deuniqueseqscommand.cpp

   1 /*
   2  *  deuniqueseqscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 10/19/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "deuniqueseqscommand.h"
  11 #include "sequence.hpp"
  12
  13 //**********************************************************************************************************************
  14 vector<string> DeUniqueSeqsCommand::setParameters(){
  15         try {
  16                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfasta);
  17                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pname);
  18                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  19                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  20
  21                 vector<string> myArray;
  22                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  23                 return myArray;
  24         }
  25         catch(exception& e) {
  26                 m->errorOut(e, "DeUniqueSeqsCommand", "setParameters");
  27                 exit(1);
  28         }
  29 }
  30 //**********************************************************************************************************************
  31 string DeUniqueSeqsCommand::getHelpString(){
  32         try {
  33                 string helpString = "";
  34                 helpString += "The deunique.seqs command reads a fastafile and namefile, and creates a fastafile containing all the sequences.\n";
  35                 helpString += "The deunique.seqs command parameters are fasta and name, both are required, unless you have valid current name and fasta files.\n";
  36                 helpString += "The deunique.seqs command should be in the following format: \n";
  37                 helpString += "deunique.seqs(fasta=yourFastaFile, name=yourNameFile) \n";
  38                 helpString += "Example deunique.seqs(fasta=abrecovery.unique.fasta, name=abrecovery.names).\n";
  39                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
  40                 return helpString;
  41         }
  42         catch(exception& e) {
  43                 m->errorOut(e, "DeUniqueSeqsCommand", "getHelpString");
  44                 exit(1);
  45         }
  46 }
  47 //**********************************************************************************************************************
  48 DeUniqueSeqsCommand::DeUniqueSeqsCommand(){
  49         try {
  50                 abort = true; calledHelp = true;
  51                 setParameters();
  52                 vector<string> tempOutNames;
  53                 outputTypes["fasta"] = tempOutNames;
  54         }
  55         catch(exception& e) {
  56                 m->errorOut(e, "DeUniqueSeqsCommand", "DeconvoluteCommand");
  57                 exit(1);
  58         }
  59 }
  60 /**************************************************************************************/
  61 DeUniqueSeqsCommand::DeUniqueSeqsCommand(string option)  {
  62         try {
  63                 abort = false; calledHelp = false;
  64
  65                 //allow user to run help
  66                 if(option == "help") { help(); abort = true; calledHelp = true; }
  67                 else if(option == "citation") { citation(); 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                         //initialize outputTypes
  84                         vector<string> tempOutNames;
  85                         outputTypes["fasta"] = tempOutNames;
  86
  87                         //if the user changes the input directory command factory will send this info to us in the output parameter
  88                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  89                         if (inputDir == "not found"){   inputDir = "";          }
  90                         else {
  91                                 string path;
  92                                 it = parameters.find("fasta");
  93                                 //user has given a template file
  94                                 if(it != parameters.end()){
  95                                         path = m->hasPath(it->second);
  96                                         //if the user has not given a path then, add inputdir. else leave path alone.
  97                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
  98                                 }
  99
 100                                 it = parameters.find("name");
 101                                 //user has given a template file
 102                                 if(it != parameters.end()){
 103                                         path = m->hasPath(it->second);
 104                                         //if the user has not given a path then, add inputdir. else leave path alone.
 105                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 106                                 }
 107                         }
 108
 109
 110                         //check for required parameters
 111                         fastaFile = validParameter.validFile(parameters, "fasta", true);
 112                         if (fastaFile == "not open") { abort = true; }
 113                         else if (fastaFile == "not found") {
 114                                 fastaFile = m->getFastaFile();
 115                                 if (fastaFile != "") { m->mothurOut("Using " + fastaFile + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 116                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 117                         }
 118
 119                         //if the user changes the output directory command factory will send this info to us in the output parameter
 120                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 121                                 outputDir = "";
 122                                 outputDir += m->hasPath(fastaFile); //if user entered a file with a path then preserve it
 123                         }
 124
 125                         nameFile = validParameter.validFile(parameters, "name", true);
 126                         if (nameFile == "not open") { abort = true; }
 127                         else if (nameFile == "not found"){
 128                                 nameFile = m->getNameFile();
 129                                 if (nameFile != "") { m->mothurOut("Using " + nameFile + " as input file for the name parameter."); m->mothurOutEndLine(); }
 130                                 else {  m->mothurOut("You have no current namefile and the name parameter is required."); m->mothurOutEndLine(); abort = true; }
 131                         }
 132                 }
 133
 134         }
 135         catch(exception& e) {
 136                 m->errorOut(e, "DeUniqueSeqsCommand", "DeUniqueSeqsCommand");
 137                 exit(1);
 138         }
 139 }
 140 /**************************************************************************************/
 141 int DeUniqueSeqsCommand::execute() {
 142         try {
 143
 144                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 145
 146                 //prepare filenames and open files
 147                 ofstream out;
 148                 string outFastaFile = m->getRootName(m->getSimpleName(fastaFile));
 149                 int pos = outFastaFile.find("unique");
 150                 if (pos != string::npos) {
 151                         outFastaFile = outputDir + outFastaFile.substr(0, pos) + "redundant" + m->getExtension(fastaFile);
 152                 }else{
 153                         outFastaFile = outputDir + m->getRootName(m->getSimpleName(fastaFile)) + "redundant" + m->getExtension(fastaFile);
 154                 }
 155                 m->openOutputFile(outFastaFile, out);
 156
 157                 readNamesFile();
 158                 if (m->control_pressed) {  out.close(); outputTypes.clear(); remove(outFastaFile.c_str()); return 0; }
 159
 160                 ifstream in;
 161                 m->openInputFile(fastaFile, in);
 162
 163                 while (!in.eof()) {
 164
 165                         if (m->control_pressed) { in.close(); out.close(); outputTypes.clear(); remove(outFastaFile.c_str()); return 0; }
 166
 167                         Sequence seq(in); m->gobble(in);
 168
 169                         if (seq.getName() != "") {
 170
 171                                 //look for sequence name in nameMap
 172                                 map<string, string>::iterator it = nameMap.find(seq.getName());
 173
 174                                 if (it == nameMap.end()) {      m->mothurOut("[ERROR]: Your namefile does not contain " + seq.getName() + ", aborting."); m->mothurOutEndLine(); m->control_pressed = true; }
 175                                 else {
 176                                         vector<string> names;
 177                                         m->splitAtComma(it->second, names);
 178
 179                                         //output sequences
 180                                         for (int i = 0; i < names.size(); i++) {
 181                                                 out << ">" << names[i] << endl;
 182                                                 out << seq.getAligned() << endl;
 183                                         }
 184
 185                                         //remove seq from name map so we can check for seqs in namefile not in fastafile later
 186                                         nameMap.erase(it);
 187                                 }
 188                         }
 189                 }
 190                 in.close();
 191                 out.close();
 192
 193                 if (nameMap.size() != 0) { //then there are names in the namefile not in the fastafile
 194                         for (map<string, string>::iterator it = nameMap.begin(); it != nameMap.end(); it++) {
 195                                 m->mothurOut(it->first + " is not in your fasta file, but is in your name file. Please correct."); m->mothurOutEndLine();
 196                         }
 197                 }
 198
 199                 if (m->control_pressed) { outputTypes.clear(); remove(outFastaFile.c_str()); return 0; }
 200
 201                 m->mothurOutEndLine();
 202                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 203                 m->mothurOut(outFastaFile); m->mothurOutEndLine();
 204                 outputNames.push_back(outFastaFile);  outputTypes["fasta"].push_back(outFastaFile);
 205                 m->mothurOutEndLine();
 206
 207                 //set fasta file as new current fastafile
 208                 string current = "";
 209                 itTypes = outputTypes.find("fasta");
 210                 if (itTypes != outputTypes.end()) {
 211                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 212                 }
 213
 214                 return 0;
 215         }
 216         catch(exception& e) {
 217                 m->errorOut(e, "DeUniqueSeqsCommand", "execute");
 218                 exit(1);
 219         }
 220 }
 221 //**********************************************************************************************************************
 222 int DeUniqueSeqsCommand::readNamesFile() {
 223         try {
 224
 225                 ifstream inNames;
 226                 m->openInputFile(nameFile, inNames);
 227
 228                 string name, names;
 229                 map<string, string>::iterator it;
 230
 231                 while(inNames){
 232
 233                         if(m->control_pressed) { break; }
 234
 235                         inNames >> name;        m->gobble(inNames);
 236                         inNames >> names;
 237
 238                         it = nameMap.find(name);
 239
 240                         if (it == nameMap.end()) {      nameMap[name] = names; }
 241                         else { m->mothurOut("[ERROR]: Your namefile already contains " + name + ", aborting."); m->mothurOutEndLine(); m->control_pressed = true; }
 242
 243                         m->gobble(inNames);
 244                 }
 245                 inNames.close();
 246
 247                 return 0;
 248
 249         }
 250         catch(exception& e) {
 251                 m->errorOut(e, "DeUniqueSeqsCommand", "readNamesFile");
 252                 exit(1);
 253         }
 254 }
 255
 256 /**************************************************************************************/