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
 115                         if (parameters.size() > okay) { m->mothurOut("You may only enter one file."); m->mothurOutEndLine(); abort = true;  }
 116                 }
 117
 118         }
 119         catch(exception& e) {
 120                 m->errorOut(e, "ListSeqsCommand", "ListSeqsCommand");
 121                 exit(1);
 122         }
 123 }
 124 //**********************************************************************************************************************
 125
 126 void ListSeqsCommand::help(){
 127         try {
 128                 m->mothurOut("The list.seqs command reads a fasta, name, group, list or alignreport file and outputs a .accnos file containing sequence names.\n");
 129                 m->mothurOut("The list.seqs command parameters are fasta, name, group and alignreport.  You must provide one of these parameters.\n");
 130                 m->mothurOut("The list.seqs command should be in the following format: list.seqs(fasta=yourFasta).\n");
 131                 m->mothurOut("Example list.seqs(fasta=amazon.fasta).\n");
 132                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
 133         }
 134         catch(exception& e) {
 135                 m->errorOut(e, "ListSeqsCommand", "help");
 136                 exit(1);
 137         }
 138 }
 139
 140 //**********************************************************************************************************************
 141
 142 int ListSeqsCommand::execute(){
 143         try {
 144
 145                 if (abort == true) { return 0; }
 146
 147                 //read functions fill names vector
 148                 if (fastafile != "")            {       inputFileName = fastafile;      readFasta();    }
 149                 else if (namefile != "")        {       inputFileName = namefile;       readName();             }
 150                 else if (groupfile != "")       {       inputFileName = groupfile;      readGroup();    }
 151                 else if (alignfile != "")       {       inputFileName = alignfile;      readAlign();    }
 152                 else if (listfile != "")        {       inputFileName = listfile;       readList();             }
 153
 154                 //sort in alphabetical order
 155                 sort(names.begin(), names.end());
 156
 157                 if (outputDir == "") {  outputDir += hasPath(inputFileName);  }
 158
 159                 string outputFileName = outputDir + getRootName(getSimpleName(inputFileName)) + "accnos";
 160
 161                 ofstream out;
 162                 openOutputFile(outputFileName, out);
 163
 164                 //output to .accnos file
 165                 for (int i = 0; i < names.size(); i++) {
 166                         out << names[i] << endl;
 167                 }
 168                 out.close();
 169
 170                 m->mothurOutEndLine();
 171                 m->mothurOut("Output File Name: "); m->mothurOutEndLine();
 172                 m->mothurOut(outputFileName); m->mothurOutEndLine();
 173                 m->mothurOutEndLine();
 174
 175                 return 0;
 176         }
 177
 178         catch(exception& e) {
 179                 m->errorOut(e, "ListSeqsCommand", "execute");
 180                 exit(1);
 181         }
 182 }
 183
 184 //**********************************************************************************************************************
 185 void ListSeqsCommand::readFasta(){
 186         try {
 187
 188                 ifstream in;
 189                 openInputFile(fastafile, in);
 190                 string name;
 191
 192                 while(!in.eof()){
 193                         Sequence currSeq(in);
 194                         name = currSeq.getName();
 195
 196                         if (name != "") {  names.push_back(name);  }
 197
 198                         gobble(in);
 199                 }
 200                 in.close();
 201
 202         }
 203         catch(exception& e) {
 204                 m->errorOut(e, "ListSeqsCommand", "readFasta");
 205                 exit(1);
 206         }
 207 }
 208 //**********************************************************************************************************************
 209 void ListSeqsCommand::readList(){
 210         try {
 211                 ifstream in;
 212                 openInputFile(listfile, in);
 213
 214                 if(!in.eof()){
 215                         //read in list vector
 216                         ListVector list(in);
 217
 218                         //for each bin
 219                         for (int i = 0; i < list.getNumBins(); i++) {
 220                                 string binnames = list.get(i);
 221
 222                                 while (binnames.find_first_of(',') != -1) {
 223                                         string name = binnames.substr(0,binnames.find_first_of(','));
 224                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 225                                         names.push_back(name);
 226                                 }
 227
 228                                 names.push_back(binnames);
 229                         }
 230                 }
 231                 in.close();
 232
 233         }
 234         catch(exception& e) {
 235                 m->errorOut(e, "ListSeqsCommand", "readList");
 236                 exit(1);
 237         }
 238 }
 239
 240 //**********************************************************************************************************************
 241 void ListSeqsCommand::readName(){
 242         try {
 243
 244                 ifstream in;
 245                 openInputFile(namefile, in);
 246                 string name, firstCol, secondCol;
 247
 248                 while(!in.eof()){
 249
 250                         in >> firstCol;
 251                         in >> secondCol;
 252
 253                         //parse second column saving each name
 254                         while (secondCol.find_first_of(',') != -1) {
 255                                 name = secondCol.substr(0,secondCol.find_first_of(','));
 256                                 secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
 257                                 names.push_back(name);
 258                         }
 259
 260                         //get name after last ,
 261                         names.push_back(secondCol);
 262
 263                         gobble(in);
 264                 }
 265                 in.close();
 266
 267         }
 268         catch(exception& e) {
 269                 m->errorOut(e, "ListSeqsCommand", "readName");
 270                 exit(1);
 271         }
 272 }
 273
 274 //**********************************************************************************************************************
 275 void ListSeqsCommand::readGroup(){
 276         try {
 277
 278                 ifstream in;
 279                 openInputFile(groupfile, in);
 280                 string name, group;
 281
 282                 while(!in.eof()){
 283
 284                         in >> name;                             //read from first column
 285                         in >> group;                    //read from second column
 286
 287                         names.push_back(name);
 288
 289                         gobble(in);
 290                 }
 291                 in.close();
 292
 293         }
 294         catch(exception& e) {
 295                 m->errorOut(e, "ListSeqsCommand", "readGroup");
 296                 exit(1);
 297         }
 298 }
 299
 300 //**********************************************************************************************************************
 301 //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
 302 void ListSeqsCommand::readAlign(){
 303         try {
 304
 305                 ifstream in;
 306                 openInputFile(alignfile, in);
 307                 string name, junk;
 308
 309                 //read column headers
 310                 for (int i = 0; i < 16; i++) {
 311                         if (!in.eof())  {       in >> junk;             }
 312                         else                    {       break;                  }
 313                 }
 314
 315
 316                 while(!in.eof()){
 317
 318                         in >> name;                             //read from first column
 319
 320                         //read rest
 321                         for (int i = 0; i < 15; i++) {
 322                                 if (!in.eof())  {       in >> junk;             }
 323                                 else                    {       break;                  }
 324                         }
 325
 326                         names.push_back(name);
 327
 328                         gobble(in);
 329                 }
 330                 in.close();
 331
 332
 333         }
 334         catch(exception& e) {
 335                 m->errorOut(e, "ListSeqsCommand", "readAlign");
 336                 exit(1);
 337         }
 338 }
 339 //**********************************************************************************************************************