makegroupcommand.cpp

   1 /*
   2  *  makegroupcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 5/7/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "makegroupcommand.h"
  11 #include "sequence.hpp"
  12
  13
  14 //**********************************************************************************************************************
  15 vector<string> MakeGroupCommand::setParameters(){
  16         try {
  17                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none","group",false,true,true); parameters.push_back(pfasta);
  18                 CommandParameter pgroups("groups", "String", "", "", "", "", "","",false,false,true); parameters.push_back(pgroups);
  19                 CommandParameter poutput("output", "String", "", "", "", "", "","",false,false); parameters.push_back(poutput);
  20                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
  21                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
  22
  23                 vector<string> myArray;
  24                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  25                 return myArray;
  26         }
  27         catch(exception& e) {
  28                 m->errorOut(e, "MakeGroupCommand", "setParameters");
  29                 exit(1);
  30         }
  31 }
  32 //**********************************************************************************************************************
  33 string MakeGroupCommand::getHelpString(){
  34         try {
  35                 string helpString = "";
  36                 helpString += "The make.group command reads a fasta file or series of fasta files and creates a groupfile.\n";
  37                 helpString += "The make.group command parameters are fasta, groups and output. Fasta and group are required.\n";
  38                 helpString += "The output parameter allows you to specify the name of groupfile created. \n";
  39                 helpString += "The make.group command should be in the following format: \n";
  40                 helpString += "make.group(fasta=yourFastaFiles, groups=yourGroups). \n";
  41                 helpString += "Example make.group(fasta=seqs1.fasta-seq2.fasta-seqs3.fasta, groups=A-B-C)\n";
  42                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFiles).\n";
  43                 return helpString;
  44         }
  45         catch(exception& e) {
  46                 m->errorOut(e, "MakeGroupCommand", "getHelpString");
  47                 exit(1);
  48         }
  49 }
  50 //**********************************************************************************************************************
  51 string MakeGroupCommand::getOutputPattern(string type) {
  52     try {
  53         string pattern = "";
  54
  55         if (type == "group") {  pattern = "[filename],groups"; }
  56         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  57
  58         return pattern;
  59     }
  60     catch(exception& e) {
  61         m->errorOut(e, "MakeGroupCommand", "getOutputPattern");
  62         exit(1);
  63     }
  64 }
  65 //**********************************************************************************************************************
  66 MakeGroupCommand::MakeGroupCommand(){
  67         try {
  68                 abort = true; calledHelp = true;
  69                 setParameters();
  70                 vector<string> tempOutNames;
  71                 outputTypes["group"] = tempOutNames;
  72         }
  73         catch(exception& e) {
  74                 m->errorOut(e, "MakeGroupCommand", "MakeGroupCommand");
  75                 exit(1);
  76         }
  77 }
  78
  79 //**********************************************************************************************************************
  80
  81 MakeGroupCommand::MakeGroupCommand(string option)  {
  82         try {
  83
  84                 abort = false; calledHelp = false;
  85
  86                 //allow user to run help
  87                 if(option == "help") { help(); abort = true; calledHelp = true; }
  88                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  89
  90                 else {
  91                         vector<string> myArray = setParameters();
  92
  93                         OptionParser parser(option);
  94                         map<string, string> parameters = parser.getParameters();
  95
  96                         ValidParameters validParameter;
  97                         map<string, string>::iterator it;
  98
  99                         //check to make sure all parameters are valid for command
 100                         for (it = parameters.begin(); it != parameters.end(); it++) {
 101                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 102                         }
 103
 104                         //initialize outputTypes
 105                         vector<string> tempOutNames;
 106                         outputTypes["group"] = tempOutNames;
 107
 108                         //if the user changes the input directory command factory will send this info to us in the output parameter
 109                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 110                         if (inputDir == "not found"){   inputDir = "";          }
 111
 112                         fastaFileName = validParameter.validFile(parameters, "fasta", false);
 113                         if (fastaFileName == "not found") {                             //if there is a current fasta file, use it
 114                                 string filename = m->getFastaFile();
 115                                 if (filename != "") { fastaFileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 116                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 117                         }else {
 118                                 m->splitAtDash(fastaFileName, fastaFileNames);
 119
 120                                 //go through files and make sure they are good, if not, then disregard them
 121                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 122
 123                                         bool ignore = false;
 124                                         if (fastaFileNames[i] == "current") {
 125                                                 fastaFileNames[i] = m->getFastaFile();
 126                                                 if (fastaFileNames[i] != "") {
 127                                                         m->mothurOut("Using " + fastaFileNames[i] + " as input file for the fasta parameter where you had given current."); m->mothurOutEndLine();
 128                                                         filename += m->getRootName(m->getSimpleName(fastaFileNames[i]));
 129                                                 }
 130                                                 else {
 131                                                         m->mothurOut("You have no current fastafile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 132                                                         //erase from file list
 133                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 134                                                         i--;
 135                                                 }
 136                                         }
 137
 138                                         if (!ignore) {
 139                                                 if (inputDir != "") {
 140                                                         string path = m->hasPath(fastaFileNames[i]);
 141                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 142                                                         if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 143                                                 }
 144
 145                                                 ifstream in;
 146                                                 int ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 147
 148                                                 //if you can't open it, try default location
 149                                                 if (ableToOpen == 1) {
 150                                                         if (m->getDefaultPath() != "") { //default path is set
 151                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 152                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 153                                                                 ifstream in2;
 154                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 155                                                                 in2.close();
 156                                                                 fastaFileNames[i] = tryPath;
 157                                                         }
 158                                                 }
 159
 160                                                 //if you can't open it, try default location
 161                                                 if (ableToOpen == 1) {
 162                                                         if (m->getOutputDir() != "") { //default path is set
 163                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 164                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 165                                                                 ifstream in2;
 166                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 167                                                                 in2.close();
 168                                                                 fastaFileNames[i] = tryPath;
 169                                                         }
 170                                                 }
 171                                                 in.close();
 172
 173                                                 if (ableToOpen == 1) {
 174                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 175                                                         //erase from file list
 176                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 177                                                         i--;
 178                                                 }else{  filename += m->getRootName(m->getSimpleName(fastaFileNames[i]));  m->setFastaFile(fastaFileNames[i]); }
 179                                         }
 180                                 }
 181
 182                                 //prevent giantic file name
 183                 map<string, string> variables;
 184                 variables["[filename]"] = filename;
 185                                 if (fastaFileNames.size() > 3) { variables["[filename]"] = outputDir + "merge"; }
 186                                 filename = getOutputFileName("group",variables);
 187
 188                                 //make sure there is at least one valid file left
 189                                 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 190                         }
 191
 192                         //if the user changes the output directory command factory will send this info to us in the output parameter
 193                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 194
 195                         output = validParameter.validFile(parameters, "output", false);
 196                         if (output == "not found") { output = "";  }
 197                         else{ filename = output; }
 198
 199                         groups = validParameter.validFile(parameters, "groups", false);
 200                         if (groups == "not found") { m->mothurOut("groups is a required parameter for the make.group command."); m->mothurOutEndLine(); abort = true;  }
 201                         else { m->splitAtDash(groups, groupsNames);     }
 202
 203                         if (groupsNames.size() != fastaFileNames.size()) { m->mothurOut("You do not have the same number of valid fastfile files as groups.  This could be because we could not open a fastafile."); m->mothurOutEndLine(); abort = true;  }
 204                 }
 205         }
 206         catch(exception& e) {
 207                 m->errorOut(e, "MakeGroupCommand", "MakeGroupCommand");
 208                 exit(1);
 209         }
 210 }
 211 //**********************************************************************************************************************
 212
 213 int MakeGroupCommand::execute(){
 214         try {
 215                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 216
 217                 if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[0]); }
 218
 219                 filename = outputDir + filename;
 220
 221                 ofstream out;
 222                 m->openOutputFile(filename, out);
 223
 224                 for (int i = 0; i < fastaFileNames.size(); i++) {
 225
 226                         if (m->control_pressed) { outputTypes.clear(); out.close(); m->mothurRemove(filename); return 0; }
 227
 228                         ifstream in;
 229                         m->openInputFile(fastaFileNames[i], in);
 230
 231                         while (!in.eof()) {
 232
 233                                 Sequence seq(in, "no align"); m->gobble(in);
 234
 235                                 if (m->control_pressed) { outputTypes.clear();  in.close(); out.close(); m->mothurRemove(filename); return 0; }
 236
 237                                 if (seq.getName() != "") {      out << seq.getName() << '\t' << groupsNames[i] << endl;         }
 238                         }
 239                         in.close();
 240                 }
 241
 242                 out.close();
 243
 244                 m->mothurOutEndLine();
 245                 m->mothurOut("Output File Names: " + filename); m->mothurOutEndLine(); outputNames.push_back(filename); outputTypes["group"].push_back(filename);
 246                 m->mothurOutEndLine();
 247
 248                 //set group file as new current groupfile
 249                 string current = "";
 250                 itTypes = outputTypes.find("group");
 251                 if (itTypes != outputTypes.end()) {
 252                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 253                 }
 254
 255                 return 0;
 256         }
 257         catch(exception& e) {
 258                 m->errorOut(e, "MakeGroupCommand", "execute");
 259                 exit(1);
 260         }
 261 }
 262 //**********************************************************************************************************************
 263
 264