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::setParameters(){
  15         try {
  16                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfasta);
  17                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  18                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  19
  20                 vector<string> myArray;
  21                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  22                 return myArray;
  23         }
  24         catch(exception& e) {
  25                 m->errorOut(e, "DegapSeqsCommand", "setParameters");
  26                 exit(1);
  27         }
  28 }
  29 //**********************************************************************************************************************
  30 string DegapSeqsCommand::getHelpString(){
  31         try {
  32                 string helpString = "";
  33                 helpString += "The degap.seqs command reads a fastafile and removes all gap characters.\n";
  34                 helpString += "The degap.seqs command parameter is fasta.\n";
  35                 helpString += "The fasta parameter allows you to enter the fasta file containing your sequences, and is required unless you have a valid current fasta file. \n";
  36                 helpString += "You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amzon.fasta \n";
  37                 helpString += "The degap.seqs command should be in the following format: \n";
  38                 helpString += "degap.seqs(fasta=yourFastaFile) \n";
  39                 helpString += "Example: degap.seqs(fasta=abrecovery.align) \n";
  40                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n";
  41                 return helpString;
  42         }
  43         catch(exception& e) {
  44                 m->errorOut(e, "DegapSeqsCommand", "getHelpString");
  45                 exit(1);
  46         }
  47 }
  48 //**********************************************************************************************************************
  49 DegapSeqsCommand::DegapSeqsCommand(){
  50         try {
  51                 abort = true; calledHelp = true;
  52                 setParameters();
  53                 vector<string> tempOutNames;
  54                 outputTypes["fasta"] = tempOutNames;
  55         }
  56         catch(exception& e) {
  57                 m->errorOut(e, "DegapSeqsCommand", "DegapSeqsCommand");
  58                 exit(1);
  59         }
  60 }
  61 //***************************************************************************************************************
  62 DegapSeqsCommand::DegapSeqsCommand(string option)  {
  63         try {
  64                 abort = false; calledHelp = false;
  65
  66                 //allow user to run help
  67                 if(option == "help") { help(); abort = true; calledHelp = true; }
  68
  69                 else {
  70                         vector<string> myArray = setParameters();
  71
  72                         OptionParser parser(option);
  73                         map<string,string> parameters = parser.getParameters();
  74
  75                         ValidParameters validParameter;
  76                         map<string,string>::iterator it;
  77
  78                         //check to make sure all parameters are valid for command
  79                         for (it = parameters.begin(); it != parameters.end(); it++) {
  80                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  81                         }
  82
  83
  84                         //initialize outputTypes
  85                         vector<string> tempOutNames;
  86                         outputTypes["fasta"] = tempOutNames;
  87
  88                         //if the user changes the input directory command factory will send this info to us in the output parameter
  89                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  90                         if (inputDir == "not found"){   inputDir = "";          }
  91
  92                         //check for required parameters
  93                         fastafile = validParameter.validFile(parameters, "fasta", false);
  94                         if (fastafile == "not found") {
  95                                 fastafile = m->getFastaFile();
  96                                 if (fastafile != "") { fastaFileNames.push_back(fastafile); m->mothurOut("Using " + fastafile + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
  97                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
  98                         }
  99                         else {
 100                                 m->splitAtDash(fastafile, fastaFileNames);
 101
 102                                 //go through files and make sure they are good, if not, then disregard them
 103                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 104                                         if (inputDir != "") {
 105                                                 string path = m->hasPath(fastaFileNames[i]);
 106                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 107                                                 if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 108                                         }
 109
 110                                         ifstream in;
 111                                         int ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 112
 113                                         //if you can't open it, try default location
 114                                         if (ableToOpen == 1) {
 115                                                 if (m->getDefaultPath() != "") { //default path is set
 116                                                         string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 117                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 118                                                         ifstream in2;
 119                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 120                                                         in2.close();
 121                                                         fastaFileNames[i] = tryPath;
 122                                                 }
 123                                         }
 124
 125                                         //if you can't open it, try default location
 126                                         if (ableToOpen == 1) {
 127                                                 if (m->getOutputDir() != "") { //default path is set
 128                                                         string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 129                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 130                                                         ifstream in2;
 131                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 132                                                         in2.close();
 133                                                         fastaFileNames[i] = tryPath;
 134                                                 }
 135                                         }
 136
 137                                         in.close();
 138
 139                                         if (ableToOpen == 1) {
 140                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 141                                                 //erase from file list
 142                                                 fastaFileNames.erase(fastaFileNames.begin()+i);
 143                                                 i--;
 144                                         }
 145                                 }
 146
 147                                 //make sure there is at least one valid file left
 148                                 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 149                         }
 150
 151
 152                         //if the user changes the output directory command factory will send this info to us in the output parameter
 153                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 154                                 outputDir = "";
 155                                 outputDir += m->hasPath(fastafile); //if user entered a file with a path then preserve it
 156                         }
 157
 158                 }
 159         }
 160         catch(exception& e) {
 161                 m->errorOut(e, "DegapSeqsCommand", "DegapSeqsCommand");
 162                 exit(1);
 163         }
 164 }
 165 //***************************************************************************************************************
 166 int DegapSeqsCommand::execute(){
 167         try{
 168
 169                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 170
 171                 for (int s = 0; s < fastaFileNames.size(); s++) {
 172
 173                         m->mothurOut("Degapping sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
 174                         ifstream inFASTA;
 175                         m->openInputFile(fastaFileNames[s], inFASTA);
 176
 177                         ofstream outFASTA;
 178                         string tempOutputDir = outputDir;
 179                         if (outputDir == "") { tempOutputDir = m->hasPath(fastaFileNames[s]); }
 180                         string degapFile = tempOutputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "ng.fasta";
 181                         m->openOutputFile(degapFile, outFASTA);
 182
 183                         while(!inFASTA.eof()){
 184                                 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; }
 185
 186                                 Sequence currSeq(inFASTA);  m->gobble(inFASTA);
 187                                 if (currSeq.getName() != "") {
 188                                         outFASTA << ">" << currSeq.getName() << endl;
 189                                         outFASTA << currSeq.getUnaligned() << endl;
 190                                 }
 191                         }
 192                         inFASTA.close();
 193                         outFASTA.close();
 194
 195                         outputNames.push_back(degapFile); outputTypes["fasta"].push_back(degapFile);
 196
 197                         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; }
 198                 }
 199
 200                 //set fasta file as new current fastafile
 201                 string current = "";
 202                 itTypes = outputTypes.find("fasta");
 203                 if (itTypes != outputTypes.end()) {
 204                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 205                 }
 206
 207                 m->mothurOutEndLine();
 208                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 209                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 210                 m->mothurOutEndLine();
 211
 212
 213                 return 0;
 214
 215         }
 216         catch(exception& e) {
 217                 m->errorOut(e, "DegapSeqsCommand", "execute");
 218                 exit(1);
 219         }
 220 }
 221
 222 //***************************************************************************************************************
 223