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\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
  75                 else {
  76                         vector<string> myArray = setParameters();
  77
  78                         OptionParser parser(option);
  79                         map<string, string> parameters = parser.getParameters();
  80
  81                         ValidParameters validParameter;
  82                         map<string, string>::iterator it;
  83
  84                         //check to make sure all parameters are valid for command
  85                         for (it = parameters.begin(); it != parameters.end(); it++) {
  86                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  87                         }
  88
  89                         //initialize outputTypes
  90                         vector<string> tempOutNames;
  91                         outputTypes["fasta"] = tempOutNames;
  92
  93                         //if the user changes the input directory command factory will send this info to us in the output parameter
  94                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  95                         if (inputDir == "not found"){   inputDir = "";          }
  96                         else {
  97                                 string path;
  98                                 it = parameters.find("fasta");
  99                                 //user has given a template file
 100                                 if(it != parameters.end()){
 101                                         path = m->hasPath(it->second);
 102                                         //if the user has not given a path then, add inputdir. else leave path alone.
 103                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 104                                 }
 105
 106                                 it = parameters.find("list");
 107                                 //user has given a template file
 108                                 if(it != parameters.end()){
 109                                         path = m->hasPath(it->second);
 110                                         //if the user has not given a path then, add inputdir. else leave path alone.
 111                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 112                                 }
 113
 114                                 it = parameters.find("name");
 115                                 //user has given a template file
 116                                 if(it != parameters.end()){
 117                                         path = m->hasPath(it->second);
 118                                         //if the user has not given a path then, add inputdir. else leave path alone.
 119                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 120                                 }
 121
 122                                 it = parameters.find("group");
 123                                 //user has given a template file
 124                                 if(it != parameters.end()){
 125                                         path = m->hasPath(it->second);
 126                                         //if the user has not given a path then, add inputdir. else leave path alone.
 127                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 128                                 }
 129                         }
 130
 131
 132                         //check for required parameters
 133                         fastafile = validParameter.validFile(parameters, "fasta", true);
 134                         if (fastafile == "not found") {                                 //if there is a current phylip file, use it
 135                                 fastafile = m->getFastaFile();
 136                                 if (fastafile != "") { m->mothurOut("Using " + fastafile + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 137                                 else {  m->mothurOut("You have no current fasta file and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 138                         }
 139                         else if (fastafile == "not open") { abort = true; }
 140
 141                         listfile = validParameter.validFile(parameters, "list", true);
 142                         if (listfile == "not found") {
 143                                 listfile = m->getListFile();
 144                                 if (listfile != "") { m->mothurOut("Using " + listfile + " as input file for the list parameter."); m->mothurOutEndLine(); }
 145                                 else {  m->mothurOut("You have no current list file and the list parameter is required."); m->mothurOutEndLine(); abort = true; }
 146                         }
 147                         else if (listfile == "not open") { listfile = ""; abort = true; }
 148
 149                         //if the user changes the output directory command factory will send this info to us in the output parameter
 150                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 151                                 outputDir = "";
 152                                 outputDir += m->hasPath(listfile); //if user entered a file with a path then preserve it
 153                         }
 154
 155
 156                         //check for optional parameter and set defaults
 157                         // ...at some point should added some additional type checking...
 158
 159                         label = validParameter.validFile(parameters, "label", false);
 160                         if (label == "not found") { label = ""; }
 161                         else {
 162                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 163                                 else { allLines = 1;  }
 164                         }
 165
 166                         namesfile = validParameter.validFile(parameters, "name", true);
 167                         if (namesfile == "not open") { abort = true; }
 168                         else if (namesfile == "not found") { namesfile = ""; }
 169
 170                         groupfile = validParameter.validFile(parameters, "group", true);
 171                         if (groupfile == "not open") { abort = true; }
 172                         else if (groupfile == "not found") { groupfile = ""; }
 173
 174                 }
 175         }
 176         catch(exception& e) {
 177                 m->errorOut(e, "BinSeqCommand", "BinSeqCommand");
 178                 exit(1);
 179         }
 180 }
 181 //**********************************************************************************************************************
 182
 183 BinSeqCommand::~BinSeqCommand(){}
 184 //**********************************************************************************************************************
 185
 186 int BinSeqCommand::execute(){
 187         try {
 188                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 189
 190                 int error = 0;
 191
 192                 fasta = new FastaMap();
 193                 if (groupfile != "") {
 194                         groupMap = new GroupMap(groupfile);
 195                         groupMap->readMap();
 196                 }
 197
 198                 //read fastafile
 199                 fasta->readFastaFile(fastafile);
 200
 201                 //if user gave a namesfile then use it
 202                 if (namesfile != "") {
 203                         readNamesFile();
 204                 }
 205
 206                 input = new InputData(listfile, "list");
 207                 list = input->getListVector();
 208                 string lastLabel = list->getLabel();
 209
 210                 if (m->control_pressed) {  delete input;  delete fasta; if (groupfile != "") {  delete groupMap;   } return 0; }
 211
 212                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 213                 set<string> processedLabels;
 214                 set<string> userLabels = labels;
 215
 216
 217                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 218
 219                         if(m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str());         } delete input;  delete fasta; if (groupfile != "") {  delete groupMap;   } return 0; }
 220
 221                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
 222
 223                                 error = process(list);
 224                                 if (error == 1) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());         } delete input;  delete fasta; if (groupfile != "") {  delete groupMap;   } return 0; }
 225
 226                                 processedLabels.insert(list->getLabel());
 227                                 userLabels.erase(list->getLabel());
 228                         }
 229
 230                         if ((m->anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 231                                 string saveLabel = list->getLabel();
 232
 233                                 delete list;
 234                                 list = input->getListVector(lastLabel);
 235
 236                                 error = process(list);
 237                                 if (error == 1) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());         } delete input;  delete fasta; if (groupfile != "") {  delete groupMap;   } return 0; }
 238
 239                                 processedLabels.insert(list->getLabel());
 240                                 userLabels.erase(list->getLabel());
 241
 242                                 //restore real lastlabel to save below
 243                                 list->setLabel(saveLabel);
 244                         }
 245
 246                         lastLabel = list->getLabel();
 247
 248                         delete list;
 249                         list = input->getListVector();
 250                 }
 251
 252                 if(m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str());         } delete input;  delete fasta; if (groupfile != "") {  delete groupMap;   } return 0; }
 253
 254                 //output error messages about any remaining user labels
 255                 set<string>::iterator it;
 256                 bool needToRun = false;
 257                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 258                         m->mothurOut("Your file does not include the label " + *it);
 259                         if (processedLabels.count(lastLabel) != 1) {
 260                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 261                                 needToRun = true;
 262                         }else {
 263                                 m->mothurOut(". Please refer to " + lastLabel + ".");  m->mothurOutEndLine();
 264                         }
 265                 }
 266
 267                 //run last label if you need to
 268                 if (needToRun == true)  {
 269                         if (list != NULL) {             delete list;    }
 270                         list = input->getListVector(lastLabel);
 271
 272                         error = process(list);
 273                         if (error == 1) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());         } delete input;  delete fasta; if (groupfile != "") {  delete groupMap;   } return 0; }
 274
 275                         delete list;
 276                 }
 277
 278                 delete input;
 279                 delete fasta;
 280                 if (groupfile != "") {  delete groupMap;   }
 281
 282                 if(m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str());         }  return 0; }
 283
 284                 delete input;
 285                 delete fasta;
 286                 if (groupfile != "") {  delete groupMap;   }
 287
 288                 m->mothurOutEndLine();
 289                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 290                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 291                 m->mothurOutEndLine();
 292
 293
 294                 return 0;
 295         }
 296         catch(exception& e) {
 297                 m->errorOut(e, "BinSeqCommand", "execute");
 298                 exit(1);
 299         }
 300 }
 301
 302 //**********************************************************************************************************************
 303 void BinSeqCommand::readNamesFile() {
 304         try {
 305                 vector<string> dupNames;
 306                 m->openInputFile(namesfile, inNames);
 307
 308                 string name, names, sequence;
 309
 310                 while(inNames){
 311                         inNames >> name;                        //read from first column  A
 312                         inNames >> names;               //read from second column  A,B,C,D
 313
 314                         dupNames.clear();
 315
 316                         //parse names into vector
 317                         m->splitAtComma(names, dupNames);
 318
 319                         //store names in fasta map
 320                         sequence = fasta->getSequence(name);
 321                         for (int i = 0; i < dupNames.size(); i++) {
 322                                 fasta->push_back(dupNames[i], sequence);
 323                         }
 324
 325                         m->gobble(inNames);
 326                 }
 327                 inNames.close();
 328
 329         }
 330         catch(exception& e) {
 331                 m->errorOut(e, "BinSeqCommand", "readNamesFile");
 332                 exit(1);
 333         }
 334 }
 335 //**********************************************************************************************************************
 336 //return 1 if error, 0 otherwise
 337 int BinSeqCommand::process(ListVector* list) {
 338         try {
 339                                 string binnames, name, sequence;
 340
 341                                 string outputFileName = outputDir + m->getRootName(m->getSimpleName(listfile)) + list->getLabel() + ".fasta";
 342                                 m->openOutputFile(outputFileName, out);
 343
 344                                 //save to output list of output file names
 345                                 outputNames.push_back(outputFileName);  outputTypes["fasta"].push_back(outputFileName);
 346
 347                                 m->mothurOut(list->getLabel()); m->mothurOutEndLine();
 348
 349                                 //for each bin in the list vector
 350                                 for (int i = 0; i < list->size(); i++) {
 351
 352                                         if (m->control_pressed) {  return 1; }
 353
 354                                         binnames = list->get(i);
 355                                         while (binnames.find_first_of(',') != -1) {
 356                                                 name = binnames.substr(0,binnames.find_first_of(','));
 357                                                 binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 358
 359                                                 //do work for that name
 360                                                 sequence = fasta->getSequence(name);
 361                                                 if (sequence != "not found") {
 362                                                         //if you don't have groups
 363                                                         if (groupfile == "") {
 364                                                                 name = name + "\t" + toString(i+1);
 365                                                                 out << ">" << name << endl;
 366                                                                 out << sequence << endl;
 367                                                         }else {//if you do have groups
 368                                                                 string group = groupMap->getGroup(name);
 369                                                                 if (group == "not found") {
 370                                                                         m->mothurOut(name + " is missing from your group file. Please correct. ");  m->mothurOutEndLine();
 371                                                                         return 1;
 372                                                                 }else{
 373                                                                         name = name + "\t" + group + "\t" + toString(i+1);
 374                                                                         out << ">" << name << endl;
 375                                                                         out << sequence << endl;
 376                                                                 }
 377                                                         }
 378                                                 }else {
 379                                                         m->mothurOut(name + " is missing from your fasta or name file. Please correct. "); m->mothurOutEndLine();
 380                                                         return 1;
 381                                                 }
 382
 383                                         }
 384
 385                                         //get last name
 386                                         sequence = fasta->getSequence(binnames);
 387                                         if (sequence != "not found") {
 388                                                 //if you don't have groups
 389                                                 if (groupfile == "") {
 390                                                         binnames = binnames + "\t" + toString(i+1);
 391                                                         out << ">" << binnames << endl;
 392                                                         out << sequence << endl;
 393                                                 }else {//if you do have groups
 394                                                         string group = groupMap->getGroup(binnames);
 395                                                         if (group == "not found") {
 396                                                                 m->mothurOut(binnames + " is missing from your group file. Please correct. "); m->mothurOutEndLine();
 397                                                                 return 1;
 398                                                         }else{
 399                                                                 binnames = binnames + "\t" + group + "\t" + toString(i+1);
 400                                                                 out << ">" << binnames << endl;
 401                                                                 out << sequence << endl;
 402                                                         }
 403                                                 }
 404                                         }else {
 405                                                 m->mothurOut(binnames + " is missing from your fasta or name file. Please correct. "); m->mothurOutEndLine();
 406                                                 return 1;
 407                                         }
 408                                 }
 409
 410                                 out.close();
 411                                 return 0;
 412
 413         }
 414         catch(exception& e) {
 415                 m->errorOut(e, "BinSeqCommand", "process");
 416                 exit(1);
 417         }
 418 }
 419 //**********************************************************************************************************************
 420
 421