listseqscommand.cpp

   1 /*
   2  *  listseqscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 7/8/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "listseqscommand.h"
  11 #include "sequence.hpp"
  12 #include "listvector.hpp"
  13
  14 //**********************************************************************************************************************
  15
  16 ListSeqsCommand::ListSeqsCommand(string option)  {
  17         try {
  18                 abort = false;
  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 Array[] =  {"fasta","name", "group", "alignreport","list","taxonomy","outputdir","inputdir"};
  26                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  27
  28                         OptionParser parser(option);
  29                         map<string,string> parameters = parser.getParameters();
  30
  31                         ValidParameters validParameter;
  32                         map<string,string>::iterator it;
  33
  34                         //check to make sure all parameters are valid for command
  35                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  36                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  37                         }
  38
  39                         //if the user changes the output directory command factory will send this info to us in the output parameter
  40                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
  41
  42                         //if the user changes the input directory command factory will send this info to us in the output parameter
  43                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  44                         if (inputDir == "not found"){   inputDir = "";          }
  45                         else {
  46                                 string path;
  47                                 it = parameters.find("alignreport");
  48                                 //user has given a template file
  49                                 if(it != parameters.end()){
  50                                         path = m->hasPath(it->second);
  51                                         //if the user has not given a path then, add inputdir. else leave path alone.
  52                                         if (path == "") {       parameters["alignreport"] = inputDir + it->second;              }
  53                                 }
  54
  55                                 it = parameters.find("fasta");
  56                                 //user has given a template file
  57                                 if(it != parameters.end()){
  58                                         path = m->hasPath(it->second);
  59                                         //if the user has not given a path then, add inputdir. else leave path alone.
  60                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
  61                                 }
  62
  63                                 it = parameters.find("list");
  64                                 //user has given a template file
  65                                 if(it != parameters.end()){
  66                                         path = m->hasPath(it->second);
  67                                         //if the user has not given a path then, add inputdir. else leave path alone.
  68                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
  69                                 }
  70
  71                                 it = parameters.find("name");
  72                                 //user has given a template file
  73                                 if(it != parameters.end()){
  74                                         path = m->hasPath(it->second);
  75                                         //if the user has not given a path then, add inputdir. else leave path alone.
  76                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
  77                                 }
  78
  79                                 it = parameters.find("group");
  80                                 //user has given a template file
  81                                 if(it != parameters.end()){
  82                                         path = m->hasPath(it->second);
  83                                         //if the user has not given a path then, add inputdir. else leave path alone.
  84                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
  85                                 }
  86
  87                                 it = parameters.find("taxonomy");
  88                                 //user has given a template file
  89                                 if(it != parameters.end()){
  90                                         path = m->hasPath(it->second);
  91                                         //if the user has not given a path then, add inputdir. else leave path alone.
  92                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
  93                                 }
  94                         }
  95
  96                         //check for required parameters
  97                         fastafile = validParameter.validFile(parameters, "fasta", true);
  98                         if (fastafile == "not open") { abort = true; }
  99                         else if (fastafile == "not found") {  fastafile = "";  }
 100
 101                         namefile = validParameter.validFile(parameters, "name", true);
 102                         if (namefile == "not open") { abort = true; }
 103                         else if (namefile == "not found") {  namefile = "";  }
 104
 105                         groupfile = validParameter.validFile(parameters, "group", true);
 106                         if (groupfile == "not open") { abort = true; }
 107                         else if (groupfile == "not found") {  groupfile = "";  }
 108
 109                         alignfile = validParameter.validFile(parameters, "alignreport", true);
 110                         if (alignfile == "not open") { abort = true; }
 111                         else if (alignfile == "not found") {  alignfile = "";  }
 112
 113                         listfile = validParameter.validFile(parameters, "list", true);
 114                         if (listfile == "not open") { abort = true; }
 115                         else if (listfile == "not found") {  listfile = "";  }
 116
 117                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 118                         if (taxfile == "not open") { abort = true; }
 119                         else if (taxfile == "not found") {  taxfile = "";  }
 120
 121                         if ((fastafile == "") && (namefile == "") && (listfile == "") && (groupfile == "") && (alignfile == "") && (taxfile == ""))  { m->mothurOut("You must provide a file."); m->mothurOutEndLine(); abort = true; }
 122
 123                         int okay = 1;
 124                         if (outputDir != "") { okay++; }
 125                         if (inputDir != "") { okay++; }
 126
 127                         if (parameters.size() > okay) { m->mothurOut("You may only enter one file."); m->mothurOutEndLine(); abort = true;  }
 128                 }
 129
 130         }
 131         catch(exception& e) {
 132                 m->errorOut(e, "ListSeqsCommand", "ListSeqsCommand");
 133                 exit(1);
 134         }
 135 }
 136 //**********************************************************************************************************************
 137
 138 void ListSeqsCommand::help(){
 139         try {
 140                 m->mothurOut("The list.seqs command reads a fasta, name, group, list, taxonomy or alignreport file and outputs a .accnos file containing sequence names.\n");
 141                 m->mothurOut("The list.seqs command parameters are fasta, name, group, list, taxonomy and alignreport.  You must provide one of these parameters.\n");
 142                 m->mothurOut("The list.seqs command should be in the following format: list.seqs(fasta=yourFasta).\n");
 143                 m->mothurOut("Example list.seqs(fasta=amazon.fasta).\n");
 144                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
 145         }
 146         catch(exception& e) {
 147                 m->errorOut(e, "ListSeqsCommand", "help");
 148                 exit(1);
 149         }
 150 }
 151
 152 //**********************************************************************************************************************
 153
 154 int ListSeqsCommand::execute(){
 155         try {
 156
 157                 if (abort == true) { return 0; }
 158
 159                 //read functions fill names vector
 160                 if (fastafile != "")            {       inputFileName = fastafile;      readFasta();    }
 161                 else if (namefile != "")        {       inputFileName = namefile;       readName();             }
 162                 else if (groupfile != "")       {       inputFileName = groupfile;      readGroup();    }
 163                 else if (alignfile != "")       {       inputFileName = alignfile;      readAlign();    }
 164                 else if (listfile != "")        {       inputFileName = listfile;       readList();             }
 165                 else if (taxfile != "")         {       inputFileName = taxfile;        readTax();              }
 166
 167                 if (m->control_pressed) { return 0; }
 168
 169                 //sort in alphabetical order
 170                 sort(names.begin(), names.end());
 171
 172                 if (outputDir == "") {  outputDir += m->hasPath(inputFileName);  }
 173
 174                 string outputFileName = outputDir + m->getRootName(m->getSimpleName(inputFileName)) + "accnos";
 175
 176                 ofstream out;
 177                 m->openOutputFile(outputFileName, out);
 178
 179                 //output to .accnos file
 180                 for (int i = 0; i < names.size(); i++) {
 181
 182                         if (m->control_pressed) { out.close(); remove(outputFileName.c_str()); return 0; }
 183
 184                         out << names[i] << endl;
 185                 }
 186                 out.close();
 187
 188                 if (m->control_pressed) { remove(outputFileName.c_str()); return 0; }
 189
 190                 m->mothurOutEndLine();
 191                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 192                 m->mothurOut(outputFileName); m->mothurOutEndLine();
 193                 m->mothurOutEndLine();
 194
 195                 return 0;
 196         }
 197
 198         catch(exception& e) {
 199                 m->errorOut(e, "ListSeqsCommand", "execute");
 200                 exit(1);
 201         }
 202 }
 203
 204 //**********************************************************************************************************************
 205 int ListSeqsCommand::readFasta(){
 206         try {
 207
 208                 ifstream in;
 209                 m->openInputFile(fastafile, in);
 210                 string name;
 211
 212                 while(!in.eof()){
 213
 214                         if (m->control_pressed) { in.close(); return 0; }
 215
 216                         Sequence currSeq(in);
 217                         name = currSeq.getName();
 218
 219                         if (name != "") {  names.push_back(name);  }
 220
 221                         m->gobble(in);
 222                 }
 223                 in.close();
 224
 225                 return 0;
 226
 227         }
 228         catch(exception& e) {
 229                 m->errorOut(e, "ListSeqsCommand", "readFasta");
 230                 exit(1);
 231         }
 232 }
 233 //**********************************************************************************************************************
 234 int ListSeqsCommand::readList(){
 235         try {
 236                 ifstream in;
 237                 m->openInputFile(listfile, in);
 238
 239                 if(!in.eof()){
 240                         //read in list vector
 241                         ListVector list(in);
 242
 243                         //for each bin
 244                         for (int i = 0; i < list.getNumBins(); i++) {
 245                                 string binnames = list.get(i);
 246
 247                                 if (m->control_pressed) { in.close(); return 0; }
 248
 249                                 while (binnames.find_first_of(',') != -1) {
 250                                         string name = binnames.substr(0,binnames.find_first_of(','));
 251                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 252                                         names.push_back(name);
 253                                 }
 254
 255                                 names.push_back(binnames);
 256                         }
 257                 }
 258                 in.close();
 259
 260                 return 0;
 261
 262         }
 263         catch(exception& e) {
 264                 m->errorOut(e, "ListSeqsCommand", "readList");
 265                 exit(1);
 266         }
 267 }
 268
 269 //**********************************************************************************************************************
 270 int ListSeqsCommand::readName(){
 271         try {
 272
 273                 ifstream in;
 274                 m->openInputFile(namefile, in);
 275                 string name, firstCol, secondCol;
 276
 277                 while(!in.eof()){
 278
 279                         if (m->control_pressed) { in.close(); return 0; }
 280
 281                         in >> firstCol;
 282                         in >> secondCol;
 283
 284                         //parse second column saving each name
 285                         while (secondCol.find_first_of(',') != -1) {
 286                                 name = secondCol.substr(0,secondCol.find_first_of(','));
 287                                 secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
 288                                 names.push_back(name);
 289                         }
 290
 291                         //get name after last ,
 292                         names.push_back(secondCol);
 293
 294                         m->gobble(in);
 295                 }
 296                 in.close();
 297                 return 0;
 298
 299         }
 300         catch(exception& e) {
 301                 m->errorOut(e, "ListSeqsCommand", "readName");
 302                 exit(1);
 303         }
 304 }
 305
 306 //**********************************************************************************************************************
 307 int ListSeqsCommand::readGroup(){
 308         try {
 309
 310                 ifstream in;
 311                 m->openInputFile(groupfile, in);
 312                 string name, group;
 313
 314                 while(!in.eof()){
 315
 316                         if (m->control_pressed) { in.close(); return 0; }
 317
 318                         in >> name;                             //read from first column
 319                         in >> group;                    //read from second column
 320
 321                         names.push_back(name);
 322
 323                         m->gobble(in);
 324                 }
 325                 in.close();
 326                 return 0;
 327
 328         }
 329         catch(exception& e) {
 330                 m->errorOut(e, "ListSeqsCommand", "readGroup");
 331                 exit(1);
 332         }
 333 }
 334
 335 //**********************************************************************************************************************
 336 //alignreport file has a column header line then all other lines contain 16 columns.  we just want the first column since that contains the name
 337 int ListSeqsCommand::readAlign(){
 338         try {
 339
 340                 ifstream in;
 341                 m->openInputFile(alignfile, in);
 342                 string name, junk;
 343
 344                 //read column headers
 345                 for (int i = 0; i < 16; i++) {
 346                         if (!in.eof())  {       in >> junk;             }
 347                         else                    {       break;                  }
 348                 }
 349
 350
 351                 while(!in.eof()){
 352
 353                         if (m->control_pressed) { in.close(); return 0; }
 354
 355                         in >> name;                             //read from first column
 356
 357                         //read rest
 358                         for (int i = 0; i < 15; i++) {
 359                                 if (!in.eof())  {       in >> junk;             }
 360                                 else                    {       break;                  }
 361                         }
 362
 363                         names.push_back(name);
 364
 365                         m->gobble(in);
 366                 }
 367                 in.close();
 368
 369                 return 0;
 370
 371
 372         }
 373         catch(exception& e) {
 374                 m->errorOut(e, "ListSeqsCommand", "readAlign");
 375                 exit(1);
 376         }
 377 }
 378 //**********************************************************************************************************************
 379 int ListSeqsCommand::readTax(){
 380         try {
 381
 382                 ifstream in;
 383                 m->openInputFile(taxfile, in);
 384                 string name, firstCol, secondCol;
 385
 386                 while(!in.eof()){
 387
 388                         if (m->control_pressed) { in.close(); return 0; }
 389
 390                         in >> firstCol;
 391                         in >> secondCol;
 392
 393                         names.push_back(firstCol);
 394
 395                         m->gobble(in);
 396                 }
 397                 in.close();
 398
 399                 return 0;
 400
 401         }
 402         catch(exception& e) {
 403                 m->errorOut(e, "ListSeqsCommand", "readTax");
 404                 exit(1);
 405         }
 406 }
 407 //**********************************************************************************************************************