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