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 string DeUniqueSeqsCommand::getOutputFileNameTag(string type, string inputName=""){
  49         try {
  50         string outputFileName = "";
  51                 map<string, vector<string> >::iterator it;
  52
  53         //is this a type this command creates
  54         it = outputTypes.find(type);
  55         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  56         else {
  57             if (type == "fasta") {  outputFileName =  "redundant.fasta"; }
  58             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  59         }
  60         return outputFileName;
  61         }
  62         catch(exception& e) {
  63                 m->errorOut(e, "DeUniqueSeqsCommand", "getOutputFileNameTag");
  64                 exit(1);
  65         }
  66 }
  67 //**********************************************************************************************************************
  68 DeUniqueSeqsCommand::DeUniqueSeqsCommand(){
  69         try {
  70                 abort = true; calledHelp = true;
  71                 setParameters();
  72                 vector<string> tempOutNames;
  73                 outputTypes["fasta"] = tempOutNames;
  74         }
  75         catch(exception& e) {
  76                 m->errorOut(e, "DeUniqueSeqsCommand", "DeconvoluteCommand");
  77                 exit(1);
  78         }
  79 }
  80 /**************************************************************************************/
  81 DeUniqueSeqsCommand::DeUniqueSeqsCommand(string option)  {
  82         try {
  83                 abort = false; calledHelp = false;
  84
  85                 //allow user to run help
  86                 if(option == "help") { help(); abort = true; calledHelp = true; }
  87                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  88
  89                 else {
  90                         vector<string> myArray = setParameters();
  91
  92                         OptionParser parser(option);
  93                         map<string,string> parameters = parser.getParameters();
  94
  95                         ValidParameters validParameter;
  96                         map<string, string>::iterator it;
  97
  98                         //check to make sure all parameters are valid for command
  99                         for (it = parameters.begin(); it != parameters.end(); it++) {
 100                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 101                         }
 102
 103                         //initialize outputTypes
 104                         vector<string> tempOutNames;
 105                         outputTypes["fasta"] = tempOutNames;
 106
 107                         //if the user changes the input directory command factory will send this info to us in the output parameter
 108                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 109                         if (inputDir == "not found"){   inputDir = "";          }
 110                         else {
 111                                 string path;
 112                                 it = parameters.find("fasta");
 113                                 //user has given a template file
 114                                 if(it != parameters.end()){
 115                                         path = m->hasPath(it->second);
 116                                         //if the user has not given a path then, add inputdir. else leave path alone.
 117                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 118                                 }
 119
 120                                 it = parameters.find("name");
 121                                 //user has given a template file
 122                                 if(it != parameters.end()){
 123                                         path = m->hasPath(it->second);
 124                                         //if the user has not given a path then, add inputdir. else leave path alone.
 125                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 126                                 }
 127                         }
 128
 129
 130                         //check for required parameters
 131                         fastaFile = validParameter.validFile(parameters, "fasta", true);
 132                         if (fastaFile == "not open") { abort = true; }
 133                         else if (fastaFile == "not found") {
 134                                 fastaFile = m->getFastaFile();
 135                                 if (fastaFile != "") { m->mothurOut("Using " + fastaFile + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 136                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 137                         }else { m->setFastaFile(fastaFile); }
 138
 139                         //if the user changes the output directory command factory will send this info to us in the output parameter
 140                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 141                                 outputDir = "";
 142                                 outputDir += m->hasPath(fastaFile); //if user entered a file with a path then preserve it
 143                         }
 144
 145                         nameFile = validParameter.validFile(parameters, "name", true);
 146                         if (nameFile == "not open") { abort = true; }
 147                         else if (nameFile == "not found"){
 148                                 nameFile = m->getNameFile();
 149                                 if (nameFile != "") { m->mothurOut("Using " + nameFile + " as input file for the name parameter."); m->mothurOutEndLine(); }
 150                                 else {  m->mothurOut("You have no current namefile and the name parameter is required."); m->mothurOutEndLine(); abort = true; }
 151                         }else { m->setNameFile(nameFile); }
 152                 }
 153
 154         }
 155         catch(exception& e) {
 156                 m->errorOut(e, "DeUniqueSeqsCommand", "DeUniqueSeqsCommand");
 157                 exit(1);
 158         }
 159 }
 160 /**************************************************************************************/
 161 int DeUniqueSeqsCommand::execute() {
 162         try {
 163
 164                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 165
 166                 //prepare filenames and open files
 167                 ofstream out;
 168                 string outFastaFile = m->getRootName(m->getSimpleName(fastaFile));
 169                 int pos = outFastaFile.find("unique");
 170                 if (pos != string::npos) {
 171                         outFastaFile = outputDir + outFastaFile.substr(0, pos) + getOutputFileNameTag("fasta");
 172                 }else{
 173                         outFastaFile = outputDir + m->getRootName(m->getSimpleName(fastaFile)) + getOutputFileNameTag("fasta");
 174                 }
 175                 m->openOutputFile(outFastaFile, out);
 176
 177                 readNamesFile();
 178                 if (m->control_pressed) {  out.close(); outputTypes.clear(); m->mothurRemove(outFastaFile); return 0; }
 179
 180                 ifstream in;
 181                 m->openInputFile(fastaFile, in);
 182
 183                 while (!in.eof()) {
 184
 185                         if (m->control_pressed) { in.close(); out.close(); outputTypes.clear(); m->mothurRemove(outFastaFile); return 0; }
 186
 187                         Sequence seq(in); m->gobble(in);
 188
 189                         if (seq.getName() != "") {
 190
 191                                 //look for sequence name in nameMap
 192                                 map<string, string>::iterator it = nameMap.find(seq.getName());
 193
 194                                 if (it == nameMap.end()) {      m->mothurOut("[ERROR]: Your namefile does not contain " + seq.getName() + ", aborting."); m->mothurOutEndLine(); m->control_pressed = true; }
 195                                 else {
 196                                         vector<string> names;
 197                                         m->splitAtComma(it->second, names);
 198
 199                                         //output sequences
 200                                         for (int i = 0; i < names.size(); i++) {
 201                                                 out << ">" << names[i] << endl;
 202                                                 out << seq.getAligned() << endl;
 203                                         }
 204
 205                                         //remove seq from name map so we can check for seqs in namefile not in fastafile later
 206                                         nameMap.erase(it);
 207                                 }
 208                         }
 209                 }
 210                 in.close();
 211                 out.close();
 212
 213                 if (nameMap.size() != 0) { //then there are names in the namefile not in the fastafile
 214                         for (map<string, string>::iterator it = nameMap.begin(); it != nameMap.end(); it++) {
 215                                 m->mothurOut(it->first + " is not in your fasta file, but is in your name file. Please correct."); m->mothurOutEndLine();
 216                         }
 217                 }
 218
 219                 if (m->control_pressed) { outputTypes.clear(); m->mothurRemove(outFastaFile); return 0; }
 220
 221                 m->mothurOutEndLine();
 222                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 223                 m->mothurOut(outFastaFile); m->mothurOutEndLine();
 224                 outputNames.push_back(outFastaFile);  outputTypes["fasta"].push_back(outFastaFile);
 225                 m->mothurOutEndLine();
 226
 227                 //set fasta file as new current fastafile
 228                 string current = "";
 229                 itTypes = outputTypes.find("fasta");
 230                 if (itTypes != outputTypes.end()) {
 231                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 232                 }
 233
 234                 return 0;
 235         }
 236         catch(exception& e) {
 237                 m->errorOut(e, "DeUniqueSeqsCommand", "execute");
 238                 exit(1);
 239         }
 240 }
 241 //**********************************************************************************************************************
 242 int DeUniqueSeqsCommand::readNamesFile() {
 243         try {
 244
 245                 ifstream inNames;
 246                 m->openInputFile(nameFile, inNames);
 247
 248                 string name, names;
 249                 map<string, string>::iterator it;
 250
 251                 while(inNames){
 252
 253                         if(m->control_pressed) { break; }
 254
 255                         inNames >> name;        m->gobble(inNames);
 256                         inNames >> names;
 257
 258                         it = nameMap.find(name);
 259
 260                         if (it == nameMap.end()) {      nameMap[name] = names; }
 261                         else { m->mothurOut("[ERROR]: Your namefile already contains " + name + ", aborting."); m->mothurOutEndLine(); m->control_pressed = true; }
 262
 263                         m->gobble(inNames);
 264                 }
 265                 inNames.close();
 266
 267                 return 0;
 268
 269         }
 270         catch(exception& e) {
 271                 m->errorOut(e, "DeUniqueSeqsCommand", "readNamesFile");
 272                 exit(1);
 273         }
 274 }
 275
 276 /**************************************************************************************/