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 vector<string> DegapSeqsCommand::getValidParameters(){
  15         try {
  16                 string Array[] =  {"fasta", "outputdir","inputdir"};
  17                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  18                 return myArray;
  19         }
  20         catch(exception& e) {
  21                 m->errorOut(e, "DegapSeqsCommand", "getValidParameters");
  22                 exit(1);
  23         }
  24 }
  25 //**********************************************************************************************************************
  26 DegapSeqsCommand::DegapSeqsCommand(){
  27         try {
  28                 abort = true;
  29                 //initialize outputTypes
  30                 vector<string> tempOutNames;
  31                 outputTypes["fasta"] = tempOutNames;
  32         }
  33         catch(exception& e) {
  34                 m->errorOut(e, "DegapSeqsCommand", "DegapSeqsCommand");
  35                 exit(1);
  36         }
  37 }
  38 //**********************************************************************************************************************
  39 vector<string> DegapSeqsCommand::getRequiredParameters(){
  40         try {
  41                 string Array[] =  {"fasta"};
  42                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  43                 return myArray;
  44         }
  45         catch(exception& e) {
  46                 m->errorOut(e, "DegapSeqsCommand", "getRequiredParameters");
  47                 exit(1);
  48         }
  49 }
  50 //**********************************************************************************************************************
  51 vector<string> DegapSeqsCommand::getRequiredFiles(){
  52         try {
  53                 vector<string> myArray;
  54                 return myArray;
  55         }
  56         catch(exception& e) {
  57                 m->errorOut(e, "DegapSeqsCommand", "getRequiredFiles");
  58                 exit(1);
  59         }
  60 }
  61 //***************************************************************************************************************
  62 DegapSeqsCommand::DegapSeqsCommand(string option)  {
  63         try {
  64                 abort = false;
  65
  66                 //allow user to run help
  67                 if(option == "help") { help(); abort = true; }
  68
  69                 else {
  70                         //valid paramters for this command
  71                         string Array[] =  {"fasta", "outputdir","inputdir"};
  72                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  73
  74                         OptionParser parser(option);
  75                         map<string,string> parameters = parser.getParameters();
  76
  77                         ValidParameters validParameter;
  78                         map<string,string>::iterator it;
  79
  80                         //check to make sure all parameters are valid for command
  81                         for (it = parameters.begin(); it != parameters.end(); it++) {
  82                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  83                         }
  84
  85
  86                         //initialize outputTypes
  87                         vector<string> tempOutNames;
  88                         outputTypes["fasta"] = tempOutNames;
  89
  90                         //if the user changes the input directory command factory will send this info to us in the output parameter
  91                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  92                         if (inputDir == "not found"){   inputDir = "";          }
  93
  94                         //check for required parameters
  95                         fastafile = validParameter.validFile(parameters, "fasta", false);
  96                         if (fastafile == "not found") { fastafile = ""; m->mothurOut("fasta is a required parameter for the degap.seqs command."); m->mothurOutEndLine(); abort = true;  }
  97                         else {
  98                                 m->splitAtDash(fastafile, fastaFileNames);
  99
 100                                 //go through files and make sure they are good, if not, then disregard them
 101                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 102                                         if (inputDir != "") {
 103                                                 string path = m->hasPath(fastaFileNames[i]);
 104                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 105                                                 if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 106                                         }
 107
 108                                         ifstream in;
 109                                         int ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 110
 111                                         //if you can't open it, try default location
 112                                         if (ableToOpen == 1) {
 113                                                 if (m->getDefaultPath() != "") { //default path is set
 114                                                         string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 115                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 116                                                         ifstream in2;
 117                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 118                                                         in2.close();
 119                                                         fastaFileNames[i] = tryPath;
 120                                                 }
 121                                         }
 122
 123                                         //if you can't open it, try default location
 124                                         if (ableToOpen == 1) {
 125                                                 if (m->getOutputDir() != "") { //default path is set
 126                                                         string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 127                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 128                                                         ifstream in2;
 129                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 130                                                         in2.close();
 131                                                         fastaFileNames[i] = tryPath;
 132                                                 }
 133                                         }
 134
 135                                         in.close();
 136
 137                                         if (ableToOpen == 1) {
 138                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 139                                                 //erase from file list
 140                                                 fastaFileNames.erase(fastaFileNames.begin()+i);
 141                                                 i--;
 142                                         }
 143                                 }
 144
 145                                 //make sure there is at least one valid file left
 146                                 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 147                         }
 148
 149
 150                         //if the user changes the output directory command factory will send this info to us in the output parameter
 151                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 152                                 outputDir = "";
 153                                 outputDir += m->hasPath(fastafile); //if user entered a file with a path then preserve it
 154                         }
 155
 156                 }
 157         }
 158         catch(exception& e) {
 159                 m->errorOut(e, "DegapSeqsCommand", "DegapSeqsCommand");
 160                 exit(1);
 161         }
 162 }
 163 //**********************************************************************************************************************
 164
 165 void DegapSeqsCommand::help(){
 166         try {
 167                 m->mothurOut("The degap.seqs command reads a fastafile and removes all gap characters.\n");
 168                 m->mothurOut("The degap.seqs command parameter is fasta.\n");
 169                 m->mothurOut("The fasta parameter allows you to enter the fasta file containing your sequences, and is required. \n");
 170                 m->mothurOut("You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amzon.fasta \n");
 171                 m->mothurOut("The degap.seqs command should be in the following format: \n");
 172                 m->mothurOut("degap.seqs(fasta=yourFastaFile) \n");
 173                 m->mothurOut("Example: degap.seqs(fasta=abrecovery.align) \n");
 174                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n\n");
 175         }
 176         catch(exception& e) {
 177                 m->errorOut(e, "DegapSeqsCommand", "help");
 178                 exit(1);
 179         }
 180 }
 181
 182 //***************************************************************************************************************
 183
 184 DegapSeqsCommand::~DegapSeqsCommand(){  /*      do nothing      */      }
 185
 186 //***************************************************************************************************************
 187
 188
 189 int DegapSeqsCommand::execute(){
 190         try{
 191
 192                 if (abort == true) { return 0; }
 193
 194                 for (int s = 0; s < fastaFileNames.size(); s++) {
 195
 196                         m->mothurOut("Degapping sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
 197                         ifstream inFASTA;
 198                         m->openInputFile(fastaFileNames[s], inFASTA);
 199
 200                         ofstream outFASTA;
 201                         string tempOutputDir = outputDir;
 202                         if (outputDir == "") { tempOutputDir = m->hasPath(fastaFileNames[s]); }
 203                         string degapFile = tempOutputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "ng.fasta";
 204                         m->openOutputFile(degapFile, outFASTA);
 205
 206                         while(!inFASTA.eof()){
 207                                 if (m->control_pressed) {   outputTypes.clear(); inFASTA.close();  outFASTA.close(); remove(degapFile.c_str()); for (int j = 0; j < outputNames.size(); j++) {  remove(outputNames[j].c_str()); } return 0; }
 208
 209                                 Sequence currSeq(inFASTA);  m->gobble(inFASTA);
 210                                 if (currSeq.getName() != "") {
 211                                         outFASTA << ">" << currSeq.getName() << endl;
 212                                         outFASTA << currSeq.getUnaligned() << endl;
 213                                 }
 214                         }
 215                         inFASTA.close();
 216                         outFASTA.close();
 217
 218                         outputNames.push_back(degapFile); outputTypes["fasta"].push_back(degapFile);
 219
 220                         if (m->control_pressed) {  outputTypes.clear(); remove(degapFile.c_str()); for (int j = 0; j < outputNames.size(); j++) {       remove(outputNames[j].c_str()); } return 0; }
 221                 }
 222
 223                 m->mothurOutEndLine();
 224                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 225                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 226                 m->mothurOutEndLine();
 227
 228
 229                 return 0;
 230
 231         }
 232         catch(exception& e) {
 233                 m->errorOut(e, "DegapSeqsCommand", "execute");
 234                 exit(1);
 235         }
 236 }
 237
 238 //***************************************************************************************************************
 239