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
  27         string getHelpString();
  28     string getOutputPattern(string);
  29         string getCitation() { return "http://www.mothur.org/wiki/Screen.seqs"; }
  30         string getDescription()         { return "enables you to keep sequences that fulfill certain user defined criteria"; }
  31
  32         int execute();
  33         void help() { m->mothurOut(getHelpString()); }
  34
  35
  36 private:
  37
  38         struct linePair {
  39                 unsigned long long start;
  40                 unsigned long long end;
  41                 linePair(unsigned long long i, unsigned long long j) : start(i), end(j) {}
  42         };
  43
  44         vector<linePair> lines;
  45
  46         int screenNameGroupFile(set<string>);
  47         int screenGroupFile(set<string>);
  48     int screenCountFile(set<string>);
  49         int screenAlignReport(set<string>);
  50         int screenQual(set<string>);
  51         int screenTaxonomy(set<string>);
  52
  53         int driver(linePair, string, string, string, set<string>&);
  54         int createProcesses(string, string, string, set<string>&);
  55
  56         #ifdef USE_MPI
  57         int driverMPI(int, int, MPI_File&, MPI_File&, MPI_File&, vector<unsigned long long>&, set<string>&);
  58         #endif
  59
  60         bool abort;
  61         string fastafile, namefile, groupfile, alignreport, outputDir, qualfile, taxonomy, countfile;
  62         int startPos, endPos, maxAmbig, maxHomoP, minLength, maxLength, processors, criteria;
  63         vector<string> outputNames;
  64         vector<string> optimize;
  65         map<string, int> nameMap;
  66
  67         int getSummary(vector<unsigned long long>&);
  68         int createProcessesCreateSummary(vector<int>&, vector<int>&, vector<int>&, vector<int>&, vector<int>&, string);
  69         int driverCreateSummary(vector<int>&, vector<int>&, vector<int>&, vector<int>&, vector<int>&, string, linePair);
  70 };
  71
  72 /**************************************************************************************************/
  73 //custom data structure for threads to use.
  74 // This is passed by void pointer so it can be any data type
  75 // that can be passed using a single void pointer (LPVOID).
  76 struct sumData {
  77         vector<int> startPosition;
  78         vector<int> endPosition;
  79         vector<int> seqLength;
  80         vector<int> ambigBases;
  81         vector<int> longHomoPolymer;
  82         string filename, namefile;
  83         unsigned long long start;
  84         unsigned long long end;
  85         int count;
  86         MothurOut* m;
  87         map<string, int> nameMap;
  88
  89
  90         sumData(){}
  91         sumData(string f, MothurOut* mout, unsigned long long st, unsigned long long en, string nf, map<string, int> nam) {
  92                 filename = f;
  93         namefile = nf;
  94                 m = mout;
  95                 start = st;
  96                 end = en;
  97                 nameMap = nam;
  98                 count = 0;
  99         }
 100 };
 101 /**************************************************************************************************/
 102 //custom data structure for threads to use.
 103 // This is passed by void pointer so it can be any data type
 104 // that can be passed using a single void pointer (LPVOID).
 105 struct sumScreenData {
 106     int startPos, endPos, maxAmbig, maxHomoP, minLength, maxLength;
 107         unsigned long long start;
 108         unsigned long long end;
 109         int count;
 110         MothurOut* m;
 111         string goodFName, badAccnosFName, filename;
 112     set<string> badSeqNames;
 113
 114
 115         sumScreenData(){}
 116         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) {
 117                 startPos = s;
 118                 endPos = e;
 119                 minLength = minl;
 120         maxLength = maxl;
 121                 maxAmbig = a;
 122                 maxHomoP = h;
 123                 filename = f;
 124         goodFName = gf;
 125         badAccnosFName = bf;
 126                 m = mout;
 127                 start = st;
 128                 end = en;
 129                 count = 0;
 130         }
 131 };
 132
 133
 134 /**************************************************************************************************/
 135 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 136 #else
 137 static DWORD WINAPI MySumThreadFunction(LPVOID lpParam){
 138         sumData* pDataArray;
 139         pDataArray = (sumData*)lpParam;
 140
 141         try {
 142                 ifstream in;
 143                 pDataArray->m->openInputFile(pDataArray->filename, in);
 144
 145                 //print header if you are process 0
 146                 if ((pDataArray->start == 0) || (pDataArray->start == 1)) {
 147                         in.seekg(0);
 148                 }else { //this accounts for the difference in line endings.
 149                         in.seekg(pDataArray->start-1); pDataArray->m->gobble(in);
 150                 }
 151
 152                 pDataArray->count = pDataArray->end;
 153                 for(int i = 0; i < pDataArray->end; i++){ //end is the number of sequences to process
 154
 155                         if (pDataArray->m->control_pressed) { in.close();  pDataArray->count = 1; return 1; }
 156
 157                         Sequence current(in); pDataArray->m->gobble(in);
 158
 159                         if (current.getName() != "") {
 160
 161                                 int num = 1;
 162                                 if (pDataArray->namefile != "") {
 163                                         //make sure this sequence is in the namefile, else error
 164                                         map<string, int>::iterator it = pDataArray->nameMap.find(current.getName());
 165
 166                                         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; }
 167                                         else { num = it->second; }
 168                                 }
 169
 170                                 //for each sequence this sequence represents
 171                                 for (int i = 0; i < num; i++) {
 172                                         pDataArray->startPosition.push_back(current.getStartPos());
 173                                         pDataArray->endPosition.push_back(current.getEndPos());
 174                                         pDataArray->seqLength.push_back(current.getNumBases());
 175                                         pDataArray->ambigBases.push_back(current.getAmbigBases());
 176                                         pDataArray->longHomoPolymer.push_back(current.getLongHomoPolymer());
 177                                 }
 178             }
 179                 }
 180
 181                 in.close();
 182
 183                 return 0;
 184
 185         }
 186         catch(exception& e) {
 187                 pDataArray->m->errorOut(e, "ScreenSeqsCommand", "MySumThreadFunction");
 188                 exit(1);
 189         }
 190 }
 191
 192 /**************************************************************************************************/
 193
 194 static DWORD WINAPI MySumScreenThreadFunction(LPVOID lpParam){
 195         sumScreenData* pDataArray;
 196         pDataArray = (sumScreenData*)lpParam;
 197
 198         try {
 199
 200         ofstream goodFile;
 201                 pDataArray->m->openOutputFile(pDataArray->goodFName, goodFile);
 202
 203                 ofstream badAccnosFile;
 204                 pDataArray->m->openOutputFile(pDataArray->badAccnosFName, badAccnosFile);
 205
 206                 ifstream in;
 207                 pDataArray->m->openInputFile(pDataArray->filename, in);
 208
 209                 //print header if you are process 0
 210                 if ((pDataArray->start == 0) || (pDataArray->start == 1)) {
 211                         in.seekg(0);
 212                 }else { //this accounts for the difference in line endings.
 213                         in.seekg(pDataArray->start-1); pDataArray->m->gobble(in);
 214                 }
 215
 216                 pDataArray->count = pDataArray->end;
 217                 for(int i = 0; i < pDataArray->end; i++){ //end is the number of sequences to process
 218
 219                         if (pDataArray->m->control_pressed) { in.close(); badAccnosFile.close(); goodFile.close(); pDataArray->count = 1; return 1; }
 220
 221                         Sequence currSeq(in); pDataArray->m->gobble(in);
 222
 223                         if (currSeq.getName() != "") {
 224                                 bool goodSeq = 1;               //      innocent until proven guilty
 225                                 if(goodSeq == 1 && pDataArray->startPos != -1 && pDataArray->startPos < currSeq.getStartPos())                  {       goodSeq = 0;    }
 226                                 if(goodSeq == 1 && pDataArray->endPos != -1 && pDataArray->endPos > currSeq.getEndPos())                                {       goodSeq = 0;    }
 227                                 if(goodSeq == 1 && pDataArray->maxAmbig != -1 && pDataArray->maxAmbig < currSeq.getAmbigBases())                {       goodSeq = 0;    }
 228                                 if(goodSeq == 1 && pDataArray->maxHomoP != -1 && pDataArray->maxHomoP < currSeq.getLongHomoPolymer())   {       goodSeq = 0;    }
 229                                 if(goodSeq == 1 && pDataArray->minLength != -1 && pDataArray->minLength > currSeq.getNumBases())                {       goodSeq = 0;    }
 230                                 if(goodSeq == 1 && pDataArray->maxLength != -1 && pDataArray->maxLength < currSeq.getNumBases())                {       goodSeq = 0;    }
 231
 232                                 if(goodSeq == 1){
 233                                         currSeq.printSequence(goodFile);
 234                                 }
 235                                 else{
 236                                         badAccnosFile << currSeq.getName() << endl;
 237                                         pDataArray->badSeqNames.insert(currSeq.getName());
 238                                 }
 239
 240                         }
 241             //report progress
 242                         if((i+1) % 100 == 0){   pDataArray->m->mothurOut("Processing sequence: " + toString(i+1)); pDataArray->m->mothurOutEndLine();           }
 243                 }
 244                 //report progress
 245                 if((pDataArray->count) % 100 != 0){     pDataArray->m->mothurOut("Processing sequence: " + toString(pDataArray->count)); pDataArray->m->mothurOutEndLine();             }
 246
 247
 248
 249                 in.close();
 250         goodFile.close();
 251         badAccnosFile.close();
 252
 253                 return 0;
 254
 255         }
 256         catch(exception& e) {
 257                 pDataArray->m->errorOut(e, "ScreenSeqsCommand", "MySumScreenThreadFunction");
 258                 exit(1);
 259         }
 260 }
 261
 262 #endif
 263
 264 /**************************************************************************************************/
 265
 266
 267
 268 #endif