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 == ""))  { mothurOut("You must provide a file."); mothurOutEndLine(); abort = true; }
 111
 112                         int okay = 1;
 113                         if (outputDir != "") { okay++; }
 114
 115                         if (parameters.size() > okay) { mothurOut("You may only enter one file."); mothurOutEndLine(); abort = true;  }
 116                 }
 117
 118         }
 119         catch(exception& e) {
 120                 errorOut(e, "ListSeqsCommand", "ListSeqsCommand");
 121                 exit(1);
 122         }
 123 }
 124 //**********************************************************************************************************************
 125
 126 void ListSeqsCommand::help(){
 127         try {
 128                 mothurOut("The list.seqs command reads a fasta, name, group, list or alignreport file and outputs a .accnos file containing sequence names.\n");
 129                 mothurOut("The list.seqs command parameters are fasta, name, group and alignreport.  You must provide one of these parameters.\n");
 130                 mothurOut("The list.seqs command should be in the following format: list.seqs(fasta=yourFasta).\n");
 131                 mothurOut("Example list.seqs(fasta=amazon.fasta).\n");
 132                 mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
 133         }
 134         catch(exception& e) {
 135                 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                 return 0;
 171         }
 172
 173         catch(exception& e) {
 174                 errorOut(e, "ListSeqsCommand", "execute");
 175                 exit(1);
 176         }
 177 }
 178
 179 //**********************************************************************************************************************
 180 void ListSeqsCommand::readFasta(){
 181         try {
 182
 183                 ifstream in;
 184                 openInputFile(fastafile, in);
 185                 string name;
 186
 187                 while(!in.eof()){
 188                         Sequence currSeq(in);
 189                         name = currSeq.getName();
 190
 191                         if (name != "") {  names.push_back(name);  }
 192
 193                         gobble(in);
 194                 }
 195                 in.close();
 196
 197         }
 198         catch(exception& e) {
 199                 errorOut(e, "ListSeqsCommand", "readFasta");
 200                 exit(1);
 201         }
 202 }
 203 //**********************************************************************************************************************
 204 void ListSeqsCommand::readList(){
 205         try {
 206                 ifstream in;
 207                 openInputFile(listfile, in);
 208
 209                 if(!in.eof()){
 210                         //read in list vector
 211                         ListVector list(in);
 212
 213                         //for each bin
 214                         for (int i = 0; i < list.getNumBins(); i++) {
 215                                 string binnames = list.get(i);
 216
 217                                 while (binnames.find_first_of(',') != -1) {
 218                                         string name = binnames.substr(0,binnames.find_first_of(','));
 219                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 220                                         names.push_back(name);
 221                                 }
 222
 223                                 names.push_back(binnames);
 224                         }
 225                 }
 226                 in.close();
 227
 228         }
 229         catch(exception& e) {
 230                 errorOut(e, "ListSeqsCommand", "readList");
 231                 exit(1);
 232         }
 233 }
 234
 235 //**********************************************************************************************************************
 236 void ListSeqsCommand::readName(){
 237         try {
 238
 239                 ifstream in;
 240                 openInputFile(namefile, in);
 241                 string name, firstCol, secondCol;
 242
 243                 while(!in.eof()){
 244
 245                         in >> firstCol;
 246                         in >> secondCol;
 247
 248                         //parse second column saving each name
 249                         while (secondCol.find_first_of(',') != -1) {
 250                                 name = secondCol.substr(0,secondCol.find_first_of(','));
 251                                 secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
 252                                 names.push_back(name);
 253                         }
 254
 255                         //get name after last ,
 256                         names.push_back(secondCol);
 257
 258                         gobble(in);
 259                 }
 260                 in.close();
 261
 262         }
 263         catch(exception& e) {
 264                 errorOut(e, "ListSeqsCommand", "readName");
 265                 exit(1);
 266         }
 267 }
 268
 269 //**********************************************************************************************************************
 270 void ListSeqsCommand::readGroup(){
 271         try {
 272
 273                 ifstream in;
 274                 openInputFile(groupfile, in);
 275                 string name, group;
 276
 277                 while(!in.eof()){
 278
 279                         in >> name;                             //read from first column
 280                         in >> group;                    //read from second column
 281
 282                         names.push_back(name);
 283
 284                         gobble(in);
 285                 }
 286                 in.close();
 287
 288         }
 289         catch(exception& e) {
 290                 errorOut(e, "ListSeqsCommand", "readGroup");
 291                 exit(1);
 292         }
 293 }
 294
 295 //**********************************************************************************************************************
 296 //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
 297 void ListSeqsCommand::readAlign(){
 298         try {
 299
 300                 ifstream in;
 301                 openInputFile(alignfile, in);
 302                 string name, junk;
 303
 304                 //read column headers
 305                 for (int i = 0; i < 16; i++) {
 306                         if (!in.eof())  {       in >> junk;             }
 307                         else                    {       break;                  }
 308                 }
 309
 310
 311                 while(!in.eof()){
 312
 313                         in >> name;                             //read from first column
 314
 315                         //read rest
 316                         for (int i = 0; i < 15; i++) {
 317                                 if (!in.eof())  {       in >> junk;             }
 318                                 else                    {       break;                  }
 319                         }
 320
 321                         names.push_back(name);
 322
 323                         gobble(in);
 324                 }
 325                 in.close();
 326
 327
 328         }
 329         catch(exception& e) {
 330                 errorOut(e, "ListSeqsCommand", "readAlign");
 331                 exit(1);
 332         }
 333 }
 334 //**********************************************************************************************************************