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","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 = 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 = 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 = 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 = 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 = 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
  88                         //check for required parameters
  89                         fastafile = validParameter.validFile(parameters, "fasta", true);
  90                         if (fastafile == "not open") { abort = true; }
  91                         else if (fastafile == "not found") {  fastafile = "";  }
  92
  93                         namefile = validParameter.validFile(parameters, "name", true);
  94                         if (namefile == "not open") { abort = true; }
  95                         else if (namefile == "not found") {  namefile = "";  }
  96
  97                         groupfile = validParameter.validFile(parameters, "group", true);
  98                         if (groupfile == "not open") { abort = true; }
  99                         else if (groupfile == "not found") {  groupfile = "";  }
 100
 101                         alignfile = validParameter.validFile(parameters, "alignreport", true);
 102                         if (alignfile == "not open") { abort = true; }
 103                         else if (alignfile == "not found") {  alignfile = "";  }
 104
 105                         listfile = validParameter.validFile(parameters, "list", true);
 106                         if (listfile == "not open") { abort = true; }
 107                         else if (listfile == "not found") {  listfile = "";  }
 108
 109
 110                         if ((fastafile == "") && (namefile == "") && (listfile == "") && (groupfile == "") && (alignfile == ""))  { m->mothurOut("You must provide a file."); m->mothurOutEndLine(); abort = true; }
 111
 112                         int okay = 1;
 113                         if (outputDir != "") { okay++; }
 114                         if (inputDir != "") { okay++; }
 115
 116                         if (parameters.size() > okay) { m->mothurOut("You may only enter one file."); m->mothurOutEndLine(); abort = true;  }
 117                 }
 118
 119         }
 120         catch(exception& e) {
 121                 m->errorOut(e, "ListSeqsCommand", "ListSeqsCommand");
 122                 exit(1);
 123         }
 124 }
 125 //**********************************************************************************************************************
 126
 127 void ListSeqsCommand::help(){
 128         try {
 129                 m->mothurOut("The list.seqs command reads a fasta, name, group, list or alignreport file and outputs a .accnos file containing sequence names.\n");
 130                 m->mothurOut("The list.seqs command parameters are fasta, name, group and alignreport.  You must provide one of these parameters.\n");
 131                 m->mothurOut("The list.seqs command should be in the following format: list.seqs(fasta=yourFasta).\n");
 132                 m->mothurOut("Example list.seqs(fasta=amazon.fasta).\n");
 133                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
 134         }
 135         catch(exception& e) {
 136                 m->errorOut(e, "ListSeqsCommand", "help");
 137                 exit(1);
 138         }
 139 }
 140
 141 //**********************************************************************************************************************
 142
 143 int ListSeqsCommand::execute(){
 144         try {
 145
 146                 if (abort == true) { return 0; }
 147
 148                 //read functions fill names vector
 149                 if (fastafile != "")            {       inputFileName = fastafile;      readFasta();    }
 150                 else if (namefile != "")        {       inputFileName = namefile;       readName();             }
 151                 else if (groupfile != "")       {       inputFileName = groupfile;      readGroup();    }
 152                 else if (alignfile != "")       {       inputFileName = alignfile;      readAlign();    }
 153                 else if (listfile != "")        {       inputFileName = listfile;       readList();             }
 154
 155                 if (m->control_pressed) { return 0; }
 156
 157                 //sort in alphabetical order
 158                 sort(names.begin(), names.end());
 159
 160                 if (outputDir == "") {  outputDir += hasPath(inputFileName);  }
 161
 162                 string outputFileName = outputDir + getRootName(getSimpleName(inputFileName)) + "accnos";
 163
 164                 ofstream out;
 165                 openOutputFile(outputFileName, out);
 166
 167                 //output to .accnos file
 168                 for (int i = 0; i < names.size(); i++) {
 169
 170                         if (m->control_pressed) { out.close(); remove(outputFileName.c_str()); return 0; }
 171
 172                         out << names[i] << endl;
 173                 }
 174                 out.close();
 175
 176                 if (m->control_pressed) { remove(outputFileName.c_str()); return 0; }
 177
 178                 m->mothurOutEndLine();
 179                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 180                 m->mothurOut(outputFileName); m->mothurOutEndLine();
 181                 m->mothurOutEndLine();
 182
 183                 return 0;
 184         }
 185
 186         catch(exception& e) {
 187                 m->errorOut(e, "ListSeqsCommand", "execute");
 188                 exit(1);
 189         }
 190 }
 191
 192 //**********************************************************************************************************************
 193 int ListSeqsCommand::readFasta(){
 194         try {
 195
 196                 ifstream in;
 197                 openInputFile(fastafile, in);
 198                 string name;
 199
 200                 while(!in.eof()){
 201
 202                         if (m->control_pressed) { in.close(); return 0; }
 203
 204                         Sequence currSeq(in);
 205                         name = currSeq.getName();
 206
 207                         if (name != "") {  names.push_back(name);  }
 208
 209                         gobble(in);
 210                 }
 211                 in.close();
 212
 213                 return 0;
 214
 215         }
 216         catch(exception& e) {
 217                 m->errorOut(e, "ListSeqsCommand", "readFasta");
 218                 exit(1);
 219         }
 220 }
 221 //**********************************************************************************************************************
 222 int ListSeqsCommand::readList(){
 223         try {
 224                 ifstream in;
 225                 openInputFile(listfile, in);
 226
 227                 if(!in.eof()){
 228                         //read in list vector
 229                         ListVector list(in);
 230
 231                         //for each bin
 232                         for (int i = 0; i < list.getNumBins(); i++) {
 233                                 string binnames = list.get(i);
 234
 235                                 if (m->control_pressed) { in.close(); return 0; }
 236
 237                                 while (binnames.find_first_of(',') != -1) {
 238                                         string name = binnames.substr(0,binnames.find_first_of(','));
 239                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 240                                         names.push_back(name);
 241                                 }
 242
 243                                 names.push_back(binnames);
 244                         }
 245                 }
 246                 in.close();
 247
 248                 return 0;
 249
 250         }
 251         catch(exception& e) {
 252                 m->errorOut(e, "ListSeqsCommand", "readList");
 253                 exit(1);
 254         }
 255 }
 256
 257 //**********************************************************************************************************************
 258 int ListSeqsCommand::readName(){
 259         try {
 260
 261                 ifstream in;
 262                 openInputFile(namefile, in);
 263                 string name, firstCol, secondCol;
 264
 265                 while(!in.eof()){
 266
 267                         if (m->control_pressed) { in.close(); return 0; }
 268
 269                         in >> firstCol;
 270                         in >> secondCol;
 271
 272                         //parse second column saving each name
 273                         while (secondCol.find_first_of(',') != -1) {
 274                                 name = secondCol.substr(0,secondCol.find_first_of(','));
 275                                 secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
 276                                 names.push_back(name);
 277                         }
 278
 279                         //get name after last ,
 280                         names.push_back(secondCol);
 281
 282                         gobble(in);
 283                 }
 284                 in.close();
 285                 return 0;
 286
 287         }
 288         catch(exception& e) {
 289                 m->errorOut(e, "ListSeqsCommand", "readName");
 290                 exit(1);
 291         }
 292 }
 293
 294 //**********************************************************************************************************************
 295 int ListSeqsCommand::readGroup(){
 296         try {
 297
 298                 ifstream in;
 299                 openInputFile(groupfile, in);
 300                 string name, group;
 301
 302                 while(!in.eof()){
 303
 304                         if (m->control_pressed) { in.close(); return 0; }
 305
 306                         in >> name;                             //read from first column
 307                         in >> group;                    //read from second column
 308
 309                         names.push_back(name);
 310
 311                         gobble(in);
 312                 }
 313                 in.close();
 314                 return 0;
 315
 316         }
 317         catch(exception& e) {
 318                 m->errorOut(e, "ListSeqsCommand", "readGroup");
 319                 exit(1);
 320         }
 321 }
 322
 323 //**********************************************************************************************************************
 324 //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
 325 int ListSeqsCommand::readAlign(){
 326         try {
 327
 328                 ifstream in;
 329                 openInputFile(alignfile, in);
 330                 string name, junk;
 331
 332                 //read column headers
 333                 for (int i = 0; i < 16; i++) {
 334                         if (!in.eof())  {       in >> junk;             }
 335                         else                    {       break;                  }
 336                 }
 337
 338
 339                 while(!in.eof()){
 340
 341                         if (m->control_pressed) { in.close(); return 0; }
 342
 343                         in >> name;                             //read from first column
 344
 345                         //read rest
 346                         for (int i = 0; i < 15; i++) {
 347                                 if (!in.eof())  {       in >> junk;             }
 348                                 else                    {       break;                  }
 349                         }
 350
 351                         names.push_back(name);
 352
 353                         gobble(in);
 354                 }
 355                 in.close();
 356
 357                 return 0;
 358
 359
 360         }
 361         catch(exception& e) {
 362                 m->errorOut(e, "ListSeqsCommand", "readAlign");
 363                 exit(1);
 364         }
 365 }
 366 //**********************************************************************************************************************