degapseqscommand.cpp

   1 /*
   2  *  degapseqscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 6/21/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "degapseqscommand.h"
  11 #include "sequence.hpp"
  12
  13 //***************************************************************************************************************
  14
  15 DegapSeqsCommand::DegapSeqsCommand(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", "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
  42                         //check for required parameters
  43                         fastafile = validParameter.validFile(parameters, "fasta", false);
  44                         if (fastafile == "not found") { fastafile = ""; m->mothurOut("fasta is a required parameter for the degap.seqs command."); m->mothurOutEndLine(); abort = true;  }
  45                         else {
  46                                 m->splitAtDash(fastafile, fastaFileNames);
  47
  48                                 //go through files and make sure they are good, if not, then disregard them
  49                                 for (int i = 0; i < fastaFileNames.size(); i++) {
  50                                         if (inputDir != "") {
  51                                                 string path = m->hasPath(fastaFileNames[i]);
  52                                                 //if the user has not given a path then, add inputdir. else leave path alone.
  53                                                 if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
  54                                         }
  55
  56                                         ifstream in;
  57                                         int ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
  58
  59                                         //if you can't open it, try default location
  60                                         if (ableToOpen == 1) {
  61                                                 if (m->getDefaultPath() != "") { //default path is set
  62                                                         string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
  63                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
  64                                                         ableToOpen = m->openInputFile(tryPath, in, "noerror");
  65                                                         fastaFileNames[i] = tryPath;
  66                                                 }
  67                                         }
  68                                         in.close();
  69
  70                                         if (ableToOpen == 1) {
  71                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
  72                                                 //erase from file list
  73                                                 fastaFileNames.erase(fastaFileNames.begin()+i);
  74                                                 i--;
  75                                         }
  76                                 }
  77
  78                                 //make sure there is at least one valid file left
  79                                 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
  80                         }
  81
  82
  83                         //if the user changes the output directory command factory will send this info to us in the output parameter
  84                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
  85                                 outputDir = "";
  86                                 outputDir += m->hasPath(fastafile); //if user entered a file with a path then preserve it
  87                         }
  88
  89                 }
  90         }
  91         catch(exception& e) {
  92                 m->errorOut(e, "DegapSeqsCommand", "DegapSeqsCommand");
  93                 exit(1);
  94         }
  95 }
  96 //**********************************************************************************************************************
  97
  98 void DegapSeqsCommand::help(){
  99         try {
 100                 m->mothurOut("The degap.seqs command reads a fastafile and removes all gap characters.\n");
 101                 m->mothurOut("The degap.seqs command parameter is fasta.\n");
 102                 m->mothurOut("The fasta parameter allows you to enter the fasta file containing your sequences, and is required. \n");
 103                 m->mothurOut("You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amzon.fasta \n");
 104                 m->mothurOut("The degap.seqs command should be in the following format: \n");
 105                 m->mothurOut("degap.seqs(fasta=yourFastaFile) \n");
 106                 m->mothurOut("Example: degap.seqs(fasta=abrecovery.align) \n");
 107                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n\n");
 108         }
 109         catch(exception& e) {
 110                 m->errorOut(e, "DegapSeqsCommand", "help");
 111                 exit(1);
 112         }
 113 }
 114
 115 //***************************************************************************************************************
 116
 117 DegapSeqsCommand::~DegapSeqsCommand(){  /*      do nothing      */      }
 118
 119 //***************************************************************************************************************
 120
 121
 122 int DegapSeqsCommand::execute(){
 123         try{
 124
 125                 if (abort == true) { return 0; }
 126
 127                 for (int s = 0; s < fastaFileNames.size(); s++) {
 128
 129                         m->mothurOut("Degapping sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
 130                         ifstream inFASTA;
 131                         m->openInputFile(fastaFileNames[s], inFASTA);
 132
 133                         ofstream outFASTA;
 134                         string tempOutputDir = outputDir;
 135                         if (outputDir == "") { tempOutputDir = m->hasPath(fastaFileNames[s]); }
 136                         string degapFile = tempOutputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "ng.fasta";
 137                         m->openOutputFile(degapFile, outFASTA);
 138
 139                         while(!inFASTA.eof()){
 140                                 if (m->control_pressed) {  inFASTA.close();  outFASTA.close(); remove(degapFile.c_str()); for (int j = 0; j < outputNames.size(); j++) {        remove(outputNames[j].c_str()); } return 0; }
 141
 142                                 Sequence currSeq(inFASTA);  m->gobble(inFASTA);
 143                                 if (currSeq.getName() != "") {
 144                                         outFASTA << ">" << currSeq.getName() << endl;
 145                                         outFASTA << currSeq.getUnaligned() << endl;
 146                                 }
 147                         }
 148                         inFASTA.close();
 149                         outFASTA.close();
 150
 151                         outputNames.push_back(degapFile);
 152
 153                         if (m->control_pressed) {  remove(degapFile.c_str()); for (int j = 0; j < outputNames.size(); j++) {    remove(outputNames[j].c_str()); } return 0; }
 154                 }
 155
 156                 m->mothurOutEndLine();
 157                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 158                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 159                 m->mothurOutEndLine();
 160
 161
 162                 return 0;
 163
 164         }
 165         catch(exception& e) {
 166                 m->errorOut(e, "DegapSeqsCommand", "execute");
 167                 exit(1);
 168         }
 169 }
 170
 171 //***************************************************************************************************************
 172