName() != "") {
 222                                 bool goodSeq = 1;               //      innocent until proven guilty
 223                                 if(goodSeq == 1 && pDataArray->startPos != -1 && pDataArray->startPos < currSeq.getStartPos())                  {       goodSeq = 0;    }
 224                                 if(goodSeq == 1 && pDataArray->endPos != -1 && pDataArray->endPos > currSeq.getEndPos())                                {       goodSeq = 0;    }
 225                                 if(goodSeq == 1 && pDataArray->maxAmbig != -1 && pDataArray->maxAmbig < currSeq.getAmbigBases())                {       goodSeq = 0;    }
 226                                 if(goodSeq == 1 && pDataArray->maxHomoP != -1 && pDataArray->maxHomoP < currSeq.getLongHomoPolymer())   {       goodSeq = 0;    }
 227                                 if(goodSeq == 1 && pDataArray->minLength != -1 && pDataArray->minLength > currSeq.getNumBases())                {       goodSeq = 0;    }
 228                                 if(goodSeq == 1 && pDataArray->maxLength != -1 && pDataArray->maxLength < currSeq.getNumBases())                {       goodSeq = 0;    }
 229
 230                                 if(goodSeq == 1){
 231                                         currSeq.printSequence(goodFile);
 232                                 }
 233                                 else{
 234                                         badAccnosFile << currSeq.getName() << endl;
 235                                         pDataArray->badSeqNames.insert(currSeq.getName());
 236                                 }
 237
 238                         }
 239             //report progress
 240                         if((i+1) % 100 == 0){   pDataArray->m->mothurOut("Processing sequence: " + toString(i+1)); pDataArray->m->mothurOutEndLine();           }
 241                 }
 242                 //report progress
 243                 if((pDataArray->count) % 100 != 0){     pDataArray->m->mothurOut("Processing sequence: " + toString(pDataArray->count)); pDataArray->m->mothurOutEndLine();             }
 244
 245
 246
 247                 in.close();
 248         goodFile.close();
 249         badAccnosFile.close();
 250
 251                 return 0;
 252
 253         }
 254         catch(exception& e) {
 255                 pDataArray->m->errorOut(e, "ScreenSeqsCommand", "MySumScreenThreadFunction");
 256                 exit(1);
 257         }
 258 }
 259
 260 #endif
 261
 262 /**************************************************************************************************/
 263
 264
 265
 266 #endif