sracommand.cpp

   1 //
   2 //  sracommand.cpp
   3 //  Mothur
   4 //
   5 //  Created by SarahsWork on 10/28/13.
   6 //  Copyright (c) 2013 Schloss Lab. All rights reserved.
   7 //
   8
   9 #include "sracommand.h"
  10 #include "sffinfocommand.h"
  11 #include "parsefastaqcommand.h"
  12
  13 //**********************************************************************************************************************
  14 vector<string> SRACommand::setParameters(){
  15         try {
  16         CommandParameter psff("sff", "InputTypes", "", "", "sffFastQFile", "sffFastQFile", "none","sra",false,false); parameters.push_back(psff);
  17         CommandParameter pgroup("group", "InputTypes", "", "", "groupOligos", "none", "none","sra",false,false); parameters.push_back(pgroup);
  18         CommandParameter poligos("oligos", "InputTypes", "", "", "groupOligos", "none", "none","sra",false,false); parameters.push_back(poligos);
  19         CommandParameter pfile("file", "InputTypes", "", "", "sffFastQFile", "sffFastQFile", "none","sra",false,false); parameters.push_back(pfile);
  20                 CommandParameter pfastq("fastq", "InputTypes", "", "", "sffFastQFile", "sffFastQFile", "none","sra",false,false); parameters.push_back(pfastq);
  21         //choose only one multiple options
  22         CommandParameter pplatform("platform", "Multiple", "454-???-???", "454", "", "", "","",false,false); parameters.push_back(pplatform);
  23         CommandParameter ppdiffs("pdiffs", "Number", "", "0", "", "", "","",false,false); parameters.push_back(ppdiffs);
  24                 CommandParameter pbdiffs("bdiffs", "Number", "", "0", "", "", "","",false,false); parameters.push_back(pbdiffs);
  25         CommandParameter pldiffs("ldiffs", "Number", "", "0", "", "", "","",false,false); parameters.push_back(pldiffs);
  26                 CommandParameter psdiffs("sdiffs", "Number", "", "0", "", "", "","",false,false); parameters.push_back(psdiffs);
  27         CommandParameter ptdiffs("tdiffs", "Number", "", "0", "", "", "","",false,false); parameters.push_back(ptdiffs);
  28
  29          //every command must have inputdir and outputdir.  This allows mothur users to redirect input and output files.
  30                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
  31                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
  32
  33                 vector<string> myArray;
  34                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  35                 return myArray;
  36         }
  37         catch(exception& e) {
  38                 m->errorOut(e, "SRACommand", "setParameters");
  39                 exit(1);
  40         }
  41 }
  42 //**********************************************************************************************************************
  43 string SRACommand::getHelpString(){
  44         try {
  45                 string helpString = "";
  46                 helpString += "The sra command creates a sequence read archive from sff or fastq files.\n";
  47                 helpString += "The sra command parameters are: sff, fastqfiles, oligos, platform....\n";
  48                 helpString += "The sffiles parameter is used to provide a file containing a \n";
  49         helpString += "The tdiffs parameter is used to specify the total number of differences allowed in the sequence. The default is pdiffs + bdiffs + sdiffs + ldiffs.\n";
  50                 helpString += "The bdiffs parameter is used to specify the number of differences allowed in the barcode. The default is 0.\n";
  51                 helpString += "The pdiffs parameter is used to specify the number of differences allowed in the primer. The default is 0.\n";
  52         helpString += "The ldiffs parameter is used to specify the number of differences allowed in the linker. The default is 0.\n";
  53                 helpString += "The sdiffs parameter is used to specify the number of differences allowed in the spacer. The default is 0.\n";
  54
  55                 helpString += "The new command should be in the following format: \n";
  56                 helpString += "new(...)\n";
  57                 return helpString;
  58         }
  59         catch(exception& e) {
  60                 m->errorOut(e, "SRACommand", "getHelpString");
  61                 exit(1);
  62         }
  63 }
  64 //**********************************************************************************************************************
  65 string SRACommand::getOutputPattern(string type) {
  66     try {
  67         string pattern = "";
  68
  69         if (type == "sra") {  pattern = "[filename],sra"; }
  70         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  71
  72         return pattern;
  73     }
  74     catch(exception& e) {
  75         m->errorOut(e, "SRACommand", "getOutputPattern");
  76         exit(1);
  77     }
  78 }
  79 //**********************************************************************************************************************
  80 SRACommand::SRACommand(){
  81         try {
  82                 abort = true; calledHelp = true;
  83                 setParameters();
  84         vector<string> tempOutNames;
  85                 outputTypes["sra"] = tempOutNames;
  86         }
  87         catch(exception& e) {
  88                 m->errorOut(e, "SRACommand", "SRACommand");
  89                 exit(1);
  90         }
  91 }
  92 //**********************************************************************************************************************
  93 SRACommand::SRACommand(string option)  {
  94         try {
  95                 abort = false; calledHelp = false;
  96
  97                 //allow user to run help
  98                 if(option == "help") { help(); abort = true; calledHelp = true; }
  99                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 100
 101                 else {
 102                         //valid paramters for this command
 103                         vector<string> myArray = setParameters();
 104
 105                         OptionParser parser(option);
 106                         map<string,string> parameters = parser.getParameters();
 107
 108                         ValidParameters validParameter;
 109                         map<string,string>::iterator it;
 110                         //check to make sure all parameters are valid for command
 111                         for (it = parameters.begin(); it != parameters.end(); it++) {
 112                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 113                         }
 114
 115
 116                         //if the user changes the input directory command factory will send this info to us in the output parameter
 117                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 118                         if (inputDir == "not found"){   inputDir = "";          }
 119                         else {
 120
 121                 string path;
 122                                 it = parameters.find("sff");
 123                                 //user has given a template file
 124                                 if(it != parameters.end()){
 125                                         path = m->hasPath(it->second);
 126                                         //if the user has not given a path then, add inputdir. else leave path alone.
 127                                         if (path == "") {       parameters["sff"] = inputDir + it->second;              }
 128                                 }
 129
 130                                 it = parameters.find("fastq");
 131                                 //user has given a template file
 132                                 if(it != parameters.end()){
 133                                         path = m->hasPath(it->second);
 134                                         //if the user has not given a path then, add inputdir. else leave path alone.
 135                                         if (path == "") {       parameters["fastq"] = inputDir + it->second;            }
 136                                 }
 137
 138                 it = parameters.find("file");
 139                                 //user has given a template file
 140                                 if(it != parameters.end()){
 141                                         path = m->hasPath(it->second);
 142                                         //if the user has not given a path then, add inputdir. else leave path alone.
 143                                         if (path == "") {       parameters["file"] = inputDir + it->second;             }
 144                                 }
 145
 146                 it = parameters.find("group");
 147                                 //user has given a template file
 148                                 if(it != parameters.end()){
 149                                         path = m->hasPath(it->second);
 150                                         //if the user has not given a path then, add inputdir. else leave path alone.
 151                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 152                                 }
 153
 154                 it = parameters.find("oligos");
 155                                 //user has given a template file
 156                                 if(it != parameters.end()){
 157                                         path = m->hasPath(it->second);
 158                                         //if the user has not given a path then, add inputdir. else leave path alone.
 159                                         if (path == "") {       parameters["oligos"] = inputDir + it->second;           }
 160                                 }
 161             }
 162
 163                         //check for parameters
 164             fastqfile = validParameter.validFile(parameters, "fastq", true);
 165                         if (fastqfile == "not open") { fastqfile = "";  abort = true; }
 166                         else if (fastqfile == "not found") { fastqfile = ""; }
 167
 168                         sfffile = validParameter.validFile(parameters, "sff", true);
 169                         if (sfffile == "not open") {  sfffile = "";  abort = true; }
 170                         else if (sfffile == "not found") { sfffile = ""; }
 171
 172             file = validParameter.validFile(parameters, "file", true);
 173                         if (file == "not open") {  file = "";  abort = true; }
 174                         else if (file == "not found") { file = ""; }
 175
 176             groupfile = validParameter.validFile(parameters, "group", true);
 177                         if (groupfile == "not open") {  groupfile = "";  abort = true; }
 178                         else if (groupfile == "not found") { groupfile = ""; }
 179             else {  m->setGroupFile(groupfile); }
 180
 181             oligosfile = validParameter.validFile(parameters, "oligos", true);
 182                         if (oligosfile == "not found")      {   oligosfile = "";        }
 183                         else if(oligosfile == "not open")       {       abort = true;           }
 184                         else {  m->setOligosFile(oligosfile); }
 185
 186
 187             file = validParameter.validFile(parameters, "file", true);
 188                         if (file == "not open") {  file = "";  abort = true; }
 189                         else if (file == "not found") { file = ""; }
 190
 191                         if ((fastqfile == "") && (sfffile == "") && (sfffile == "")) {
 192                 m->mothurOut("[ERROR]: You must provide a file, sff file or fastq file before you can use the sra command."); m->mothurOutEndLine(); abort = true;
 193             }
 194
 195             if ((groupfile != "") && (oligosfile != "")) {
 196                 m->mothurOut("[ERROR]: You may not use a group file and an oligos file, only one."); m->mothurOutEndLine(); abort = true;
 197             }
 198
 199             if ((fastqfile != "") || (sfffile != "")) {
 200                 if ((groupfile == "") && (oligosfile == "")) {
 201                     oligosfile = m->getOligosFile();
 202                                         if (oligosfile != "") {  m->mothurOut("Using " + oligosfile + " as input file for the oligos parameter."); m->mothurOutEndLine(); }
 203                                         else {
 204                                                 groupfile = m->getGroupFile();
 205                         if (groupfile != "") {  m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 206                         else {
 207                             m->mothurOut("[ERROR]: You must provide groupfile or oligos file if splitting a fastq or sff file."); m->mothurOutEndLine(); abort = true;
 208                         }
 209                                         }
 210                 }
 211             }
 212
 213             //use only one Mutliple type
 214                         platform = validParameter.validFile(parameters, "platform", false);
 215                         if (platform == "not found") { platform = "454"; }
 216
 217                         if ((platform == "454") || (platform == "????") || (platform == "????") || (platform == "????")) { }
 218                         else { m->mothurOut("Not a valid platform option.  Valid platform options are 454, ...."); m->mothurOutEndLine(); abort = true; }
 219
 220
 221             string temp = validParameter.validFile(parameters, "bdiffs", false);                if (temp == "not found"){       temp = "0";             }
 222                         m->mothurConvert(temp, bdiffs);
 223
 224                         temp = validParameter.validFile(parameters, "pdiffs", false);           if (temp == "not found"){       temp = "0";             }
 225                         m->mothurConvert(temp, pdiffs);
 226
 227             temp = validParameter.validFile(parameters, "ldiffs", false);               if (temp == "not found") { temp = "0"; }
 228                         m->mothurConvert(temp, ldiffs);
 229
 230             temp = validParameter.validFile(parameters, "sdiffs", false);               if (temp == "not found") { temp = "0"; }
 231                         m->mothurConvert(temp, sdiffs);
 232
 233                         temp = validParameter.validFile(parameters, "tdiffs", false);           if (temp == "not found") { int tempTotal = pdiffs + bdiffs + ldiffs + sdiffs;  temp = toString(tempTotal); }
 234                         m->mothurConvert(temp, tdiffs);
 235
 236                         if(tdiffs == 0){        tdiffs = bdiffs + pdiffs + ldiffs + sdiffs;     }
 237
 238
 239
 240                 }
 241
 242         }
 243         catch(exception& e) {
 244                 m->errorOut(e, "SRACommand", "SRACommand");
 245                 exit(1);
 246         }
 247 }
 248 //**********************************************************************************************************************
 249 int SRACommand::execute(){
 250         try {
 251
 252                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 253
 254         //parse files
 255         vector<string> filesBySample;
 256
 257         if (file != "")             {       readFile(filesBySample);        }
 258         else if (sfffile != "")     {       parseSffFile(filesBySample);    }
 259         else if (fastqfile != "")   {       parseFastqFile(filesBySample);  }
 260
 261
 262
 263
 264         //output files created by command
 265                 m->mothurOutEndLine();
 266                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 267                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 268                 m->mothurOutEndLine();
 269         return 0;
 270
 271     }
 272         catch(exception& e) {
 273                 m->errorOut(e, "SRACommand", "SRACommand");
 274                 exit(1);
 275         }
 276 }
 277 //**********************************************************************************************************************
 278 int SRACommand::readFile(vector<string>& files){
 279         try {
 280
 281         return 0;
 282     }
 283         catch(exception& e) {
 284                 m->errorOut(e, "SRACommand", "readFile");
 285                 exit(1);
 286         }
 287 }
 288 //**********************************************************************************************************************
 289 int SRACommand::parseSffFile(vector<string>& files){
 290         try {
 291         //run sffinfo to parse sff file into individual sampled sff files
 292         string commandString = "sff=" + sfffile;
 293         if (groupfile != "") { commandString += ", group=" + groupfile; }
 294         else if (oligosfile != "") {
 295             commandString += ", oligos=" + oligosfile;
 296             //add in pdiffs, bdiffs, ldiffs, sdiffs, tdiffs
 297             if (pdiffs != 0) { commandString += ", pdiffs=" + toString(pdiffs); }
 298             if (bdiffs != 0) { commandString += ", bdiffs=" + toString(bdiffs); }
 299             if (ldiffs != 0) { commandString += ", ldiffs=" + toString(ldiffs); }
 300             if (sdiffs != 0) { commandString += ", sdiffs=" + toString(sdiffs); }
 301             if (tdiffs != 0) { commandString += ", tdiffs=" + toString(tdiffs); }
 302         }
 303         m->mothurOutEndLine();
 304         m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 305         m->mothurOut("Running command: sffinfo(" + commandString + ")"); m->mothurOutEndLine();
 306         m->mothurCalling = true;
 307
 308         Command* sffinfoCommand = new SffInfoCommand(commandString);
 309         sffinfoCommand->execute();
 310
 311         map<string, vector<string> > filenames = sffinfoCommand->getOutputFiles();
 312         map<string, vector<string> >::iterator it = filenames.find("sff");
 313         if (it != filenames.end()) { files = it->second; }
 314         else { m->control_pressed = true; } // error in sffinfo
 315
 316         delete sffinfoCommand;
 317         m->mothurCalling = false;
 318         m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 319
 320         return 0;
 321     }
 322         catch(exception& e) {
 323                 m->errorOut(e, "SRACommand", "readFile");
 324                 exit(1);
 325         }
 326 }
 327
 328 //**********************************************************************************************************************
 329 int SRACommand::parseFastqFile(vector<string>& files){
 330         try {
 331
 332         //run sffinfo to parse sff file into individual sampled sff files
 333         string commandString = "fastq=" + fastqfile;
 334         if (groupfile != "") { commandString += ", group=" + groupfile; }
 335         else if (oligosfile != "") {
 336             commandString += ", oligos=" + oligosfile;
 337             //add in pdiffs, bdiffs, ldiffs, sdiffs, tdiffs
 338             if (pdiffs != 0) { commandString += ", pdiffs=" + toString(pdiffs); }
 339             if (bdiffs != 0) { commandString += ", bdiffs=" + toString(bdiffs); }
 340             if (ldiffs != 0) { commandString += ", ldiffs=" + toString(ldiffs); }
 341             if (sdiffs != 0) { commandString += ", sdiffs=" + toString(sdiffs); }
 342             if (tdiffs != 0) { commandString += ", tdiffs=" + toString(tdiffs); }
 343         }
 344         m->mothurOutEndLine();
 345         m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 346         m->mothurOut("Running command: fastq.info(" + commandString + ")"); m->mothurOutEndLine();
 347         m->mothurCalling = true;
 348
 349         Command* fastqinfoCommand = new ParseFastaQCommand(commandString);
 350         fastqinfoCommand->execute();
 351
 352         map<string, vector<string> > filenames = fastqinfoCommand->getOutputFiles();
 353         map<string, vector<string> >::iterator it = filenames.find("fastq");
 354         if (it != filenames.end()) { files = it->second; }
 355         else { m->control_pressed = true; } // error in sffinfo
 356
 357         delete fastqinfoCommand;
 358         m->mothurCalling = false;
 359         m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 360
 361         return 0;
 362     }
 363         catch(exception& e) {
 364                 m->errorOut(e, "SRACommand", "readFile");
 365                 exit(1);
 366         }
 367 }
 368 //**********************************************************************************************************************
 369
 370