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