binsequencecommand.cpp

   1 /*
   2  *  binsequencecommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 4/3/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amhers. All rights reserved.
   7  *
   8  */
   9
  10 #include "binsequencecommand.h"
  11
  12
  13 //**********************************************************************************************************************
  14 vector<string> BinSeqCommand::setParameters(){
  15         try {
  16                 CommandParameter plist("list", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(plist);
  17                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfasta);
  18                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
  19                 CommandParameter pgroup("group", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pgroup);
  20                 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
  21                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  22                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  23
  24                 vector<string> myArray;
  25                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  26                 return myArray;
  27         }
  28         catch(exception& e) {
  29                 m->errorOut(e, "BinSeqCommand", "setParameters");
  30                 exit(1);
  31         }
  32 }
  33 //**********************************************************************************************************************
  34 string BinSeqCommand::getHelpString(){
  35         try {
  36                 string helpString = "";
  37                 helpString += "The bin.seqs command parameters are list, fasta, name, label and group.  The fasta and list are required, unless you have a valid current list and fasta file.\n";
  38                 helpString += "The label parameter allows you to select what distance levels you would like a output files created for, and are separated by dashes.\n";
  39                 helpString += "The bin.seqs command should be in the following format: bin.seqs(fasta=yourFastaFile, name=yourNamesFile, group=yourGroupFile, label=yourLabels).\n";
  40                 helpString += "Example bin.seqs(fasta=amazon.fasta, group=amazon.groups, name=amazon.names).\n";
  41                 helpString += "The default value for label is all lines in your inputfile.\n";
  42                 helpString += "The bin.seqs command outputs a .fasta file for each distance you specify appending the OTU number to each name.\n";
  43                 helpString += "If you provide a groupfile, then it also appends the sequences group to the name.\n";
  44                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n";
  45                 return helpString;
  46         }
  47         catch(exception& e) {
  48                 m->errorOut(e, "BinSeqCommand", "getHelpString");
  49                 exit(1);
  50         }
  51 }
  52 //**********************************************************************************************************************
  53 BinSeqCommand::BinSeqCommand(){
  54         try {
  55                 abort = true; calledHelp = true;
  56                 setParameters();
  57                 vector<string> tempOutNames;
  58                 outputTypes["fasta"] = tempOutNames;
  59         }
  60         catch(exception& e) {
  61                 m->errorOut(e, "BinSeqCommand", "BinSeqCommand");
  62                 exit(1);
  63         }
  64 }
  65 //**********************************************************************************************************************
  66 BinSeqCommand::BinSeqCommand(string option) {
  67         try {
  68                 abort = false; calledHelp = false;
  69                 allLines = 1;
  70                 labels.clear();
  71
  72                 //allow user to run help
  73                 if(option == "help") { help(); abort = true; calledHelp = true; }
  74                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  75
  76                 else {
  77                         vector<string> myArray = setParameters();
  78
  79                         OptionParser parser(option);
  80                         map<string, string> parameters = parser.getParameters();
  81
  82                         ValidParameters validParameter;
  83                         map<string, string>::iterator it;
  84
  85                         //check to make sure all parameters are valid for command
  86                         for (it = parameters.begin(); it != parameters.end(); it++) {
  87                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  88                         }
  89
  90                         //initialize outputTypes
  91                         vector<string> tempOutNames;
  92                         outputTypes["fasta"] = tempOutNames;
  93
  94                         //if the user changes the input directory command factory will send this info to us in the output parameter
  95                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  96                         if (inputDir == "not found"){   inputDir = "";          }
  97                         else {
  98                                 string path;
  99                                 it = parameters.find("fasta");
 100                                 //user has given a template file
 101                                 if(it != parameters.end()){
 102                                         path = m->hasPath(it->second);
 103                                         //if the user has not given a path then, add inputdir. else leave path alone.
 104                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 105                                 }
 106
 107                                 it = parameters.find("list");
 108                                 //user has given a template file
 109                                 if(it != parameters.end()){
 110                                         path = m->hasPath(it->second);
 111                                         //if the user has not given a path then, add inputdir. else leave path alone.
 112                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 113                                 }
 114
 115                                 it = parameters.find("name");
 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["name"] = inputDir + it->second;             }
 121                                 }
 122
 123                                 it = parameters.find("group");
 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["group"] = inputDir + it->second;            }
 129                                 }
 130                         }
 131
 132
 133                         //check for required parameters
 134                         fastafile = validParameter.validFile(parameters, "fasta", true);
 135                         if (fastafile == "not found") {                                 //if there is a current phylip file, use it
 136                                 fastafile = m->getFastaFile();
 137                                 if (fastafile != "") { m->mothurOut("Using " + fastafile + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 138                                 else {  m->mothurOut("You have no current fasta file and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 139                         }
 140                         else if (fastafile == "not open") { abort = true; }
 141                         else { m->setFastaFile(fastafile); }
 142
 143                         listfile = validParameter.validFile(parameters, "list", true);
 144                         if (listfile == "not found") {
 145                                 listfile = m->getListFile();
 146                                 if (listfile != "") { m->mothurOut("Using " + listfile + " as input file for the list parameter."); m->mothurOutEndLine(); }
 147                                 else {  m->mothurOut("You have no current list file and the list parameter is required."); m->mothurOutEndLine(); abort = true; }
 148                         }
 149                         else if (listfile == "not open") { listfile = ""; abort = true; }
 150                         else { m->setListFile(listfile); }
 151
 152                         //if the user changes the output directory command factory will send this info to us in the output parameter
 153                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 154                                 outputDir = "";
 155                                 outputDir += m->hasPath(listfile); //if user entered a file with a path then preserve it
 156                         }
 157
 158
 159                         //check for optional parameter and set defaults
 160                         // ...at some point should added some additional type checking...
 161
 162                         label = validParameter.validFile(parameters, "label", false);
 163                         if (label == "not found") { label = ""; }
 164                         else {
 165                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 166                                 else { allLines = 1;  }
 167                         }
 168
 169                         namesfile = validParameter.validFile(parameters, "name", true);
 170                         if (namesfile == "not open") { namesfile = ""; abort = true; }
 171                         else if (namesfile == "not found") { namesfile = ""; }
 172                         else {  m->setNameFile(namesfile); }
 173
 174                         groupfile = validParameter.validFile(parameters, "group", true);
 175                         if (groupfile == "not open") { abort = true; }
 176                         else if (groupfile == "not found") { groupfile = ""; }
 177                         else { m->setGroupFile(groupfile); }
 178
 179                         if (namesfile == ""){
 180                                 vector<string> files; files.push_back(fastafile);
 181                                 parser.getNameFile(files);
 182                         }
 183
 184                 }
 185         }
 186         catch(exception& e) {
 187                 m->errorOut(e, "BinSeqCommand", "BinSeqCommand");
 188                 exit(1);
 189         }
 190 }
 191 //**********************************************************************************************************************
 192
 193 BinSeqCommand::~BinSeqCommand(){}
 194 //**********************************************************************************************************************
 195
 196 int BinSeqCommand::execute(){
 197         try {
 198                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 199
 200                 int error = 0;
 201
 202                 fasta = new FastaMap();
 203                 if (groupfile != "") {
 204                         groupMap = new GroupMap(groupfile);
 205                         groupMap->readMap();
 206                 }
 207
 208                 //read fastafile
 209                 fasta->readFastaFile(fastafile);
 210
 211                 //if user gave a namesfile then use it
 212                 if (namesfile != "") {
 213                         readNamesFile();
 214                 }
 215
 216                 input = new InputData(listfile, "list");
 217                 list = input->getListVector();
 218                 string lastLabel = list->getLabel();
 219
 220                 if (m->control_pressed) {  delete input;  delete fasta; if (groupfile != "") {  delete groupMap;   } return 0; }
 221
 222                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 223                 set<string> processedLabels;
 224                 set<string> userLabels = labels;
 225
 226
 227                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 228
 229                         if(m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]);                } delete input;  delete fasta; if (groupfile != "") {  delete groupMap;   } return 0; }
 230
 231                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
 232
 233                                 error = process(list);
 234                                 if (error == 1) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);                } delete input;  delete fasta; if (groupfile != "") {  delete groupMap;   } return 0; }
 235
 236                                 processedLabels.insert(list->getLabel());
 237                                 userLabels.erase(list->getLabel());
 238                         }
 239
 240                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 241                                 string saveLabel = list->getLabel();
 242
 243                                 delete list;
 244                                 list = input->getListVector(lastLabel);
 245
 246                                 error = process(list);
 247                                 if (error == 1) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);                } delete input;  delete fasta; if (groupfile != "") {  delete groupMap;   } return 0; }
 248
 249                                 processedLabels.insert(list->getLabel());
 250                                 userLabels.erase(list->getLabel());
 251
 252                                 //restore real lastlabel to save below
 253                                 list->setLabel(saveLabel);
 254                         }
 255
 256                         lastLabel = list->getLabel();
 257
 258                         delete list;
 259                         list = input->getListVector();
 260                 }
 261
 262                 if(m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]);                } delete input;  delete fasta; if (groupfile != "") {  delete groupMap;   } return 0; }
 263
 264                 //output error messages about any remaining user labels
 265                 set<string>::iterator it;
 266                 bool needToRun = false;
 267                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 268                         m->mothurOut("Your file does not include the label " + *it);
 269                         if (processedLabels.count(lastLabel) != 1) {
 270                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 271                                 needToRun = true;
 272                         }else {
 273                                 m->mothurOut(". Please refer to " + lastLabel + ".");  m->mothurOutEndLine();
 274                         }
 275                 }
 276
 277                 //run last label if you need to
 278                 if (needToRun == true)  {
 279                         if (list != NULL) {             delete list;    }
 280                         list = input->getListVector(lastLabel);
 281
 282                         error = process(list);
 283                         if (error == 1) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);                } delete input;  delete fasta; if (groupfile != "") {  delete groupMap;   } return 0; }
 284
 285                         delete list;
 286                 }
 287
 288                 delete input;
 289                 delete fasta;
 290                 if (groupfile != "") {  delete groupMap;   }
 291
 292                 if(m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]);                }  return 0; }
 293
 294                 m->mothurOutEndLine();
 295                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 296                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 297                 m->mothurOutEndLine();
 298
 299
 300                 return 0;
 301         }
 302         catch(exception& e) {
 303                 m->errorOut(e, "BinSeqCommand", "execute");
 304                 exit(1);
 305         }
 306 }
 307
 308 //**********************************************************************************************************************
 309 void BinSeqCommand::readNamesFile() {
 310         try {
 311                 vector<string> dupNames;
 312                 m->openInputFile(namesfile, inNames);
 313
 314                 string name, names, sequence;
 315
 316                 while(inNames){
 317                         inNames >> name;                        //read from first column  A
 318                         inNames >> names;               //read from second column  A,B,C,D
 319
 320                         dupNames.clear();
 321
 322                         //parse names into vector
 323                         m->splitAtComma(names, dupNames);
 324
 325                         //store names in fasta map
 326                         sequence = fasta->getSequence(name);
 327                         for (int i = 0; i < dupNames.size(); i++) {
 328                                 fasta->push_back(dupNames[i], sequence);
 329                         }
 330
 331                         m->gobble(inNames);
 332                 }
 333                 inNames.close();
 334
 335         }
 336         catch(exception& e) {
 337                 m->errorOut(e, "BinSeqCommand", "readNamesFile");
 338                 exit(1);
 339         }
 340 }
 341 //**********************************************************************************************************************
 342 //return 1 if error, 0 otherwise
 343 int BinSeqCommand::process(ListVector* list) {
 344         try {
 345                                 string binnames, name, sequence;
 346
 347                                 string outputFileName = outputDir + m->getRootName(m->getSimpleName(listfile)) + list->getLabel() + ".fasta";
 348                                 m->openOutputFile(outputFileName, out);
 349
 350                                 //save to output list of output file names
 351                                 outputNames.push_back(outputFileName);  outputTypes["fasta"].push_back(outputFileName);
 352
 353                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 354
 355                                 //for each bin in the list vector
 356                                 for (int i = 0; i < list->size(); i++) {
 357
 358                                         if (m->control_pressed) {  return 1; }
 359
 360                                         binnames = list->get(i);
 361                                         while (binnames.find_first_of(',') != -1) {
 362                                                 name = binnames.substr(0,binnames.find_first_of(','));
 363                                                 binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 364
 365                                                 //do work for that name
 366                                                 sequence = fasta->getSequence(name);
 367                                                 if (sequence != "not found") {
 368                                                         //if you don't have groups
 369                                                         if (groupfile == "") {
 370                                                                 name = name + "\t" + toString(i+1);
 371                                                                 out << ">" << name << endl;
 372                                                                 out << sequence << endl;
 373                                                         }else {//if you do have groups
 374                                                                 string group = groupMap->getGroup(name);
 375                                                                 if (group == "not found") {
 376                                                                         m->mothurOut(name + " is missing from your group file. Please correct. ");  m->mothurOutEndLine();
 377                                                                         return 1;
 378                                                                 }else{
 379                                                                         name = name + "\t" + group + "\t" + toString(i+1);
 380                                                                         out << ">" << name << endl;
 381                                                                         out << sequence << endl;
 382                                                                 }
 383                                                         }
 384                                                 }else {
 385                                                         m->mothurOut(name + " is missing from your fasta or name file. Please correct. "); m->mothurOutEndLine();
 386                                                         return 1;
 387                                                 }
 388
 389                                         }
 390
 391                                         //get last name
 392                                         sequence = fasta->getSequence(binnames);
 393                                         if (sequence != "not found") {
 394                                                 //if you don't have groups
 395                                                 if (groupfile == "") {
 396                                                         binnames = binnames + "\t" + toString(i+1);
 397                                                         out << ">" << binnames << endl;
 398                                                         out << sequence << endl;
 399                                                 }else {//if you do have groups
 400                                                         string group = groupMap->getGroup(binnames);
 401                                                         if (group == "not found") {
 402                                                                 m->mothurOut(binnames + " is missing from your group file. Please correct. "); m->mothurOutEndLine();
 403                                                                 return 1;
 404                                                         }else{
 405                                                                 binnames = binnames + "\t" + group + "\t" + toString(i+1);
 406                                                                 out << ">" << binnames << endl;
 407                                                                 out << sequence << endl;
 408                                                         }
 409                                                 }
 410                                         }else {
 411                                                 m->mothurOut(binnames + " is missing from your fasta or name file. Please correct. "); m->mothurOutEndLine();
 412                                                 return 1;
 413                                         }
 414                                 }
 415
 416                                 out.close();
 417                                 return 0;
 418
 419         }
 420         catch(exception& e) {
 421                 m->errorOut(e, "BinSeqCommand", "process");
 422                 exit(1);
 423         }
 424 }
 425 //**********************************************************************************************************************
 426
 427