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;
  29                 //initialize outputTypes
  30                 vector<string> tempOutNames;
  31                 outputTypes["fasta"] = tempOutNames;
  32         }
  33         catch(exception& e) {
  34                 m->errorOut(e, "DeUniqueSeqsCommand", "DeconvoluteCommand");
  35                 exit(1);
  36         }
  37 }
  38 //**********************************************************************************************************************
  39 vector<string> DeUniqueSeqsCommand::getRequiredParameters(){
  40         try {
  41                 string Array[] =  {"fasta","name"};
  42                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  43                 return myArray;
  44         }
  45         catch(exception& e) {
  46                 m->errorOut(e, "DeUniqueSeqsCommand", "getRequiredParameters");
  47                 exit(1);
  48         }
  49 }
  50 //**********************************************************************************************************************
  51 vector<string> DeUniqueSeqsCommand::getRequiredFiles(){
  52         try {
  53                 vector<string> myArray;
  54                 return myArray;
  55         }
  56         catch(exception& e) {
  57                 m->errorOut(e, "DeUniqueSeqsCommand", "getRequiredFiles");
  58                 exit(1);
  59         }
  60 }
  61 /**************************************************************************************/
  62 DeUniqueSeqsCommand::DeUniqueSeqsCommand(string option)  {
  63         try {
  64                 abort = false;
  65
  66                 //allow user to run help
  67                 if(option == "help") { help(); abort = true; }
  68
  69                 else {
  70                         //valid paramters for this command
  71                         string Array[] =  {"fasta", "name","outputdir","inputdir"};
  72                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  73
  74                         OptionParser parser(option);
  75                         map<string,string> parameters = parser.getParameters();
  76
  77                         ValidParameters validParameter;
  78                         map<string, string>::iterator it;
  79
  80                         //check to make sure all parameters are valid for command
  81                         for (it = parameters.begin(); it != parameters.end(); it++) {
  82                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  83                         }
  84
  85                         //initialize outputTypes
  86                         vector<string> tempOutNames;
  87                         outputTypes["fasta"] = tempOutNames;
  88
  89                         //if the user changes the input directory command factory will send this info to us in the output parameter
  90                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  91                         if (inputDir == "not found"){   inputDir = "";          }
  92                         else {
  93                                 string path;
  94                                 it = parameters.find("fasta");
  95                                 //user has given a template file
  96                                 if(it != parameters.end()){
  97                                         path = m->hasPath(it->second);
  98                                         //if the user has not given a path then, add inputdir. else leave path alone.
  99                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 100                                 }
 101
 102                                 it = parameters.find("name");
 103                                 //user has given a template file
 104                                 if(it != parameters.end()){
 105                                         path = m->hasPath(it->second);
 106                                         //if the user has not given a path then, add inputdir. else leave path alone.
 107                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 108                                 }
 109                         }
 110
 111
 112                         //check for required parameters
 113                         fastaFile = validParameter.validFile(parameters, "fasta", true);
 114                         if (fastaFile == "not open") { abort = true; }
 115                         else if (fastaFile == "not found") { fastaFile = ""; m->mothurOut("fasta is a required parameter for the deunique.seqs command."); m->mothurOutEndLine(); abort = true;  }
 116
 117                         //if the user changes the output directory command factory will send this info to us in the output parameter
 118                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 119                                 outputDir = "";
 120                                 outputDir += m->hasPath(fastaFile); //if user entered a file with a path then preserve it
 121                         }
 122
 123                         nameFile = validParameter.validFile(parameters, "name", true);
 124                         if (nameFile == "not open") { abort = true; }
 125                         else if (nameFile == "not found"){      nameFile = "";  m->mothurOut("name is a required parameter for the deunique.seqs command."); m->mothurOutEndLine(); abort = true;  }
 126                 }
 127
 128         }
 129         catch(exception& e) {
 130                 m->errorOut(e, "DeUniqueSeqsCommand", "DeUniqueSeqsCommand");
 131                 exit(1);
 132         }
 133 }
 134 //**********************************************************************************************************************
 135
 136 void DeUniqueSeqsCommand::help(){
 137         try {
 138                 m->mothurOut("The deunique.seqs command reads a fastafile and namefile, and creates a fastafile containing all the sequences.\n");
 139                 m->mothurOut("The deunique.seqs command parameters are fasta and name, both are required.\n");
 140                 m->mothurOut("The deunique.seqs command should be in the following format: \n");
 141                 m->mothurOut("deunique.seqs(fasta=yourFastaFile, name=yourNameFile) \n");
 142                 m->mothurOut("Example deunique.seqs(fasta=abrecovery.unique.fasta, name=abrecovery.names).\n");
 143                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
 144
 145         }
 146         catch(exception& e) {
 147                 m->errorOut(e, "DeUniqueSeqsCommand", "help");
 148                 exit(1);
 149         }
 150 }
 151
 152 /**************************************************************************************/
 153 int DeUniqueSeqsCommand::execute() {
 154         try {
 155
 156                 if (abort == true) { return 0; }
 157
 158                 //prepare filenames and open files
 159                 ofstream out;
 160                 string outFastaFile = m->getRootName(m->getSimpleName(fastaFile));
 161                 int pos = outFastaFile.find("unique");
 162                 if (pos != string::npos) {
 163                         outFastaFile = outputDir + outFastaFile.substr(0, pos) + "redundant" + m->getExtension(fastaFile);
 164                 }else{
 165                         outFastaFile = outputDir + m->getRootName(m->getSimpleName(fastaFile)) + "redundant" + m->getExtension(fastaFile);
 166                 }
 167                 m->openOutputFile(outFastaFile, out);
 168
 169                 readNamesFile();
 170                 if (m->control_pressed) {  out.close(); outputTypes.clear(); remove(outFastaFile.c_str()); return 0; }
 171
 172                 ifstream in;
 173                 m->openInputFile(fastaFile, in);
 174
 175                 while (!in.eof()) {
 176
 177                         if (m->control_pressed) { in.close(); out.close(); outputTypes.clear(); remove(outFastaFile.c_str()); return 0; }
 178
 179                         Sequence seq(in); m->gobble(in);
 180
 181                         if (seq.getName() != "") {
 182
 183                                 //look for sequence name in nameMap
 184                                 map<string, string>::iterator it = nameMap.find(seq.getName());
 185
 186                                 if (it == nameMap.end()) {      m->mothurOut("[ERROR]: Your namefile does not contain " + seq.getName() + ", aborting."); m->mothurOutEndLine(); m->control_pressed = true; }
 187                                 else {
 188                                         vector<string> names;
 189                                         m->splitAtComma(it->second, names);
 190
 191                                         //output sequences
 192                                         for (int i = 0; i < names.size(); i++) {
 193                                                 out << ">" << names[i] << endl;
 194                                                 out << seq.getAligned() << endl;
 195                                         }
 196
 197                                         //remove seq from name map so we can check for seqs in namefile not in fastafile later
 198                                         nameMap.erase(it);
 199                                 }
 200                         }
 201                 }
 202                 in.close();
 203                 out.close();
 204
 205                 if (nameMap.size() != 0) { //then there are names in the namefile not in the fastafile
 206                         for (map<string, string>::iterator it = nameMap.begin(); it != nameMap.end(); it++) {
 207                                 m->mothurOut(it->first + " is not in your fasta file, but is in your name file. Please correct."); m->mothurOutEndLine();
 208                         }
 209                 }
 210
 211                 if (m->control_pressed) { outputTypes.clear(); remove(outFastaFile.c_str()); return 0; }
 212
 213                 m->mothurOutEndLine();
 214                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 215                 m->mothurOut(outFastaFile); m->mothurOutEndLine();
 216                 outputNames.push_back(outFastaFile);  outputTypes["fasta"].push_back(outFastaFile);
 217                 m->mothurOutEndLine();
 218
 219                 return 0;
 220         }
 221         catch(exception& e) {
 222                 m->errorOut(e, "DeUniqueSeqsCommand", "execute");
 223                 exit(1);
 224         }
 225 }
 226 //**********************************************************************************************************************
 227 int DeUniqueSeqsCommand::readNamesFile() {
 228         try {
 229
 230                 ifstream inNames;
 231                 m->openInputFile(nameFile, inNames);
 232
 233                 string name, names;
 234                 map<string, string>::iterator it;
 235
 236                 while(inNames){
 237
 238                         if(m->control_pressed) { break; }
 239
 240                         inNames >> name;        m->gobble(inNames);
 241                         inNames >> names;
 242
 243                         it = nameMap.find(name);
 244
 245                         if (it == nameMap.end()) {      nameMap[name] = names; }
 246                         else { m->mothurOut("[ERROR]: Your namefile already contains " + name + ", aborting."); m->mothurOutEndLine(); m->control_pressed = true; }
 247
 248                         m->gobble(inNames);
 249                 }
 250                 inNames.close();
 251
 252                 return 0;
 253
 254         }
 255         catch(exception& e) {
 256                 m->errorOut(e, "DeUniqueSeqsCommand", "readNamesFile");
 257                 exit(1);
 258         }
 259 }
 260
 261 /**************************************************************************************/