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::getValidParameters(){
  15         try {
  16                 string Array[] =  {"fasta", "name","outputdir","inputdir"};
  17                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  18                 return myArray;
  19         }
  20         catch(exception& e) {
  21                 m->errorOut(e, "DeUniqueSeqsCommand", "getValidParameters");
  22                 exit(1);
  23         }
  24 }
  25 //**********************************************************************************************************************
  26 DeUniqueSeqsCommand::DeUniqueSeqsCommand(){
  27         try {
  28                 abort = true; calledHelp = true;
  29                 vector<string> tempOutNames;
  30                 outputTypes["fasta"] = tempOutNames;
  31         }
  32         catch(exception& e) {
  33                 m->errorOut(e, "DeUniqueSeqsCommand", "DeconvoluteCommand");
  34                 exit(1);
  35         }
  36 }
  37 //**********************************************************************************************************************
  38 vector<string> DeUniqueSeqsCommand::getRequiredParameters(){
  39         try {
  40                 string Array[] =  {"fasta","name"};
  41                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  42                 return myArray;
  43         }
  44         catch(exception& e) {
  45                 m->errorOut(e, "DeUniqueSeqsCommand", "getRequiredParameters");
  46                 exit(1);
  47         }
  48 }
  49 //**********************************************************************************************************************
  50 vector<string> DeUniqueSeqsCommand::getRequiredFiles(){
  51         try {
  52                 vector<string> myArray;
  53                 return myArray;
  54         }
  55         catch(exception& e) {
  56                 m->errorOut(e, "DeUniqueSeqsCommand", "getRequiredFiles");
  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                         //valid paramters for this command
  70                         string Array[] =  {"fasta", "name","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                         //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                         else {
  92                                 string path;
  93                                 it = parameters.find("fasta");
  94                                 //user has given a template file
  95                                 if(it != parameters.end()){
  96                                         path = m->hasPath(it->second);
  97                                         //if the user has not given a path then, add inputdir. else leave path alone.
  98                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
  99                                 }
 100
 101                                 it = parameters.find("name");
 102                                 //user has given a template file
 103                                 if(it != parameters.end()){
 104                                         path = m->hasPath(it->second);
 105                                         //if the user has not given a path then, add inputdir. else leave path alone.
 106                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 107                                 }
 108                         }
 109
 110
 111                         //check for required parameters
 112                         fastaFile = validParameter.validFile(parameters, "fasta", true);
 113                         if (fastaFile == "not open") { abort = true; }
 114                         else if (fastaFile == "not found") { fastaFile = ""; m->mothurOut("fasta is a required parameter for the deunique.seqs command."); m->mothurOutEndLine(); abort = true;  }
 115
 116                         //if the user changes the output directory command factory will send this info to us in the output parameter
 117                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 118                                 outputDir = "";
 119                                 outputDir += m->hasPath(fastaFile); //if user entered a file with a path then preserve it
 120                         }
 121
 122                         nameFile = validParameter.validFile(parameters, "name", true);
 123                         if (nameFile == "not open") { abort = true; }
 124                         else if (nameFile == "not found"){      nameFile = "";  m->mothurOut("name is a required parameter for the deunique.seqs command."); m->mothurOutEndLine(); abort = true;  }
 125                 }
 126
 127         }
 128         catch(exception& e) {
 129                 m->errorOut(e, "DeUniqueSeqsCommand", "DeUniqueSeqsCommand");
 130                 exit(1);
 131         }
 132 }
 133 //**********************************************************************************************************************
 134
 135 void DeUniqueSeqsCommand::help(){
 136         try {
 137                 m->mothurOut("The deunique.seqs command reads a fastafile and namefile, and creates a fastafile containing all the sequences.\n");
 138                 m->mothurOut("The deunique.seqs command parameters are fasta and name, both are required.\n");
 139                 m->mothurOut("The deunique.seqs command should be in the following format: \n");
 140                 m->mothurOut("deunique.seqs(fasta=yourFastaFile, name=yourNameFile) \n");
 141                 m->mothurOut("Example deunique.seqs(fasta=abrecovery.unique.fasta, name=abrecovery.names).\n");
 142                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
 143
 144         }
 145         catch(exception& e) {
 146                 m->errorOut(e, "DeUniqueSeqsCommand", "help");
 147                 exit(1);
 148         }
 149 }
 150
 151 /**************************************************************************************/
 152 int DeUniqueSeqsCommand::execute() {
 153         try {
 154
 155                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 156
 157                 //prepare filenames and open files
 158                 ofstream out;
 159                 string outFastaFile = m->getRootName(m->getSimpleName(fastaFile));
 160                 int pos = outFastaFile.find("unique");
 161                 if (pos != string::npos) {
 162                         outFastaFile = outputDir + outFastaFile.substr(0, pos) + "redundant" + m->getExtension(fastaFile);
 163                 }else{
 164                         outFastaFile = outputDir + m->getRootName(m->getSimpleName(fastaFile)) + "redundant" + m->getExtension(fastaFile);
 165                 }
 166                 m->openOutputFile(outFastaFile, out);
 167
 168                 readNamesFile();
 169                 if (m->control_pressed) {  out.close(); outputTypes.clear(); remove(outFastaFile.c_str()); return 0; }
 170
 171                 ifstream in;
 172                 m->openInputFile(fastaFile, in);
 173
 174                 while (!in.eof()) {
 175
 176                         if (m->control_pressed) { in.close(); out.close(); outputTypes.clear(); remove(outFastaFile.c_str()); return 0; }
 177
 178                         Sequence seq(in); m->gobble(in);
 179
 180                         if (seq.getName() != "") {
 181
 182                                 //look for sequence name in nameMap
 183                                 map<string, string>::iterator it = nameMap.find(seq.getName());
 184
 185                                 if (it == nameMap.end()) {      m->mothurOut("[ERROR]: Your namefile does not contain " + seq.getName() + ", aborting."); m->mothurOutEndLine(); m->control_pressed = true; }
 186                                 else {
 187                                         vector<string> names;
 188                                         m->splitAtComma(it->second, names);
 189
 190                                         //output sequences
 191                                         for (int i = 0; i < names.size(); i++) {
 192                                                 out << ">" << names[i] << endl;
 193                                                 out << seq.getAligned() << endl;
 194                                         }
 195
 196                                         //remove seq from name map so we can check for seqs in namefile not in fastafile later
 197                                         nameMap.erase(it);
 198                                 }
 199                         }
 200                 }
 201                 in.close();
 202                 out.close();
 203
 204                 if (nameMap.size() != 0) { //then there are names in the namefile not in the fastafile
 205                         for (map<string, string>::iterator it = nameMap.begin(); it != nameMap.end(); it++) {
 206                                 m->mothurOut(it->first + " is not in your fasta file, but is in your name file. Please correct."); m->mothurOutEndLine();
 207                         }
 208                 }
 209
 210                 if (m->control_pressed) { outputTypes.clear(); remove(outFastaFile.c_str()); return 0; }
 211
 212                 m->mothurOutEndLine();
 213                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 214                 m->mothurOut(outFastaFile); m->mothurOutEndLine();
 215                 outputNames.push_back(outFastaFile);  outputTypes["fasta"].push_back(outFastaFile);
 216                 m->mothurOutEndLine();
 217
 218                 //set fasta file as new current fastafile
 219                 string current = "";
 220                 itTypes = outputTypes.find("fasta");
 221                 if (itTypes != outputTypes.end()) {
 222                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 223                 }
 224
 225                 return 0;
 226         }
 227         catch(exception& e) {
 228                 m->errorOut(e, "DeUniqueSeqsCommand", "execute");
 229                 exit(1);
 230         }
 231 }
 232 //**********************************************************************************************************************
 233 int DeUniqueSeqsCommand::readNamesFile() {
 234         try {
 235
 236                 ifstream inNames;
 237                 m->openInputFile(nameFile, inNames);
 238
 239                 string name, names;
 240                 map<string, string>::iterator it;
 241
 242                 while(inNames){
 243
 244                         if(m->control_pressed) { break; }
 245
 246                         inNames >> name;        m->gobble(inNames);
 247                         inNames >> names;
 248
 249                         it = nameMap.find(name);
 250
 251                         if (it == nameMap.end()) {      nameMap[name] = names; }
 252                         else { m->mothurOut("[ERROR]: Your namefile already contains " + name + ", aborting."); m->mothurOutEndLine(); m->control_pressed = true; }
 253
 254                         m->gobble(inNames);
 255                 }
 256                 inNames.close();
 257
 258                 return 0;
 259
 260         }
 261         catch(exception& e) {
 262                 m->errorOut(e, "DeUniqueSeqsCommand", "readNamesFile");
 263                 exit(1);
 264         }
 265 }
 266
 267 /**************************************************************************************/