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","fasta-name",false,true,true); parameters.push_back(pfasta);
  17                 CommandParameter pname("name", "InputTypes", "", "", "namecount", "none", "none","name",false,false,true); parameters.push_back(pname);
  18         CommandParameter pcount("count", "InputTypes", "", "", "namecount", "none", "none","count",false,false,true); parameters.push_back(pcount);
  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, "DeconvoluteCommand", "setParameters");
  28                 exit(1);
  29         }
  30 }
  31 //**********************************************************************************************************************
  32 string DeconvoluteCommand::getHelpString(){
  33         try {
  34                 string helpString = "";
  35                 helpString += "The unique.seqs command reads a fastafile and creates a name or count file.\n";
  36                 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";
  37                 helpString += "If the sequence is unique the second column will just contain its name. \n";
  38                 helpString += "The unique.seqs command parameters are fasta and name.  fasta is required, unless there is a valid current fasta file.\n";
  39                 helpString += "The unique.seqs command should be in the following format: \n";
  40                 helpString += "unique.seqs(fasta=yourFastaFile) \n";
  41                 return helpString;
  42         }
  43         catch(exception& e) {
  44                 m->errorOut(e, "DeconvoluteCommand", "getHelpString");
  45                 exit(1);
  46         }
  47 }
  48 //**********************************************************************************************************************
  49 string DeconvoluteCommand::getOutputPattern(string type) {
  50     try {
  51         string pattern = "";
  52
  53         if (type == "fasta") {  pattern = "[filename],unique,[extension]"; }
  54         else if (type == "name") {  pattern = "[filename],names-[filename],[tag],names"; }
  55         else if (type == "count") {  pattern = "[filename],count_table-[filename],[tag],count_table"; }
  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, "DeconvoluteCommand", "getOutputPattern");
  62         exit(1);
  63     }
  64 }
  65
  66 //**********************************************************************************************************************
  67 DeconvoluteCommand::DeconvoluteCommand(){
  68         try {
  69                 abort = true; calledHelp = true;
  70                 setParameters();
  71                 vector<string> tempOutNames;
  72                 outputTypes["fasta"] = tempOutNames;
  73                 outputTypes["name"] = tempOutNames;
  74         outputTypes["count"] = tempOutNames;
  75         }
  76         catch(exception& e) {
  77                 m->errorOut(e, "DeconvoluteCommand", "DeconvoluteCommand");
  78                 exit(1);
  79         }
  80 }
  81 /**************************************************************************************/
  82 DeconvoluteCommand::DeconvoluteCommand(string option)  {
  83         try {
  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["fasta"] = tempOutNames;
 107                         outputTypes["name"] = tempOutNames;
 108             outputTypes["count"] = tempOutNames;
 109
 110                         //if the user changes the input directory command factory will send this info to us in the output parameter
 111                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 112                         if (inputDir == "not found"){   inputDir = "";          }
 113                         else {
 114                                 string path;
 115                                 it = parameters.find("fasta");
 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["fasta"] = inputDir + it->second;            }
 121                                 }
 122
 123                                 it = parameters.find("name");
 124                                 //user has given a template file
 125                                 if(it != parameters.end()){
 126                                         path = m->hasPath(it->second);
 127                                         //if the user has not given a path then, add inputdir. else leave path alone.
 128                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 129                                 }
 130
 131                 it = parameters.find("count");
 132                                 //user has given a template file
 133                                 if(it != parameters.end()){
 134                                         path = m->hasPath(it->second);
 135                                         //if the user has not given a path then, add inputdir. else leave path alone.
 136                                         if (path == "") {       parameters["count"] = inputDir + it->second;            }
 137                                 }
 138                         }
 139
 140
 141                         //check for required parameters
 142                         inFastaName = validParameter.validFile(parameters, "fasta", true);
 143                         if (inFastaName == "not open") { abort = true; }
 144                         else if (inFastaName == "not found") {
 145                                 inFastaName = m->getFastaFile();
 146                                 if (inFastaName != "") { m->mothurOut("Using " + inFastaName + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 147                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 148                         }else { m->setFastaFile(inFastaName); }
 149
 150                         //if the user changes the output directory command factory will send this info to us in the output parameter
 151                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 152                                 outputDir = "";
 153                                 outputDir += m->hasPath(inFastaName); //if user entered a file with a path then preserve it
 154                         }
 155
 156                         oldNameMapFName = validParameter.validFile(parameters, "name", true);
 157                         if (oldNameMapFName == "not open") { oldNameMapFName = ""; abort = true; }
 158                         else if (oldNameMapFName == "not found"){       oldNameMapFName = "";   }
 159                         else { m->setNameFile(oldNameMapFName); }
 160
 161             countfile = validParameter.validFile(parameters, "count", true);
 162                         if (countfile == "not open") { abort = true; countfile = ""; }
 163                         else if (countfile == "not found") { countfile = ""; }
 164                         else { m->setCountTableFile(countfile); }
 165
 166             if ((countfile != "") && (oldNameMapFName != "")) { m->mothurOut("When executing a unique.seqs command you must enter ONLY ONE of the following: count or name."); m->mothurOutEndLine(); abort = true; }
 167
 168
 169                         if (countfile == "") {
 170                 if (oldNameMapFName == "") {
 171                     vector<string> files; files.push_back(inFastaName);
 172                     parser.getNameFile(files);
 173                 }
 174             }
 175
 176                 }
 177
 178         }
 179         catch(exception& e) {
 180                 m->errorOut(e, "DeconvoluteCommand", "DeconvoluteCommand");
 181                 exit(1);
 182         }
 183 }
 184 /**************************************************************************************/
 185 int DeconvoluteCommand::execute() {
 186         try {
 187
 188                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 189
 190                 //prepare filenames and open files
 191         map<string, string> variables;
 192         variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(inFastaName));
 193         string outNameFile = getOutputFileName("name", variables);
 194         string outCountFile = getOutputFileName("count", variables);
 195         variables["[extension]"] = m->getExtension(inFastaName);
 196                 string outFastaFile = getOutputFileName("fasta", variables);
 197
 198                 map<string, string> nameMap;
 199                 map<string, string>::iterator itNames;
 200                 if (oldNameMapFName != "")  {
 201             m->readNames(oldNameMapFName, nameMap);
 202             if (oldNameMapFName == outNameFile){
 203                 variables["[tag]"] = "unique";
 204                 outNameFile = getOutputFileName("name", variables);   }
 205         }
 206         CountTable ct;
 207         if (countfile != "")  {
 208             ct.readTable(countfile);
 209             if (countfile == outCountFile){
 210                 variables["[tag]"] = "unique";
 211                 outCountFile = getOutputFileName("count", variables);   }
 212         }
 213
 214                 if (m->control_pressed) { return 0; }
 215
 216                 ifstream in;
 217                 m->openInputFile(inFastaName, in);
 218
 219                 ofstream outFasta;
 220                 m->openOutputFile(outFastaFile, outFasta);
 221
 222                 map<string, string> sequenceStrings; //sequenceString -> list of names.  "atgc...." -> seq1,seq2,seq3.
 223                 map<string, string>::iterator itStrings;
 224                 set<string> nameInFastaFile; //for sanity checking
 225                 set<string>::iterator itname;
 226                 vector<string> nameFileOrder;
 227                 int count = 0;
 228                 while (!in.eof()) {
 229
 230                         if (m->control_pressed) { in.close(); outFasta.close(); m->mothurRemove(outFastaFile); return 0; }
 231
 232                         Sequence seq(in);
 233
 234                         if (seq.getName() != "") {
 235
 236                                 //sanity checks
 237                                 itname = nameInFastaFile.find(seq.getName());
 238                                 if (itname == nameInFastaFile.end()) { nameInFastaFile.insert(seq.getName());  }
 239                                 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(); }
 240
 241                                 itStrings = sequenceStrings.find(seq.getAligned());
 242
 243                                 if (itStrings == sequenceStrings.end()) { //this is a new unique sequence
 244                                         //output to unique fasta file
 245                                         seq.printSequence(outFasta);
 246
 247                                         if (oldNameMapFName != "") {
 248                                                 itNames = nameMap.find(seq.getName());
 249
 250                                                 if (itNames == nameMap.end()) { //namefile and fastafile do not match
 251                                                         m->mothurOut("[ERROR]: " + seq.getName() + " is in your fasta file, and not in your namefile, please correct."); m->mothurOutEndLine();
 252                                                 }else {
 253                                                         sequenceStrings[seq.getAligned()] = itNames->second;
 254                                                         nameFileOrder.push_back(seq.getAligned());
 255                                                 }
 256                                         }else if (countfile != "") {
 257                         ct.getNumSeqs(seq.getName()); //checks to make sure seq is in table
 258                         sequenceStrings[seq.getAligned()] = seq.getName();      nameFileOrder.push_back(seq.getAligned());
 259                     }else {     sequenceStrings[seq.getAligned()] = seq.getName();      nameFileOrder.push_back(seq.getAligned()); }
 260                                 }else { //this is a dup
 261                                         if (oldNameMapFName != "") {
 262                                                 itNames = nameMap.find(seq.getName());
 263
 264                                                 if (itNames == nameMap.end()) { //namefile and fastafile do not match
 265                                                         m->mothurOut("[ERROR]: " + seq.getName() + " is in your fasta file, and not in your namefile, please correct."); m->mothurOutEndLine();
 266                                                 }else {
 267                                                         sequenceStrings[seq.getAligned()] += "," + itNames->second;
 268                                                 }
 269                     }else if (countfile != "") {
 270                         int num = ct.getNumSeqs(seq.getName()); //checks to make sure seq is in table
 271                         if (num != 0) { //its in the table
 272                             ct.mergeCounts(itStrings->second, seq.getName()); //merges counts and saves in uniques name
 273                         }
 274                     }else {     sequenceStrings[seq.getAligned()] += "," + seq.getName();       }
 275                                 }
 276
 277                                 count++;
 278                         }
 279
 280                         m->gobble(in);
 281
 282                         if(count % 1000 == 0)   { m->mothurOut(toString(count) + "\t" + toString(sequenceStrings.size())); m->mothurOutEndLine();       }
 283                 }
 284
 285                 if(count % 1000 != 0)   { m->mothurOut(toString(count) + "\t" + toString(sequenceStrings.size())); m->mothurOutEndLine();       }
 286
 287                 in.close();
 288                 outFasta.close();
 289
 290                 if (m->control_pressed) { m->mothurRemove(outFastaFile); return 0; }
 291
 292                 //print new names file
 293                 ofstream outNames;
 294                 if (countfile == "") { m->openOutputFile(outNameFile, outNames); outputNames.push_back(outNameFile); outputTypes["name"].push_back(outNameFile);  }
 295         else { m->openOutputFile(outCountFile, outNames); ct.printHeaders(outNames); outputTypes["count"].push_back(outCountFile); outputNames.push_back(outCountFile); }
 296
 297                 for (int i = 0; i < nameFileOrder.size(); i++) {
 298                         if (m->control_pressed) { outputTypes.clear(); m->mothurRemove(outFastaFile); outNames.close(); for (int j = 0; j < outputNames.size(); j++) { m->mothurRemove(outputNames[j]); } return 0; }
 299
 300                         itStrings = sequenceStrings.find(nameFileOrder[i]);
 301
 302                         if (itStrings != sequenceStrings.end()) {
 303                 if (countfile == "") {
 304                     //get rep name
 305                     int pos = (itStrings->second).find_first_of(',');
 306
 307                     if (pos == string::npos) { // only reps itself
 308                         outNames << itStrings->second << '\t' << itStrings->second << endl;
 309                     }else {
 310                         outNames << (itStrings->second).substr(0, pos) << '\t' << itStrings->second << endl;
 311                     }
 312                 }else {  ct.printSeq(outNames, itStrings->second);  }
 313                         }else{ m->mothurOut("[ERROR]: mismatch in namefile print."); m->mothurOutEndLine(); m->control_pressed = true; }
 314                 }
 315                 outNames.close();
 316
 317                 if (m->control_pressed) { outputTypes.clear(); m->mothurRemove(outFastaFile); for (int j = 0; j < outputNames.size(); j++) { m->mothurRemove(outputNames[j]); }  return 0; }
 318
 319                 m->mothurOutEndLine();
 320                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 321                 outputNames.push_back(outFastaFile);   outputTypes["fasta"].push_back(outFastaFile);
 322         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 323                 m->mothurOutEndLine();
 324
 325                 //set fasta file as new current fastafile
 326                 string current = "";
 327                 itTypes = outputTypes.find("fasta");
 328                 if (itTypes != outputTypes.end()) {
 329                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 330                 }
 331
 332                 itTypes = outputTypes.find("name");
 333                 if (itTypes != outputTypes.end()) {
 334                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 335                 }
 336
 337         itTypes = outputTypes.find("count");
 338                 if (itTypes != outputTypes.end()) {
 339                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setCountTableFile(current); }
 340                 }
 341
 342                 return 0;
 343         }
 344         catch(exception& e) {
 345                 m->errorOut(e, "DeconvoluteCommand", "execute");
 346                 exit(1);
 347         }
 348 }
 349 /**************************************************************************************/