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 BinSeqCommand::BinSeqCommand(string option){
  14         try {
  15                 globaldata = GlobalData::getInstance();
  16                 abort = false;
  17                 allLines = 1;
  18                 labels.clear();
  19
  20                 //allow user to run help
  21                 if(option == "help") { help(); abort = true; }
  22
  23                 else {
  24                         //valid paramters for this command
  25                         string AlignArray[] =  {"fasta","label","name", "group"};
  26                         vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  27
  28                         OptionParser parser(option);
  29                         map<string, string> parameters = parser.getParameters();
  30
  31                         ValidParameters validParameter;
  32
  33                         //check to make sure all parameters are valid for command
  34                         for (map<string, string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  35                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  36                         }
  37
  38                         //make sure the user has already run the read.otu command
  39                         if (globaldata->getListFile() == "") {
  40                                 mothurOut("You must read a listfile before running the bin.seqs command.");
  41                                 mothurOutEndLine();
  42                                 abort = true;
  43                         }
  44
  45
  46                         //check for required parameters
  47                         fastafile = validParameter.validFile(parameters, "fasta", true);
  48                         if (fastafile == "not found") { mothurOut("fasta is a required parameter for the bin.seqs command.");  mothurOutEndLine(); abort = true; }
  49                         else if (fastafile == "not open") { abort = true; }
  50
  51                         //check for optional parameter and set defaults
  52                         // ...at some point should added some additional type checking...
  53
  54                         label = validParameter.validFile(parameters, "label", false);
  55                         if (label == "not found") { label = ""; }
  56                         else {
  57                                 if(label != "all") {  splitAtDash(label, labels);  allLines = 0;  }
  58                                 else { allLines = 1;  }
  59                         }
  60
  61                         //if the user has not specified any labels use the ones from read.otu
  62                         if (label == "") {
  63                                 allLines = globaldata->allLines;
  64                                 labels = globaldata->labels;
  65                         }
  66
  67                         namesfile = validParameter.validFile(parameters, "name", true);
  68                         if (namesfile == "not open") { abort = true; }
  69                         else if (namesfile == "not found") { namesfile = ""; }
  70
  71                         groupfile = validParameter.validFile(parameters, "group", true);
  72                         if (groupfile == "not open") { abort = true; }
  73                         else if (groupfile == "not found") { groupfile = ""; }
  74
  75                         if (abort == false) {
  76 //                              openInputFile(fastafile, in);
  77                                 fasta = new FastaMap();
  78                                 if (groupfile != "") {
  79                                         groupMap = new GroupMap(groupfile);
  80                                         groupMap->readMap();
  81                                 }
  82                         }
  83
  84                 }
  85         }
  86         catch(exception& e) {
  87                 errorOut(e, "BinSeqCommand", "BinSeqCommand");
  88                 exit(1);
  89         }
  90 }
  91 //**********************************************************************************************************************
  92
  93 void BinSeqCommand::help(){
  94         try {
  95                 mothurOut("The bin.seqs command can only be executed after a successful read.otu command of a listfile.\n");
  96                 mothurOut("The bin.seqs command parameters are fasta, name, label and group.  The fasta parameter is required.\n");
  97                 mothurOut("The label parameter allows you to select what distance levels you would like a output files created for, and are separated by dashes.\n");
  98                 mothurOut("The bin.seqs command should be in the following format: bin.seqs(fasta=yourFastaFile, name=yourNamesFile, group=yourGroupFile, label=yourLabels).\n");
  99                 mothurOut("Example bin.seqs(fasta=amazon.fasta, group=amazon.groups, name=amazon.names).\n");
 100                 mothurOut("The default value for label is all lines in your inputfile.\n");
 101                 mothurOut("The bin.seqs command outputs a .fasta file for each distance you specify appending the OTU number to each name.\n");
 102                 mothurOut("If you provide a groupfile, then it also appends the sequences group to the name.\n");
 103                 mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n\n");
 104         }
 105         catch(exception& e) {
 106                 errorOut(e, "BinSeqCommand", "help");
 107                 exit(1);
 108         }
 109 }
 110
 111 //**********************************************************************************************************************
 112
 113 BinSeqCommand::~BinSeqCommand(){
 114         //made new in execute
 115         if (abort == false) {
 116                 delete input;  globaldata->ginput = NULL;
 117                 delete read;
 118                 globaldata->gListVector = NULL;
 119                 delete fasta;
 120                 if (groupfile != "") {  delete groupMap;  globaldata->gGroupmap = NULL; }
 121         }
 122 }
 123
 124 //**********************************************************************************************************************
 125
 126 int BinSeqCommand::execute(){
 127         try {
 128                 if (abort == true) {    return 0;       }
 129
 130                 int error = 0;
 131
 132                 //read fastafile
 133                 fasta->readFastaFile(fastafile);
 134
 135
 136                 //set format to list so input can get listvector
 137 //              globaldata->setFormat("list");
 138
 139                 //if user gave a namesfile then use it
 140                 if (namesfile != "") {
 141                         readNamesFile();
 142                 }
 143
 144                 //read list file
 145                 read = new ReadOTUFile(globaldata->getListFile());
 146                 read->read(&*globaldata);
 147
 148                 input = globaldata->ginput;
 149                 list = globaldata->gListVector;
 150                 string lastLabel = list->getLabel();
 151
 152                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 153                 set<string> processedLabels;
 154                 set<string> userLabels = labels;
 155
 156
 157                 while((list != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 158
 159                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
 160
 161                                 error = process(list);
 162                                 if (error == 1) { return 0; }
 163
 164                                 processedLabels.insert(list->getLabel());
 165                                 userLabels.erase(list->getLabel());
 166                         }
 167
 168                         if ((anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 169                                 delete list;
 170                                 list = input->getListVector(lastLabel);
 171
 172                                 error = process(list);
 173                                 if (error == 1) { return 0; }
 174
 175                                 processedLabels.insert(list->getLabel());
 176                                 userLabels.erase(list->getLabel());
 177                         }
 178
 179                         lastLabel = list->getLabel();
 180
 181                         delete list;
 182                         list = input->getListVector();
 183                 }
 184
 185
 186                 //output error messages about any remaining user labels
 187                 set<string>::iterator it;
 188                 bool needToRun = false;
 189                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 190                         mothurOut("Your file does not include the label " + *it);
 191                         if (processedLabels.count(lastLabel) != 1) {
 192                                 mothurOut(". I will use " + lastLabel + "."); mothurOutEndLine();
 193                                 needToRun = true;
 194                         }else {
 195                                 mothurOut(". Please refer to " + lastLabel + ".");  mothurOutEndLine();
 196                         }
 197                 }
 198
 199                 //run last label if you need to
 200                 if (needToRun == true)  {
 201                         if (list != NULL) {             delete list;    }
 202                         list = input->getListVector(lastLabel);
 203
 204                         error = process(list);
 205                         if (error == 1) { return 0; }
 206
 207                         delete list;
 208                 }
 209
 210                 return 0;
 211         }
 212         catch(exception& e) {
 213                 errorOut(e, "BinSeqCommand", "execute");
 214                 exit(1);
 215         }
 216 }
 217
 218 //**********************************************************************************************************************
 219 void BinSeqCommand::readNamesFile() {
 220         try {
 221                 vector<string> dupNames;
 222                 openInputFile(namesfile, inNames);
 223
 224                 string name, names, sequence;
 225
 226                 while(inNames){
 227                         inNames >> name;                        //read from first column  A
 228                         inNames >> names;               //read from second column  A,B,C,D
 229
 230                         dupNames.clear();
 231
 232                         //parse names into vector
 233                         splitAtComma(names, dupNames);
 234
 235                         //store names in fasta map
 236                         sequence = fasta->getSequence(name);
 237                         for (int i = 0; i < dupNames.size(); i++) {
 238                                 fasta->push_back(dupNames[i], sequence);
 239                         }
 240
 241                         gobble(inNames);
 242                 }
 243                 inNames.close();
 244
 245         }
 246         catch(exception& e) {
 247                 errorOut(e, "BinSeqCommand", "readNamesFile");
 248                 exit(1);
 249         }
 250 }
 251 //**********************************************************************************************************************
 252 //return 1 if error, 0 otherwise
 253 int BinSeqCommand::process(ListVector* list) {
 254         try {
 255                                 string binnames, name, sequence;
 256                                 string outputFileName = getRootName(globaldata->getListFile()) + list->getLabel() + ".fasta";
 257                                 openOutputFile(outputFileName, out);
 258
 259                                 mothurOut(list->getLabel()); mothurOutEndLine();
 260
 261                                 //for each bin in the list vector
 262                                 for (int i = 0; i < list->size(); i++) {
 263
 264                                         binnames = list->get(i);
 265                                         while (binnames.find_first_of(',') != -1) {
 266                                                 name = binnames.substr(0,binnames.find_first_of(','));
 267                                                 binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 268
 269                                                 //do work for that name
 270                                                 sequence = fasta->getSequence(name);
 271                                                 if (sequence != "not found") {
 272                                                         //if you don't have groups
 273                                                         if (groupfile == "") {
 274                                                                 name = name + "|" + toString(i+1);
 275                                                                 out << ">" << name << endl;
 276                                                                 out << sequence << endl;
 277                                                         }else {//if you do have groups
 278                                                                 string group = groupMap->getGroup(name);
 279                                                                 if (group == "not found") {
 280                                                                         mothurOut(name + " is missing from your group file. Please correct. ");  mothurOutEndLine();
 281                                                                         remove(outputFileName.c_str());
 282                                                                         return 1;
 283                                                                 }else{
 284                                                                         name = name + "|" + group + "|" + toString(i+1);
 285                                                                         out << ">" << name << endl;
 286                                                                         out << sequence << endl;
 287                                                                 }
 288                                                         }
 289                                                 }else {
 290                                                         mothurOut(name + " is missing from your fasta or name file. Please correct. "); mothurOutEndLine();
 291                                                         remove(outputFileName.c_str());
 292                                                         return 1;
 293                                                 }
 294
 295                                         }
 296
 297                                         //get last name
 298                                         sequence = fasta->getSequence(binnames);
 299                                         if (sequence != "not found") {
 300                                                 //if you don't have groups
 301                                                 if (groupfile == "") {
 302                                                         binnames = binnames + "|" + toString(i+1);
 303                                                         out << ">" << binnames << endl;
 304                                                         out << sequence << endl;
 305                                                 }else {//if you do have groups
 306                                                         string group = groupMap->getGroup(binnames);
 307                                                         if (group == "not found") {
 308                                                                 mothurOut(binnames + " is missing from your group file. Please correct. "); mothurOutEndLine();
 309                                                                 remove(outputFileName.c_str());
 310                                                                 return 1;
 311                                                         }else{
 312                                                                 binnames = binnames + "|" + group + "|" + toString(i+1);
 313                                                                 out << ">" << binnames << endl;
 314                                                                 out << sequence << endl;
 315                                                         }
 316                                                 }
 317                                         }else {
 318                                                 mothurOut(binnames + " is missing from your fasta or name file. Please correct. "); mothurOutEndLine();
 319                                                 remove(outputFileName.c_str());
 320                                                 return 1;
 321                                         }
 322                                 }
 323
 324                                 out.close();
 325                                 return 0;
 326
 327         }
 328         catch(exception& e) {
 329                 errorOut(e, "BinSeqCommand", "process");
 330                 exit(1);
 331         }
 332 }
 333 //**********************************************************************************************************************
 334
 335