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 string DegapSeqsCommand::getOutputFileNameTag(string type, string inputName=""){
  50         try {
  51         string outputFileName = "";
  52                 map<string, vector<string> >::iterator it;
  53
  54         //is this a type this command creates
  55         it = outputTypes.find(type);
  56         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  57         else {
  58             if (type == "fasta") {  outputFileName =  "ng.fasta"; }
  59             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  60         }
  61         return outputFileName;
  62         }
  63         catch(exception& e) {
  64                 m->errorOut(e, "DegapSeqsCommand", "getOutputFileNameTag");
  65                 exit(1);
  66         }
  67 }
  68 //**********************************************************************************************************************
  69 DegapSeqsCommand::DegapSeqsCommand(){
  70         try {
  71                 abort = true; calledHelp = true;
  72                 setParameters();
  73                 vector<string> tempOutNames;
  74                 outputTypes["fasta"] = tempOutNames;
  75         }
  76         catch(exception& e) {
  77                 m->errorOut(e, "DegapSeqsCommand", "DegapSeqsCommand");
  78                 exit(1);
  79         }
  80 }
  81 //***************************************************************************************************************
  82 DegapSeqsCommand::DegapSeqsCommand(string option)  {
  83         try {
  84                 abort = false; calledHelp = false;
  85
  86                 //allow user to run help
  87                 if(option == "help") { help(); abort = true; calledHelp = true; }
  88                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  89
  90                 else {
  91                         vector<string> myArray = setParameters();
  92
  93                         OptionParser parser(option);
  94                         map<string,string> parameters = parser.getParameters();
  95
  96                         ValidParameters validParameter;
  97                         map<string,string>::iterator it;
  98
  99                         //check to make sure all parameters are valid for command
 100                         for (it = parameters.begin(); it != parameters.end(); it++) {
 101                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 102                         }
 103
 104
 105                         //initialize outputTypes
 106                         vector<string> tempOutNames;
 107                         outputTypes["fasta"] = tempOutNames;
 108
 109                         //if the user changes the input directory command factory will send this info to us in the output parameter
 110                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 111                         if (inputDir == "not found"){   inputDir = "";          }
 112
 113                         //check for required parameters
 114                         fastafile = validParameter.validFile(parameters, "fasta", false);
 115                         if (fastafile == "not found") {
 116                                 fastafile = m->getFastaFile();
 117                                 if (fastafile != "") { fastaFileNames.push_back(fastafile); m->mothurOut("Using " + fastafile + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 118                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 119                         }
 120                         else {
 121                                 m->splitAtDash(fastafile, fastaFileNames);
 122
 123                                 //go through files and make sure they are good, if not, then disregard them
 124                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 125
 126                                         bool ignore = false;
 127                                         if (fastaFileNames[i] == "current") {
 128                                                 fastaFileNames[i] = m->getFastaFile();
 129                                                 if (fastaFileNames[i] != "") {  m->mothurOut("Using " + fastaFileNames[i] + " as input file for the fasta parameter where you had given current."); m->mothurOutEndLine(); }
 130                                                 else {
 131                                                         m->mothurOut("You have no current fastafile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 132                                                         //erase from file list
 133                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 134                                                         i--;
 135                                                 }
 136                                         }
 137
 138                                         if (!ignore) {
 139                                                 if (inputDir != "") {
 140                                                         string path = m->hasPath(fastaFileNames[i]);
 141                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 142                                                         if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 143                                                 }
 144
 145                                                 ifstream in;
 146                                                 int ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 147
 148                                                 //if you can't open it, try default location
 149                                                 if (ableToOpen == 1) {
 150                                                         if (m->getDefaultPath() != "") { //default path is set
 151                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 152                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 153                                                                 ifstream in2;
 154                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 155                                                                 in2.close();
 156                                                                 fastaFileNames[i] = tryPath;
 157                                                         }
 158                                                 }
 159
 160                                                 //if you can't open it, try default location
 161                                                 if (ableToOpen == 1) {
 162                                                         if (m->getOutputDir() != "") { //default path is set
 163                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 164                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 165                                                                 ifstream in2;
 166                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 167                                                                 in2.close();
 168                                                                 fastaFileNames[i] = tryPath;
 169                                                         }
 170                                                 }
 171
 172                                                 in.close();
 173
 174                                                 if (ableToOpen == 1) {
 175                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 176                                                         //erase from file list
 177                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 178                                                         i--;
 179                                                 }else { m->setFastaFile(fastaFileNames[i]); }
 180                                         }
 181                                 }
 182
 183                                 //make sure there is at least one valid file left
 184                                 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 185                         }
 186
 187
 188                         //if the user changes the output directory command factory will send this info to us in the output parameter
 189                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 190                                 outputDir = "";
 191                                 outputDir += m->hasPath(fastafile); //if user entered a file with a path then preserve it
 192                         }
 193
 194                 }
 195         }
 196         catch(exception& e) {
 197                 m->errorOut(e, "DegapSeqsCommand", "DegapSeqsCommand");
 198                 exit(1);
 199         }
 200 }
 201 //***************************************************************************************************************
 202 int DegapSeqsCommand::execute(){
 203         try{
 204
 205                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 206
 207                 for (int s = 0; s < fastaFileNames.size(); s++) {
 208
 209                         m->mothurOut("Degapping sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
 210                         ifstream inFASTA;
 211                         m->openInputFile(fastaFileNames[s], inFASTA);
 212
 213                         ofstream outFASTA;
 214                         string tempOutputDir = outputDir;
 215                         if (outputDir == "") { tempOutputDir = m->hasPath(fastaFileNames[s]); }
 216                         string degapFile = tempOutputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + getOutputFileNameTag("fasta");
 217                         m->openOutputFile(degapFile, outFASTA);
 218
 219                         while(!inFASTA.eof()){
 220                                 if (m->control_pressed) {   outputTypes.clear(); inFASTA.close();  outFASTA.close(); m->mothurRemove(degapFile); for (int j = 0; j < outputNames.size(); j++) { m->mothurRemove(outputNames[j]);        } return 0; }
 221
 222                                 Sequence currSeq(inFASTA);  m->gobble(inFASTA);
 223                                 if (currSeq.getName() != "") {
 224                                         outFASTA << ">" << currSeq.getName() << endl;
 225                                         outFASTA << currSeq.getUnaligned() << endl;
 226                                 }
 227                         }
 228                         inFASTA.close();
 229                         outFASTA.close();
 230
 231                         outputNames.push_back(degapFile); outputTypes["fasta"].push_back(degapFile);
 232
 233                         if (m->control_pressed) {  outputTypes.clear(); m->mothurRemove(degapFile); for (int j = 0; j < outputNames.size(); j++) {      m->mothurRemove(outputNames[j]);        } return 0; }
 234                 }
 235
 236                 //set fasta file as new current fastafile
 237                 string current = "";
 238                 itTypes = outputTypes.find("fasta");
 239                 if (itTypes != outputTypes.end()) {
 240                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 241                 }
 242
 243                 m->mothurOutEndLine();
 244                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 245                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 246                 m->mothurOutEndLine();
 247
 248
 249                 return 0;
 250
 251         }
 252         catch(exception& e) {
 253                 m->errorOut(e, "DegapSeqsCommand", "execute");
 254                 exit(1);
 255         }
 256 }
 257
 258 //***************************************************************************************************************
 259