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