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", "", "", "namecount", "none", "none",false,false); parameters.push_back(pname);
  18         CommandParameter pcount("count", "InputTypes", "", "", "namecount", "none", "none",false,false); 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::getOutputFileNameTag(string type, string inputName=""){
  50         try {
  51         string outputFileName = "";
  52                 map<string, vector<string> >::iterator it;
  53
  54         //is this a type this command creates
  55         it = outputTypes.find(type);
  56         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  57         else {
  58             if (type == "fasta") {  outputFileName =  "unique" + m->getExtension(inputName); }
  59             else if (type == "name") {  outputFileName =  "names"; }
  60             else if (type == "count") {  outputFileName =  "count_table"; }
  61             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  62         }
  63         return outputFileName;
  64         }
  65         catch(exception& e) {
  66                 m->errorOut(e, "DeconvoluteCommand", "getOutputFileNameTag");
  67                 exit(1);
  68         }
  69 }
  70 //**********************************************************************************************************************
  71 DeconvoluteCommand::DeconvoluteCommand(){
  72         try {
  73                 abort = true; calledHelp = true;
  74                 setParameters();
  75                 vector<string> tempOutNames;
  76                 outputTypes["fasta"] = tempOutNames;
  77                 outputTypes["name"] = tempOutNames;
  78         outputTypes["count"] = tempOutNames;
  79         }
  80         catch(exception& e) {
  81                 m->errorOut(e, "DeconvoluteCommand", "DeconvoluteCommand");
  82                 exit(1);
  83         }
  84 }
  85 /**************************************************************************************/
  86 DeconvoluteCommand::DeconvoluteCommand(string option)  {
  87         try {
  88                 abort = false; calledHelp = false;
  89
  90                 //allow user to run help
  91                 if(option == "help") { help(); abort = true; calledHelp = true; }
  92                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  93
  94                 else {
  95                         vector<string> myArray = setParameters();
  96
  97                         OptionParser parser(option);
  98                         map<string,string> parameters = parser.getParameters();
  99
 100                         ValidParameters validParameter;
 101                         map<string, string>::iterator it;
 102
 103                         //check to make sure all parameters are valid for command
 104                         for (it = parameters.begin(); it != parameters.end(); it++) {
 105                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 106                         }
 107
 108                         //initialize outputTypes
 109                         vector<string> tempOutNames;
 110                         outputTypes["fasta"] = tempOutNames;
 111                         outputTypes["name"] = tempOutNames;
 112             outputTypes["count"] = tempOutNames;
 113
 114                         //if the user changes the input directory command factory will send this info to us in the output parameter
 115                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 116                         if (inputDir == "not found"){   inputDir = "";          }
 117                         else {
 118                                 string path;
 119                                 it = parameters.find("fasta");
 120                                 //user has given a template file
 121                                 if(it != parameters.end()){
 122                                         path = m->hasPath(it->second);
 123                                         //if the user has not given a path then, add inputdir. else leave path alone.
 124                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 125                                 }
 126
 127                                 it = parameters.find("name");
 128                                 //user has given a template file
 129                                 if(it != parameters.end()){
 130                                         path = m->hasPath(it->second);
 131                                         //if the user has not given a path then, add inputdir. else leave path alone.
 132                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 133                                 }
 134
 135                 it = parameters.find("count");
 136                                 //user has given a template file
 137                                 if(it != parameters.end()){
 138                                         path = m->hasPath(it->second);
 139                                         //if the user has not given a path then, add inputdir. else leave path alone.
 140                                         if (path == "") {       parameters["count"] = inputDir + it->second;            }
 141                                 }
 142                         }
 143
 144
 145                         //check for required parameters
 146                         inFastaName = validParameter.validFile(parameters, "fasta", true);
 147                         if (inFastaName == "not open") { abort = true; }
 148                         else if (inFastaName == "not found") {
 149                                 inFastaName = m->getFastaFile();
 150                                 if (inFastaName != "") { m->mothurOut("Using " + inFastaName + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 151                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 152                         }else { m->setFastaFile(inFastaName); }
 153
 154                         //if the user changes the output directory command factory will send this info to us in the output parameter
 155                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 156                                 outputDir = "";
 157                                 outputDir += m->hasPath(inFastaName); //if user entered a file with a path then preserve it
 158                         }
 159
 160                         oldNameMapFName = validParameter.validFile(parameters, "name", true);
 161                         if (oldNameMapFName == "not open") { oldNameMapFName = ""; abort = true; }
 162                         else if (oldNameMapFName == "not found"){       oldNameMapFName = "";   }
 163                         else { m->setNameFile(oldNameMapFName); }
 164
 165             countfile = validParameter.validFile(parameters, "count", true);
 166                         if (countfile == "not open") { abort = true; countfile = ""; }
 167                         else if (countfile == "not found") { countfile = ""; }
 168                         else { m->setCountTableFile(countfile); }
 169
 170             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; }
 171
 172
 173                         if (countfile == "") {
 174                 if (oldNameMapFName == "") {
 175                     vector<string> files; files.push_back(inFastaName);
 176                     parser.getNameFile(files);
 177                 }
 178             }
 179
 180                 }
 181
 182         }
 183         catch(exception& e) {
 184                 m->errorOut(e, "DeconvoluteCommand", "DeconvoluteCommand");
 185                 exit(1);
 186         }
 187 }
 188 /**************************************************************************************/
 189 int DeconvoluteCommand::execute() {
 190         try {
 191
 192                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 193
 194                 //prepare filenames and open files
 195                 string outNameFile = outputDir + m->getRootName(m->getSimpleName(inFastaName)) + getOutputFileNameTag("name");
 196         string outCountFile = outputDir + m->getRootName(m->getSimpleName(inFastaName)) + getOutputFileNameTag("count");
 197                 string outFastaFile = outputDir + m->getRootName(m->getSimpleName(inFastaName)) + getOutputFileNameTag("fasta", inFastaName);
 198
 199                 map<string, string> nameMap;
 200                 map<string, string>::iterator itNames;
 201                 if (oldNameMapFName != "")  {
 202             m->readNames(oldNameMapFName, nameMap);
 203             if (oldNameMapFName == outNameFile){ outNameFile = outputDir + m->getRootName(m->getSimpleName(inFastaName)) + "unique." + getOutputFileNameTag("name");   }
 204         }
 205         CountTable ct;
 206         if (countfile != "")  {
 207             ct.readTable(countfile);
 208             if (countfile == outCountFile){ outCountFile = outputDir + m->getRootName(m->getSimpleName(inFastaName)) + "unique." + getOutputFileNameTag("count");   }
 209         }
 210
 211                 if (m->control_pressed) { return 0; }
 212
 213                 ifstream in;
 214                 m->openInputFile(inFastaName, in);
 215
 216                 ofstream outFasta;
 217                 m->openOutputFile(outFastaFile, outFasta);
 218
 219                 map<string, string> sequenceStrings; //sequenceString -> list of names.  "atgc...." -> seq1,seq2,seq3.
 220                 map<string, string>::iterator itStrings;
 221                 set<string> nameInFastaFile; //for sanity checking
 222                 set<string>::iterator itname;
 223                 vector<string> nameFileOrder;
 224                 int count = 0;
 225                 while (!in.eof()) {
 226
 227                         if (m->control_pressed) { in.close(); outFasta.close(); m->mothurRemove(outFastaFile); return 0; }
 228
 229                         Sequence seq(in);
 230
 231                         if (seq.getName() != "") {
 232
 233                                 //sanity checks
 234                                 itname = nameInFastaFile.find(seq.getName());
 235                                 if (itname == nameInFastaFile.end()) { nameInFastaFile.insert(seq.getName());  }
 236                                 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(); }
 237
 238                                 itStrings = sequenceStrings.find(seq.getAligned());
 239
 240                                 if (itStrings == sequenceStrings.end()) { //this is a new unique sequence
 241                                         //output to unique fasta file
 242                                         seq.printSequence(outFasta);
 243
 244                                         if (oldNameMapFName != "") {
 245                                                 itNames = nameMap.find(seq.getName());
 246
 247                                                 if (itNames == nameMap.end()) { //namefile and fastafile do not match
 248                                                         m->mothurOut("[ERROR]: " + seq.getName() + " is in your fasta file, and not in your namefile, please correct."); m->mothurOutEndLine();
 249                                                 }else {
 250                                                         sequenceStrings[seq.getAligned()] = itNames->second;
 251                                                         nameFileOrder.push_back(seq.getAligned());
 252                                                 }
 253                                         }else if (countfile != "") {
 254                         ct.getNumSeqs(seq.getName()); //checks to make sure seq is in table
 255                         sequenceStrings[seq.getAligned()] = seq.getName();      nameFileOrder.push_back(seq.getAligned());
 256                     }else {     sequenceStrings[seq.getAligned()] = seq.getName();      nameFileOrder.push_back(seq.getAligned()); }
 257                                 }else { //this is a dup
 258                                         if (oldNameMapFName != "") {
 259                                                 itNames = nameMap.find(seq.getName());
 260
 261                                                 if (itNames == nameMap.end()) { //namefile and fastafile do not match
 262                                                         m->mothurOut("[ERROR]: " + seq.getName() + " is in your fasta file, and not in your namefile, please correct."); m->mothurOutEndLine();
 263                                                 }else {
 264                                                         sequenceStrings[seq.getAligned()] += "," + itNames->second;
 265                                                 }
 266                     }else if (countfile != "") {
 267                         int num = ct.getNumSeqs(seq.getName()); //checks to make sure seq is in table
 268                         if (num != 0) { //its in the table
 269                             ct.mergeCounts(itStrings->second, seq.getName()); //merges counts and saves in uniques name
 270                         }
 271                     }else {     sequenceStrings[seq.getAligned()] += "," + seq.getName();       }
 272                                 }
 273
 274                                 count++;
 275                         }
 276
 277                         m->gobble(in);
 278
 279                         if(count % 1000 == 0)   { m->mothurOut(toString(count) + "\t" + toString(sequenceStrings.size())); m->mothurOutEndLine();       }
 280                 }
 281
 282                 if(count % 1000 != 0)   { m->mothurOut(toString(count) + "\t" + toString(sequenceStrings.size())); m->mothurOutEndLine();       }
 283
 284                 in.close();
 285                 outFasta.close();
 286
 287                 if (m->control_pressed) { m->mothurRemove(outFastaFile); return 0; }
 288
 289                 //print new names file
 290                 ofstream outNames;
 291                 if (countfile == "") { m->openOutputFile(outNameFile, outNames); outputNames.push_back(outNameFile); outputTypes["name"].push_back(outNameFile);  }
 292         else { m->openOutputFile(outCountFile, outNames); ct.printHeaders(outNames); outputTypes["count"].push_back(outCountFile); outputNames.push_back(outCountFile); }
 293
 294                 for (int i = 0; i < nameFileOrder.size(); i++) {
 295                         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; }
 296
 297                         itStrings = sequenceStrings.find(nameFileOrder[i]);
 298
 299                         if (itStrings != sequenceStrings.end()) {
 300                 if (countfile == "") {
 301                     //get rep name
 302                     int pos = (itStrings->second).find_first_of(',');
 303
 304                     if (pos == string::npos) { // only reps itself
 305                         outNames << itStrings->second << '\t' << itStrings->second << endl;
 306                     }else {
 307                         outNames << (itStrings->second).substr(0, pos) << '\t' << itStrings->second << endl;
 308                     }
 309                 }else {  ct.printSeq(outNames, itStrings->second);  }
 310                         }else{ m->mothurOut("[ERROR]: mismatch in namefile print."); m->mothurOutEndLine(); m->control_pressed = true; }
 311                 }
 312                 outNames.close();
 313
 314                 if (m->control_pressed) { outputTypes.clear(); m->mothurRemove(outFastaFile); for (int j = 0; j < outputNames.size(); j++) { m->mothurRemove(outputNames[j]); }  return 0; }
 315
 316                 m->mothurOutEndLine();
 317                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 318                 outputNames.push_back(outFastaFile);   outputTypes["fasta"].push_back(outFastaFile);
 319         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 320                 m->mothurOutEndLine();
 321
 322                 //set fasta file as new current fastafile
 323                 string current = "";
 324                 itTypes = outputTypes.find("fasta");
 325                 if (itTypes != outputTypes.end()) {
 326                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 327                 }
 328
 329                 itTypes = outputTypes.find("name");
 330                 if (itTypes != outputTypes.end()) {
 331                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 332                 }
 333
 334         itTypes = outputTypes.find("count");
 335                 if (itTypes != outputTypes.end()) {
 336                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setCountTableFile(current); }
 337                 }
 338
 339                 return 0;
 340         }
 341         catch(exception& e) {
 342                 m->errorOut(e, "DeconvoluteCommand", "execute");
 343                 exit(1);
 344         }
 345 }
 346 /**************************************************************************************/