filterseqscommand.cpp

   1 /*
   2  *  filterseqscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Thomas Ryabin on 5/4/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "filterseqscommand.h"
  11 #include "sequence.hpp"
  12
  13 /**************************************************************************************/
  14
  15 FilterSeqsCommand::FilterSeqsCommand(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 Array[] =  {"fasta", "trump", "soft", "hard", "vertical"};
  25                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  26
  27                         OptionParser parser(option);
  28                         map<string,string> parameters = parser.getParameters();
  29
  30                         ValidParameters validParameter;
  31
  32                         //check to make sure all parameters are valid for command
  33                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  34                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  35                         }
  36
  37                         //check for required parameters
  38                         fastafile = validParameter.validFile(parameters, "fasta", true);
  39                         if (fastafile == "not found") { mothurOut("fasta is a required parameter for the filter.seqs command."); mothurOutEndLine(); abort = true; }
  40                         else if (fastafile == "not open") { abort = true; }
  41
  42                         //check for optional parameter and set defaults
  43                         // ...at some point should added some additional type checking...
  44
  45                         string temp;
  46                         temp = validParameter.validFile(parameters, "trump", false);                    if (temp == "not found") { temp = "*"; }
  47                         trump = temp[0];
  48
  49                         temp = validParameter.validFile(parameters, "soft", false);                             if (temp == "not found") { soft = 0; }
  50                         else {  soft = (float)atoi(temp.c_str()) / 100.0;  }
  51
  52                         hard = validParameter.validFile(parameters, "hard", true);                              if (hard == "not found") { hard = ""; }
  53                         else if (hard == "not open") { abort = true; }
  54
  55                         vertical = validParameter.validFile(parameters, "vertical", false);             if (vertical == "not found") { vertical = "T"; }
  56
  57                         numSeqs = 0;
  58
  59                         if (abort == false) {
  60
  61                                 if (soft != 0)                  {  F.setSoft(soft);             }
  62                                 if (trump != '*')               {  F.setTrump(trump);   }
  63
  64                         }
  65
  66                 }
  67
  68         }
  69         catch(exception& e) {
  70                 errorOut(e, "FilterSeqsCommand", "FilterSeqsCommand");
  71                 exit(1);
  72         }
  73 }
  74
  75 //**********************************************************************************************************************
  76
  77 void FilterSeqsCommand::help(){
  78         try {
  79                 mothurOut("The filter.seqs command reads a file containing sequences and creates a .filter and .filter.fasta file.\n");
  80                 mothurOut("The filter.seqs command parameters are fasta, trump, soft, hard and vertical. \n");
  81                 mothurOut("The fasta parameter is required.\n");
  82                 mothurOut("The trump parameter .... The default is ...\n");
  83                 mothurOut("The soft parameter .... The default is ....\n");
  84                 mothurOut("The hard parameter .... The default is ....\n");
  85                 mothurOut("The vertical parameter .... The default is T.\n");
  86                 mothurOut("The filter.seqs command should be in the following format: \n");
  87                 mothurOut("filter.seqs(fasta=yourFastaFile, trump=yourTrump, soft=yourSoft, hard=yourHard, vertical=yourVertical) \n");
  88                 mothurOut("Example filter.seqs(fasta=abrecovery.fasta, trump=..., soft=..., hard=..., vertical=T).\n");
  89                 mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
  90
  91         }
  92         catch(exception& e) {
  93                 errorOut(e, "FilterSeqsCommand", "help");
  94                 exit(1);
  95         }
  96 }
  97
  98 /**************************************************************************************/
  99
 100 int FilterSeqsCommand::execute() {
 101         try {
 102
 103                 if (abort == true) { return 0; }
 104
 105                 ifstream inFASTA;
 106                 openInputFile(fastafile, inFASTA);
 107
 108                 Sequence testSeq(inFASTA);
 109                 alignmentLength = testSeq.getAlignLength();
 110                 inFASTA.seekg(0);
 111
 112                 F.setLength(alignmentLength);
 113
 114                 if(soft != 0 || isTrue(vertical)){
 115                         F.initialize();
 116                 }
 117
 118                 if(hard.compare("") != 0)       {       F.doHard(hard);         }
 119                 else                                            {       F.setFilter(string(alignmentLength, '1'));      }
 120
 121                 if(trump != '*' || isTrue(vertical) || soft != 0){
 122                         while(!inFASTA.eof()){  //read through and create the filter...
 123                                 Sequence seq(inFASTA);
 124                                 if (seq.getName() != "") {
 125                                         if(trump != '*'){       F.doTrump(seq); }
 126                                         if(isTrue(vertical) || soft != 0){      F.getFreqs(seq);        }
 127                                         numSeqs++;
 128                                         cout.flush();
 129                                 }
 130                         }
 131
 132                 }
 133                 inFASTA.close();
 134                 F.setNumSeqs(numSeqs);
 135
 136
 137                 if(isTrue(vertical) == 1)       {       F.doVertical(); }
 138                 if(soft != 0)                           {       F.doSoft();             }
 139
 140                 filter = F.getFilter();
 141
 142                 ofstream outFilter;
 143                 string filterFile = getRootName(fastafile) + "filter";
 144                 openOutputFile(filterFile, outFilter);
 145                 outFilter << filter << endl;
 146                 outFilter.close();
 147
 148                 ifstream inFasta2;
 149                 openInputFile(fastafile, inFasta2);
 150                 string filteredFasta = getRootName(fastafile) + "filter.fasta";
 151                 ofstream outFASTA;
 152                 openOutputFile(filteredFasta, outFASTA);
 153
 154                 numSeqs = 0;
 155                 while(!inFasta2.eof()){
 156                         Sequence seq(inFasta2);
 157                         if (seq.getName() != "") {
 158                                 string align = seq.getAligned();
 159                                 string filterSeq = "";
 160
 161                                 for(int j=0;j<alignmentLength;j++){
 162                                         if(filter[j] == '1'){
 163                                                 filterSeq += align[j];
 164                                         }
 165                                 }
 166
 167                                 outFASTA << '>' << seq.getName() << endl << filterSeq << endl;
 168                                 numSeqs++;
 169                         }
 170                         gobble(inFasta2);
 171                 }
 172                 outFASTA.close();
 173                 inFasta2.close();
 174
 175
 176                 int filteredLength = 0;
 177                 for(int i=0;i<alignmentLength;i++){
 178                         if(filter[i] == '1'){   filteredLength++;       }
 179                 }
 180
 181                 mothurOutEndLine();
 182                 mothurOut("Length of filtered alignment: " + toString(filteredLength)); mothurOutEndLine();
 183                 mothurOut("Number of columns removed: " + toString((alignmentLength-filteredLength))); mothurOutEndLine();
 184                 mothurOut("Length of the original alignment: " + toString(alignmentLength)); mothurOutEndLine();
 185                 mothurOut("Number of sequences used to construct filter: " + toString(numSeqs)); mothurOutEndLine();
 186
 187                 return 0;
 188
 189         }
 190         catch(exception& e) {
 191                 errorOut(e, "FilterSeqsCommand", "execute");
 192                 exit(1);
 193         }
 194 }
 195
 196 /**************************************************************************************/