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\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
  68                 else {
  69                         vector<string> myArray = setParameters();
  70
  71                         OptionParser parser(option);
  72                         map<string,string> parameters = parser.getParameters();
  73
  74                         ValidParameters validParameter;
  75                         map<string, string>::iterator it;
  76
  77                         //check to make sure all parameters are valid for command
  78                         for (it = parameters.begin(); it != parameters.end(); it++) {
  79                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  80                         }
  81
  82                         //initialize outputTypes
  83                         vector<string> tempOutNames;
  84                         outputTypes["fasta"] = tempOutNames;
  85
  86                         //if the user changes the input directory command factory will send this info to us in the output parameter
  87                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  88                         if (inputDir == "not found"){   inputDir = "";          }
  89                         else {
  90                                 string path;
  91                                 it = parameters.find("fasta");
  92                                 //user has given a template file
  93                                 if(it != parameters.end()){
  94                                         path = m->hasPath(it->second);
  95                                         //if the user has not given a path then, add inputdir. else leave path alone.
  96                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
  97                                 }
  98
  99                                 it = parameters.find("name");
 100                                 //user has given a template file
 101                                 if(it != parameters.end()){
 102                                         path = m->hasPath(it->second);
 103                                         //if the user has not given a path then, add inputdir. else leave path alone.
 104                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 105                                 }
 106                         }
 107
 108
 109                         //check for required parameters
 110                         fastaFile = validParameter.validFile(parameters, "fasta", true);
 111                         if (fastaFile == "not open") { abort = true; }
 112                         else if (fastaFile == "not found") {
 113                                 fastaFile = m->getFastaFile();
 114                                 if (fastaFile != "") { m->mothurOut("Using " + fastaFile + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 115                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 116                         }
 117
 118                         //if the user changes the output directory command factory will send this info to us in the output parameter
 119                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 120                                 outputDir = "";
 121                                 outputDir += m->hasPath(fastaFile); //if user entered a file with a path then preserve it
 122                         }
 123
 124                         nameFile = validParameter.validFile(parameters, "name", true);
 125                         if (nameFile == "not open") { abort = true; }
 126                         else if (nameFile == "not found"){
 127                                 nameFile = m->getNameFile();
 128                                 if (nameFile != "") { m->mothurOut("Using " + nameFile + " as input file for the name parameter."); m->mothurOutEndLine(); }
 129                                 else {  m->mothurOut("You have no current namefile and the name parameter is required."); m->mothurOutEndLine(); abort = true; }
 130                         }
 131                 }
 132
 133         }
 134         catch(exception& e) {
 135                 m->errorOut(e, "DeUniqueSeqsCommand", "DeUniqueSeqsCommand");
 136                 exit(1);
 137         }
 138 }
 139 /**************************************************************************************/
 140 int DeUniqueSeqsCommand::execute() {
 141         try {
 142
 143                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 144
 145                 //prepare filenames and open files
 146                 ofstream out;
 147                 string outFastaFile = m->getRootName(m->getSimpleName(fastaFile));
 148                 int pos = outFastaFile.find("unique");
 149                 if (pos != string::npos) {
 150                         outFastaFile = outputDir + outFastaFile.substr(0, pos) + "redundant" + m->getExtension(fastaFile);
 151                 }else{
 152                         outFastaFile = outputDir + m->getRootName(m->getSimpleName(fastaFile)) + "redundant" + m->getExtension(fastaFile);
 153                 }
 154                 m->openOutputFile(outFastaFile, out);
 155
 156                 readNamesFile();
 157                 if (m->control_pressed) {  out.close(); outputTypes.clear(); remove(outFastaFile.c_str()); return 0; }
 158
 159                 ifstream in;
 160                 m->openInputFile(fastaFile, in);
 161
 162                 while (!in.eof()) {
 163
 164                         if (m->control_pressed) { in.close(); out.close(); outputTypes.clear(); remove(outFastaFile.c_str()); return 0; }
 165
 166                         Sequence seq(in); m->gobble(in);
 167
 168                         if (seq.getName() != "") {
 169
 170                                 //look for sequence name in nameMap
 171                                 map<string, string>::iterator it = nameMap.find(seq.getName());
 172
 173                                 if (it == nameMap.end()) {      m->mothurOut("[ERROR]: Your namefile does not contain " + seq.getName() + ", aborting."); m->mothurOutEndLine(); m->control_pressed = true; }
 174                                 else {
 175                                         vector<string> names;
 176                                         m->splitAtComma(it->second, names);
 177
 178                                         //output sequences
 179                                         for (int i = 0; i < names.size(); i++) {
 180                                                 out << ">" << names[i] << endl;
 181                                                 out << seq.getAligned() << endl;
 182                                         }
 183
 184                                         //remove seq from name map so we can check for seqs in namefile not in fastafile later
 185                                         nameMap.erase(it);
 186                                 }
 187                         }
 188                 }
 189                 in.close();
 190                 out.close();
 191
 192                 if (nameMap.size() != 0) { //then there are names in the namefile not in the fastafile
 193                         for (map<string, string>::iterator it = nameMap.begin(); it != nameMap.end(); it++) {
 194                                 m->mothurOut(it->first + " is not in your fasta file, but is in your name file. Please correct."); m->mothurOutEndLine();
 195                         }
 196                 }
 197
 198                 if (m->control_pressed) { outputTypes.clear(); remove(outFastaFile.c_str()); return 0; }
 199
 200                 m->mothurOutEndLine();
 201                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 202                 m->mothurOut(outFastaFile); m->mothurOutEndLine();
 203                 outputNames.push_back(outFastaFile);  outputTypes["fasta"].push_back(outFastaFile);
 204                 m->mothurOutEndLine();
 205
 206                 //set fasta file as new current fastafile
 207                 string current = "";
 208                 itTypes = outputTypes.find("fasta");
 209                 if (itTypes != outputTypes.end()) {
 210                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 211                 }
 212
 213                 return 0;
 214         }
 215         catch(exception& e) {
 216                 m->errorOut(e, "DeUniqueSeqsCommand", "execute");
 217                 exit(1);
 218         }
 219 }
 220 //**********************************************************************************************************************
 221 int DeUniqueSeqsCommand::readNamesFile() {
 222         try {
 223
 224                 ifstream inNames;
 225                 m->openInputFile(nameFile, inNames);
 226
 227                 string name, names;
 228                 map<string, string>::iterator it;
 229
 230                 while(inNames){
 231
 232                         if(m->control_pressed) { break; }
 233
 234                         inNames >> name;        m->gobble(inNames);
 235                         inNames >> names;
 236
 237                         it = nameMap.find(name);
 238
 239                         if (it == nameMap.end()) {      nameMap[name] = names; }
 240                         else { m->mothurOut("[ERROR]: Your namefile already contains " + name + ", aborting."); m->mothurOutEndLine(); m->control_pressed = true; }
 241
 242                         m->gobble(inNames);
 243                 }
 244                 inNames.close();
 245
 246                 return 0;
 247
 248         }
 249         catch(exception& e) {
 250                 m->errorOut(e, "DeUniqueSeqsCommand", "readNamesFile");
 251                 exit(1);
 252         }
 253 }
 254
 255 /**************************************************************************************/