deconvolutecommand.cpp

   1 /*
   2  *  deconvolute.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 1/21/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "deconvolutecommand.h"
  11 #include "sequence.hpp"
  12
  13 //**********************************************************************************************************************
  14 vector<string> DeconvoluteCommand::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,false); 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, "DeconvoluteCommand", "setParameters");
  27                 exit(1);
  28         }
  29 }
  30 //**********************************************************************************************************************
  31 string DeconvoluteCommand::getHelpString(){
  32         try {
  33                 string helpString = "";
  34                 helpString += "The unique.seqs command reads a fastafile and creates a namesfile.\n";
  35                 helpString += "It creates a file where the first column is the groupname and the second column is a list of sequence names who have the same sequence. \n";
  36                 helpString += "If the sequence is unique the second column will just contain its name. \n";
  37                 helpString += "The unique.seqs command parameters are fasta and name.  fasta is required, unless there is a valid current fasta file.\n";
  38                 helpString += "The unique.seqs command should be in the following format: \n";
  39                 helpString += "unique.seqs(fasta=yourFastaFile) \n";
  40                 return helpString;
  41         }
  42         catch(exception& e) {
  43                 m->errorOut(e, "DeconvoluteCommand", "getHelpString");
  44                 exit(1);
  45         }
  46 }
  47 //**********************************************************************************************************************
  48 DeconvoluteCommand::DeconvoluteCommand(){
  49         try {
  50                 abort = true; calledHelp = true;
  51                 setParameters();
  52                 vector<string> tempOutNames;
  53                 outputTypes["fasta"] = tempOutNames;
  54                 outputTypes["name"] = tempOutNames;
  55         }
  56         catch(exception& e) {
  57                 m->errorOut(e, "DeconvoluteCommand", "DeconvoluteCommand");
  58                 exit(1);
  59         }
  60 }
  61 /**************************************************************************************/
  62 DeconvoluteCommand::DeconvoluteCommand(string option)  {
  63         try {
  64                 abort = false; calledHelp = false;
  65
  66                 //allow user to run help
  67                 if(option == "help") { help(); abort = true; calledHelp = true; }
  68                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  69
  70                 else {
  71                         vector<string> myArray = setParameters();
  72
  73                         OptionParser parser(option);
  74                         map<string,string> parameters = parser.getParameters();
  75
  76                         ValidParameters validParameter;
  77                         map<string, string>::iterator it;
  78
  79                         //check to make sure all parameters are valid for command
  80                         for (it = parameters.begin(); it != parameters.end(); it++) {
  81                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  82                         }
  83
  84                         //initialize outputTypes
  85                         vector<string> tempOutNames;
  86                         outputTypes["fasta"] = tempOutNames;
  87                         outputTypes["name"] = 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                         inFastaName = validParameter.validFile(parameters, "fasta", true);
 114                         if (inFastaName == "not open") { abort = true; }
 115                         else if (inFastaName == "not found") {
 116                                 inFastaName = m->getFastaFile();
 117                                 if (inFastaName != "") { m->mothurOut("Using " + inFastaName + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 118                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 119                         }else { m->setFastaFile(inFastaName); }
 120
 121                         //if the user changes the output directory command factory will send this info to us in the output parameter
 122                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 123                                 outputDir = "";
 124                                 outputDir += m->hasPath(inFastaName); //if user entered a file with a path then preserve it
 125                         }
 126
 127                         oldNameMapFName = validParameter.validFile(parameters, "name", true);
 128                         if (oldNameMapFName == "not open") { abort = true; }
 129                         else if (oldNameMapFName == "not found"){       oldNameMapFName = "";   }
 130                         else { m->setNameFile(oldNameMapFName); }
 131                 }
 132
 133         }
 134         catch(exception& e) {
 135                 m->errorOut(e, "DeconvoluteCommand", "DeconvoluteCommand");
 136                 exit(1);
 137         }
 138 }
 139 /**************************************************************************************/
 140 int DeconvoluteCommand::execute() {
 141         try {
 142
 143                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 144
 145                 //prepare filenames and open files
 146                 string outNameFile = outputDir + m->getRootName(m->getSimpleName(inFastaName)) + "names";
 147                 string outFastaFile = outputDir + m->getRootName(m->getSimpleName(inFastaName)) + "unique" + m->getExtension(inFastaName);
 148
 149                 map<string, string> nameMap;
 150                 map<string, string>::iterator itNames;
 151                 if (oldNameMapFName != "")  {  m->readNames(oldNameMapFName, nameMap); }
 152
 153                 if (m->control_pressed) { return 0; }
 154
 155                 ifstream in;
 156                 m->openInputFile(inFastaName, in);
 157
 158                 ofstream outFasta;
 159                 m->openOutputFile(outFastaFile, outFasta);
 160
 161                 map<string, string> sequenceStrings; //sequenceString -> list of names.  "atgc...." -> seq1,seq2,seq3.
 162                 map<string, string>::iterator itStrings;
 163                 set<string> nameInFastaFile; //for sanity checking
 164                 set<string>::iterator itname;
 165                 vector<string> nameFileOrder;
 166                 int count = 0;
 167                 while (!in.eof()) {
 168
 169                         if (m->control_pressed) { in.close(); outFasta.close(); m->mothurRemove(outFastaFile); return 0; }
 170
 171                         Sequence seq(in);
 172
 173                         if (seq.getName() != "") {
 174
 175                                 //sanity checks
 176                                 itname = nameInFastaFile.find(seq.getName());
 177                                 if (itname == nameInFastaFile.end()) { nameInFastaFile.insert(seq.getName());  }
 178                                 else { m->mothurOut("[ERROR]: You already have a sequence named " + seq.getName() + " in your fasta file, sequence names must be unique, please correct."); m->mothurOutEndLine(); }
 179
 180                                 itStrings = sequenceStrings.find(seq.getAligned());
 181
 182                                 if (itStrings == sequenceStrings.end()) { //this is a new unique sequence
 183                                         //output to unique fasta file
 184                                         seq.printSequence(outFasta);
 185
 186                                         if (oldNameMapFName != "") {
 187                                                 itNames = nameMap.find(seq.getName());
 188
 189                                                 if (itNames == nameMap.end()) { //namefile and fastafile do not match
 190                                                         m->mothurOut("[ERROR]: " + seq.getName() + " is in your fasta file, and not in your namefile, please correct."); m->mothurOutEndLine();
 191                                                 }else {
 192                                                         sequenceStrings[seq.getAligned()] = itNames->second;
 193                                                         nameFileOrder.push_back(seq.getAligned());
 194                                                 }
 195                                         }else { sequenceStrings[seq.getAligned()] = seq.getName();      nameFileOrder.push_back(seq.getAligned()); }
 196                                 }else { //this is a dup
 197                                         if (oldNameMapFName != "") {
 198                                                 itNames = nameMap.find(seq.getName());
 199
 200                                                 if (itNames == nameMap.end()) { //namefile and fastafile do not match
 201                                                         m->mothurOut("[ERROR]: " + seq.getName() + " is in your fasta file, and not in your namefile, please correct."); m->mothurOutEndLine();
 202                                                 }else {
 203                                                         sequenceStrings[seq.getAligned()] += "," + itNames->second;
 204                                                 }
 205                                         }else { sequenceStrings[seq.getAligned()] += "," + seq.getName();       }
 206                                 }
 207
 208                                 count++;
 209                         }
 210
 211                         m->gobble(in);
 212
 213                         if(count % 1000 == 0)   { m->mothurOut(toString(count) + "\t" + toString(sequenceStrings.size())); m->mothurOutEndLine();       }
 214                 }
 215
 216                 if(count % 1000 != 0)   { m->mothurOut(toString(count) + "\t" + toString(sequenceStrings.size())); m->mothurOutEndLine();       }
 217
 218                 in.close();
 219                 outFasta.close();
 220
 221                 if (m->control_pressed) { m->mothurRemove(outFastaFile); return 0; }
 222
 223                 //print new names file
 224                 ofstream outNames;
 225                 m->openOutputFile(outNameFile, outNames);
 226
 227                 for (int i = 0; i < nameFileOrder.size(); i++) {
 228                 //for (itStrings = sequenceStrings.begin(); itStrings != sequenceStrings.end(); itStrings++) {
 229                         if (m->control_pressed) { outputTypes.clear(); m->mothurRemove(outFastaFile); outNames.close(); m->mothurRemove(outNameFile); return 0; }
 230
 231                         itStrings = sequenceStrings.find(nameFileOrder[i]);
 232
 233                         if (itStrings != sequenceStrings.end()) {
 234                                 //get rep name
 235                                 int pos = (itStrings->second).find_first_of(',');
 236
 237                                 if (pos == string::npos) { // only reps itself
 238                                         outNames << itStrings->second << '\t' << itStrings->second << endl;
 239                                 }else {
 240                                         outNames << (itStrings->second).substr(0, pos) << '\t' << itStrings->second << endl;
 241                                 }
 242                         }else{ m->mothurOut("[ERROR]: mismatch in namefile print."); m->mothurOutEndLine(); m->control_pressed = true; }
 243                 }
 244                 outNames.close();
 245
 246                 if (m->control_pressed) { outputTypes.clear(); m->mothurRemove(outFastaFile); m->mothurRemove(outNameFile); return 0; }
 247
 248                 m->mothurOutEndLine();
 249                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 250                 m->mothurOut(outFastaFile); m->mothurOutEndLine();
 251                 m->mothurOut(outNameFile); m->mothurOutEndLine();
 252                 outputNames.push_back(outFastaFile);  outputNames.push_back(outNameFile); outputTypes["fasta"].push_back(outFastaFile);  outputTypes["name"].push_back(outNameFile);
 253                 m->mothurOutEndLine();
 254
 255                 //set fasta file as new current fastafile
 256                 string current = "";
 257                 itTypes = outputTypes.find("fasta");
 258                 if (itTypes != outputTypes.end()) {
 259                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 260                 }
 261
 262                 itTypes = outputTypes.find("name");
 263                 if (itTypes != outputTypes.end()) {
 264                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 265                 }
 266
 267                 return 0;
 268         }
 269         catch(exception& e) {
 270                 m->errorOut(e, "DeconvoluteCommand", "execute");
 271                 exit(1);
 272         }
 273 }
 274 /**************************************************************************************/