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", "outputdir","inputdir"};
  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                         map<string,string>::iterator it;
  32
  33                         //check to make sure all parameters are valid for command
  34                         for (it = parameters.begin(); it != parameters.end(); it++) {
  35                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  36                         }
  37
  38                         //if the user changes the input directory command factory will send this info to us in the output parameter
  39                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  40                         if (inputDir == "not found"){   inputDir = "";          }
  41                         else {
  42                                 string path;
  43                                 it = parameters.find("fasta");
  44                                 //user has given a template file
  45                                 if(it != parameters.end()){
  46                                         path = hasPath(it->second);
  47                                         //if the user has not given a path then, add inputdir. else leave path alone.
  48                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
  49                                 }
  50
  51                                 it = parameters.find("hard");
  52                                 //user has given a template file
  53                                 if(it != parameters.end()){
  54                                         path = hasPath(it->second);
  55                                         //if the user has not given a path then, add inputdir. else leave path alone.
  56                                         if (path == "") {       parameters["hard"] = inputDir + it->second;             }
  57                                 }
  58                         }
  59
  60                         //check for required parameters
  61                         fastafile = validParameter.validFile(parameters, "fasta", true);
  62                         if (fastafile == "not found") { mothurOut("fasta is a required parameter for the filter.seqs command."); mothurOutEndLine(); abort = true; }
  63                         else if (fastafile == "not open") { abort = true; }
  64
  65                         //if the user changes the output directory command factory will send this info to us in the output parameter
  66                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
  67                                 outputDir = "";
  68                                 outputDir += hasPath(fastafile); //if user entered a file with a path then preserve it
  69                         }
  70
  71                         //check for optional parameter and set defaults
  72                         // ...at some point should added some additional type checking...
  73
  74                         string temp;
  75                         temp = validParameter.validFile(parameters, "trump", false);                    if (temp == "not found") { temp = "*"; }
  76                         trump = temp[0];
  77
  78                         temp = validParameter.validFile(parameters, "soft", false);                             if (temp == "not found") { soft = 0; }
  79                         else {  soft = (float)atoi(temp.c_str()) / 100.0;  }
  80
  81                         hard = validParameter.validFile(parameters, "hard", true);                              if (hard == "not found") { hard = ""; }
  82                         else if (hard == "not open") { abort = true; }
  83
  84                         vertical = validParameter.validFile(parameters, "vertical", false);             if (vertical == "not found") { vertical = "T"; }
  85
  86                         numSeqs = 0;
  87
  88                         if (abort == false) {
  89
  90                                 if (soft != 0)                  {  F.setSoft(soft);             }
  91                                 if (trump != '*')               {  F.setTrump(trump);   }
  92
  93                         }
  94
  95                 }
  96
  97         }
  98         catch(exception& e) {
  99                 errorOut(e, "FilterSeqsCommand", "FilterSeqsCommand");
 100                 exit(1);
 101         }
 102 }
 103
 104 //**********************************************************************************************************************
 105
 106 void FilterSeqsCommand::help(){
 107         try {
 108                 mothurOut("The filter.seqs command reads a file containing sequences and creates a .filter and .filter.fasta file.\n");
 109                 mothurOut("The filter.seqs command parameters are fasta, trump, soft, hard and vertical. \n");
 110                 mothurOut("The fasta parameter is required.\n");
 111                 mothurOut("The trump parameter .... The default is ...\n");
 112                 mothurOut("The soft parameter .... The default is ....\n");
 113                 mothurOut("The hard parameter .... The default is ....\n");
 114                 mothurOut("The vertical parameter .... The default is T.\n");
 115                 mothurOut("The filter.seqs command should be in the following format: \n");
 116                 mothurOut("filter.seqs(fasta=yourFastaFile, trump=yourTrump, soft=yourSoft, hard=yourHard, vertical=yourVertical) \n");
 117                 mothurOut("Example filter.seqs(fasta=abrecovery.fasta, trump=..., soft=..., hard=..., vertical=T).\n");
 118                 mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
 119
 120         }
 121         catch(exception& e) {
 122                 errorOut(e, "FilterSeqsCommand", "help");
 123                 exit(1);
 124         }
 125 }
 126
 127 /**************************************************************************************/
 128
 129 int FilterSeqsCommand::execute() {
 130         try {
 131
 132                 if (abort == true) { return 0; }
 133
 134                 ifstream inFASTA;
 135                 openInputFile(fastafile, inFASTA);
 136
 137                 Sequence testSeq(inFASTA);
 138                 alignmentLength = testSeq.getAlignLength();
 139                 inFASTA.seekg(0);
 140
 141                 F.setLength(alignmentLength);
 142
 143                 if(soft != 0 || isTrue(vertical)){
 144                         F.initialize();
 145                 }
 146
 147                 if(hard.compare("") != 0)       {       F.doHard(hard);         }
 148                 else                                            {       F.setFilter(string(alignmentLength, '1'));      }
 149
 150                 if(trump != '*' || isTrue(vertical) || soft != 0){
 151                         while(!inFASTA.eof()){  //read through and create the filter...
 152                                 Sequence seq(inFASTA);
 153                                 if (seq.getName() != "") {
 154                                         if(trump != '*'){       F.doTrump(seq); }
 155                                         if(isTrue(vertical) || soft != 0){      F.getFreqs(seq);        }
 156                                         numSeqs++;
 157                                         cout.flush();
 158                                 }
 159                         }
 160
 161                 }
 162                 inFASTA.close();
 163                 F.setNumSeqs(numSeqs);
 164
 165
 166                 if(isTrue(vertical) == 1)       {       F.doVertical(); }
 167                 if(soft != 0)                           {       F.doSoft();             }
 168
 169                 filter = F.getFilter();
 170
 171                 ofstream outFilter;
 172                 string filterFile = outputDir + getRootName(getSimpleName(fastafile)) + "filter";
 173                 openOutputFile(filterFile, outFilter);
 174                 outFilter << filter << endl;
 175                 outFilter.close();
 176
 177                 ifstream inFasta2;
 178                 openInputFile(fastafile, inFasta2);
 179                 string filteredFasta = outputDir + getRootName(getSimpleName(fastafile)) + "filter.fasta";
 180                 ofstream outFASTA;
 181                 openOutputFile(filteredFasta, outFASTA);
 182
 183                 numSeqs = 0;
 184                 while(!inFasta2.eof()){
 185                         Sequence seq(inFasta2);
 186                         if (seq.getName() != "") {
 187                                 string align = seq.getAligned();
 188                                 string filterSeq = "";
 189
 190                                 for(int j=0;j<alignmentLength;j++){
 191                                         if(filter[j] == '1'){
 192                                                 filterSeq += align[j];
 193                                         }
 194                                 }
 195
 196                                 outFASTA << '>' << seq.getName() << endl << filterSeq << endl;
 197                                 numSeqs++;
 198                         }
 199                         gobble(inFasta2);
 200                 }
 201                 outFASTA.close();
 202                 inFasta2.close();
 203
 204
 205                 int filteredLength = 0;
 206                 for(int i=0;i<alignmentLength;i++){
 207                         if(filter[i] == '1'){   filteredLength++;       }
 208                 }
 209
 210                 mothurOutEndLine();
 211                 mothurOut("Length of filtered alignment: " + toString(filteredLength)); mothurOutEndLine();
 212                 mothurOut("Number of columns removed: " + toString((alignmentLength-filteredLength))); mothurOutEndLine();
 213                 mothurOut("Length of the original alignment: " + toString(alignmentLength)); mothurOutEndLine();
 214                 mothurOut("Number of sequences used to construct filter: " + toString(numSeqs)); mothurOutEndLine();
 215
 216                 return 0;
 217
 218         }
 219         catch(exception& e) {
 220                 errorOut(e, "FilterSeqsCommand", "execute");
 221                 exit(1);
 222         }
 223 }
 224
 225 /**************************************************************************************/