screenseqscommand.cpp

   1 /*
   2  *  screenseqscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Pat Schloss on 6/3/09.
   6  *  Copyright 2009 Patrick D. Schloss. All rights reserved.
   7  *
   8  */
   9
  10 #include "screenseqscommand.h"
  11 #include "sequence.hpp"
  12
  13 //***************************************************************************************************************
  14
  15 ScreenSeqsCommand::ScreenSeqsCommand(string option){
  16         try {
  17                 abort = false;
  18
  19                 //allow user to run help
  20                 if(option == "help") { help(); abort = true; }
  21
  22                 else {
  23                         //valid paramters for this command
  24                         string AlignArray[] =  {"fasta", "start", "end", "maxambig", "maxhomop", "minlength", "maxlength",
  25                                                                         "name", "group", "alignreport"};
  26                         vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  27
  28                         OptionParser parser(option);
  29                         map<string,string> parameters = parser.getParameters();
  30
  31                         ValidParameters validParameter;
  32
  33                         //check to make sure all parameters are valid for command
  34                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  35                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  36                         }
  37
  38                         //check for required parameters
  39                         fastafile = validParameter.validFile(parameters, "fasta", true);
  40                         if (fastafile == "not found") { cout << "fasta is a required parameter for the screen.seqs command." << endl; abort = true; }
  41                         else if (fastafile == "not open") { abort = true; }
  42
  43                         groupfile = validParameter.validFile(parameters, "group", true);
  44                         if (groupfile == "not open") { abort = true; }
  45                         else if (groupfile == "not found") { groupfile = ""; }
  46
  47                         namefile = validParameter.validFile(parameters, "name", true);
  48                         if (namefile == "not open") { abort = true; }
  49                         else if (namefile == "not found") { namefile = ""; }
  50
  51                         alignreport = validParameter.validFile(parameters, "alignreport", true);
  52                         if (alignreport == "not open") { abort = true; }
  53                         else if (alignreport == "not found") { namefile = ""; }
  54
  55                         //check for optional parameter and set defaults
  56                         // ...at some point should added some additional type checking...
  57                         string temp;
  58                         temp = validParameter.validFile(parameters, "start", false);            if (temp == "not found") { temp = "-1"; }
  59                         convert(temp, startPos);
  60
  61                         temp = validParameter.validFile(parameters, "end", false);                      if (temp == "not found") { temp = "-1"; }
  62                         convert(temp, endPos);
  63
  64                         temp = validParameter.validFile(parameters, "maxambig", false);         if (temp == "not found") { temp = "-1"; }
  65                         convert(temp, maxAmbig);
  66
  67                         temp = validParameter.validFile(parameters, "maxhomop", false);         if (temp == "not found") { temp = "-1"; }
  68                         convert(temp, maxHomoP);
  69
  70                         temp = validParameter.validFile(parameters, "minlength", false);        if (temp == "not found") { temp = "-1"; }
  71                         convert(temp, minLength);
  72
  73                         temp = validParameter.validFile(parameters, "maxlength", false);        if (temp == "not found") { temp = "-1"; }
  74                         convert(temp, maxLength);
  75                 }
  76
  77         }
  78         catch(exception& e) {
  79                 cout << "Standard Error: " << e.what() << " has occurred in the ScreenSeqsCommand class Function ScreenSeqsCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  80                 exit(1);
  81         }
  82         catch(...) {
  83                 cout << "An unknown error has occurred in the ScreenSeqsCommand class function ScreenSeqsCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  84                 exit(1);
  85         }
  86 }
  87 //**********************************************************************************************************************
  88
  89 void ScreenSeqsCommand::help(){
  90         try {
  91                 cout << "The screen.seqs command reads a fastafile and creates ....." << "\n";
  92                 cout << "The screen.seqs command parameters are fasta, start, end, maxambig, maxhomop, minlength, maxlength, name, and group." << "\n";
  93                 cout << "The fasta parameter is required." << "\n";
  94                 cout << "The start parameter .... The default is -1." << "\n";
  95                 cout << "The end parameter .... The default is -1." << "\n";
  96                 cout << "The maxambig parameter .... The default is -1." << "\n";
  97                 cout << "The maxhomop parameter .... The default is -1." << "\n";
  98                 cout << "The minlength parameter .... The default is -1." << "\n";
  99                 cout << "The maxlength parameter .... The default is -1." << "\n";
 100                 cout << "The name parameter allows you to provide a namesfile, and the group parameter allows you to provide a groupfile." << "\n";
 101                 cout << "The screen.seqs command should be in the following format: " << "\n";
 102                 cout << "screen.seqs(fasta=yourFastaFile, name=youNameFile, group=yourGroupFIle, start=yourStart, end=yourEnd, maxambig=yourMaxambig,  " << "\n";
 103                 cout << "maxhomop=yourMaxhomop, minlength=youMinlength, maxlength=yourMaxlength)  " << "\n";
 104                 cout << "Example screen.seqs(fasta=abrecovery.fasta, name=abrecovery.names, group=abrecovery.groups, start=..., end=..., maxambig=..., maxhomop=..., minlength=..., maxlength=...)." << "\n";
 105                 cout << "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta)." << "\n" << "\n";
 106
 107         }
 108         catch(exception& e) {
 109                 cout << "Standard Error: " << e.what() << " has occurred in the ScreenSeqsCommand class Function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 110                 exit(1);
 111         }
 112         catch(...) {
 113                 cout << "An unknown error has occurred in the ScreenSeqsCommand class function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 114                 exit(1);
 115         }
 116 }
 117
 118 //***************************************************************************************************************
 119
 120 ScreenSeqsCommand::~ScreenSeqsCommand(){        /*      do nothing      */      }
 121
 122 //***************************************************************************************************************
 123
 124 int ScreenSeqsCommand::execute(){
 125         try{
 126
 127                 if (abort == true) { return 0; }
 128
 129                 ifstream inFASTA;
 130                 openInputFile(fastafile, inFASTA);
 131
 132                 set<string> badSeqNames;
 133
 134                 string goodSeqFile = getRootName(fastafile) + "good" + getExtension(fastafile);
 135                 string badSeqFile = getRootName(fastafile) + "bad" + getExtension(fastafile);
 136
 137                 ofstream goodSeqOut;    openOutputFile(goodSeqFile, goodSeqOut);
 138                 ofstream badSeqOut;             openOutputFile(badSeqFile, badSeqOut);
 139
 140                 while(!inFASTA.eof()){
 141                         Sequence currSeq(inFASTA);
 142                         bool goodSeq = 1;               //      innocent until proven guilty
 143                         if(goodSeq == 1 && startPos != -1 && startPos < currSeq.getStartPos())                  {       goodSeq = 0;    }
 144                         if(goodSeq == 1 && endPos != -1 && endPos > currSeq.getEndPos())                                {       goodSeq = 0;    }
 145                         if(goodSeq == 1 && maxAmbig != -1 && maxAmbig < currSeq.getAmbigBases())                {       goodSeq = 0;    }
 146                         if(goodSeq == 1 && maxHomoP != -1 && maxHomoP < currSeq.getLongHomoPolymer())   {       goodSeq = 0;    }
 147                         if(goodSeq == 1 && minLength != -1 && minLength > currSeq.getNumBases())                {       goodSeq = 0;    }
 148                         if(goodSeq == 1 && maxLength != -1 && maxLength < currSeq.getNumBases())                {       goodSeq = 0;    }
 149
 150                         if(goodSeq == 1){
 151                                 currSeq.printSequence(goodSeqOut);
 152                         }
 153                         else{
 154                                 currSeq.printSequence(badSeqOut);
 155                                 badSeqNames.insert(currSeq.getName());
 156                         }
 157                         gobble(inFASTA);
 158                 }
 159                 if(namefile != "")              {       screenNameGroupFile(badSeqNames);       }
 160                 if(groupfile != "")             {       screenGroupFile(badSeqNames);           }
 161                 if(alignreport != "")   {       screenAlignReport(badSeqNames);         }
 162
 163                 return 0;
 164         }
 165         catch(exception& e) {
 166                 cout << "Standard Error: " << e.what() << " has occurred in the ScreenSeqsCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 167                 exit(1);
 168         }
 169         catch(...) {
 170                 cout << "An unknown error has occurred in the ScreenSeqsCommand class function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 171                 exit(1);
 172         }
 173
 174 }
 175
 176 //***************************************************************************************************************
 177
 178 void ScreenSeqsCommand::screenNameGroupFile(set<string> badSeqNames){
 179
 180         ifstream inputNames;
 181         openInputFile(namefile, inputNames);
 182         set<string> badSeqGroups;
 183         string seqName, seqList, group;
 184         set<string>::iterator it;
 185
 186         string goodNameFile = getRootName(namefile) + "good" + getExtension(namefile);
 187         string badNameFile = getRootName(namefile) + "bad" + getExtension(namefile);
 188
 189         ofstream goodNameOut;   openOutputFile(goodNameFile, goodNameOut);
 190         ofstream badNameOut;    openOutputFile(badNameFile, badNameOut);
 191
 192         while(!inputNames.eof()){
 193                 inputNames >> seqName >> seqList;
 194                 it = badSeqNames.find(seqName);
 195
 196                 if(it != badSeqNames.end()){
 197                         badSeqNames.erase(it);
 198                         badNameOut << seqName << '\t' << seqList << endl;
 199                         if(namefile != ""){
 200                                 int start = 0;
 201                                 for(int i=0;i<seqList.length();i++){
 202                                         if(seqList[i] == ','){
 203                                                 badSeqGroups.insert(seqList.substr(start,i-start));
 204                                                 start = i+1;
 205                                         }
 206                                 }
 207                                 badSeqGroups.insert(seqList.substr(start,seqList.length()-start));
 208                         }
 209                 }
 210                 else{
 211                         goodNameOut << seqName << '\t' << seqList << endl;
 212                 }
 213                 gobble(inputNames);
 214         }
 215         inputNames.close();
 216         goodNameOut.close();
 217         badNameOut.close();
 218
 219         if(groupfile != ""){
 220
 221                 ifstream inputGroups;
 222                 openInputFile(groupfile, inputGroups);
 223
 224                 string goodGroupFile = getRootName(groupfile) + "good" + getExtension(groupfile);
 225                 string badGroupFile = getRootName(groupfile) + "bad" + getExtension(groupfile);
 226
 227                 ofstream goodGroupOut;  openOutputFile(goodGroupFile, goodGroupOut);
 228                 ofstream badGroupOut;   openOutputFile(badGroupFile, badGroupOut);
 229
 230                 while(!inputGroups.eof()){
 231                         inputGroups >> seqName >> group;
 232
 233                         it = badSeqGroups.find(seqName);
 234
 235                         if(it != badSeqGroups.end()){
 236                                 badSeqGroups.erase(it);
 237                                 badGroupOut << seqName << '\t' << group << endl;
 238                         }
 239                         else{
 240                                 goodGroupOut << seqName << '\t' << group << endl;
 241                         }
 242                         gobble(inputGroups);
 243                 }
 244                 inputGroups.close();
 245                 goodGroupOut.close();
 246                 badGroupOut.close();
 247         }
 248 }
 249
 250 //***************************************************************************************************************
 251
 252 void ScreenSeqsCommand::screenGroupFile(set<string> badSeqNames){
 253
 254         ifstream inputGroups;
 255         openInputFile(groupfile, inputGroups);
 256         string seqName, group;
 257         set<string>::iterator it;
 258
 259         string goodGroupFile = getRootName(groupfile) + "good" + getExtension(groupfile);
 260         string badGroupFile = getRootName(groupfile) + "bad" + getExtension(groupfile);
 261
 262         ofstream goodGroupOut;  openOutputFile(goodGroupFile, goodGroupOut);
 263         ofstream badGroupOut;   openOutputFile(badGroupFile, badGroupOut);
 264
 265         while(!inputGroups.eof()){
 266                 inputGroups >> seqName >> group;
 267                 it = badSeqNames.find(seqName);
 268
 269                 if(it != badSeqNames.end()){
 270                         badSeqNames.erase(it);
 271                         badGroupOut << seqName << '\t' << group << endl;
 272                 }
 273                 else{
 274                         goodGroupOut << seqName << '\t' << group << endl;
 275                 }
 276                 gobble(inputGroups);
 277         }
 278         inputGroups.close();
 279         goodGroupOut.close();
 280         badGroupOut.close();
 281
 282 }
 283
 284 //***************************************************************************************************************
 285
 286 void ScreenSeqsCommand::screenAlignReport(set<string> badSeqNames){
 287
 288         ifstream inputAlignReport;
 289         openInputFile(alignreport, inputAlignReport);
 290         string seqName, group;
 291         set<string>::iterator it;
 292
 293         string goodAlignReportFile = getRootName(alignreport) + "good" + getExtension(alignreport);
 294         string badAlignReportFile = getRootName(alignreport) + "bad" + getExtension(alignreport);
 295
 296         ofstream goodAlignReportOut;    openOutputFile(goodAlignReportFile, goodAlignReportOut);
 297         ofstream badAlignReportOut;             openOutputFile(badAlignReportFile, badAlignReportOut);
 298
 299         while (!inputAlignReport.eof()) {               //      need to copy header
 300                 char c = inputAlignReport.get();
 301                 goodAlignReportOut << c;
 302                 badAlignReportOut << c;
 303                 if (c == 10 || c == 13){        break;  }
 304         }
 305
 306         while(!inputAlignReport.eof()){
 307                 inputAlignReport >> seqName;
 308                 it = badSeqNames.find(seqName);
 309                 string line;
 310                 while (!inputAlignReport.eof()) {               //      need to copy header
 311                         char c = inputAlignReport.get();
 312                         line += c;
 313                         if (c == 10 || c == 13){        break;  }
 314                 }
 315
 316                 if(it != badSeqNames.end()){
 317                         badSeqNames.erase(it);
 318                         badAlignReportOut << seqName << '\t' << line;;
 319                 }
 320                 else{
 321                         goodAlignReportOut << seqName << '\t' << line;
 322                 }
 323                 gobble(inputAlignReport);
 324         }
 325         inputAlignReport.close();
 326         goodAlignReportOut.close();
 327         badAlignReportOut.close();
 328
 329 }
 330
 331 //***************************************************************************************************************
 332
 333