splitgroupscommand.cpp

   1 /*
   2  *  splitgroupscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 9/20/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "splitgroupscommand.h"
  11 #include "sharedutilities.h"
  12 #include "sequenceparser.h"
  13 #include "counttable.h"
  14
  15 //**********************************************************************************************************************
  16 vector<string> SplitGroupCommand::setParameters(){
  17         try {
  18                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none","fasta",false,true,true); parameters.push_back(pfasta);
  19         CommandParameter pname("name", "InputTypes", "", "", "NameCount", "none", "none","name",false,false,true); parameters.push_back(pname);
  20         CommandParameter pcount("count", "InputTypes", "", "", "NameCount-CountGroup", "CountGroup", "none","count",false,false,true); parameters.push_back(pcount);
  21                 CommandParameter pgroup("group", "InputTypes", "", "", "CountGroup", "CountGroup", "none","group",false,false,true); parameters.push_back(pgroup);
  22                 CommandParameter pgroups("groups", "String", "", "", "", "", "","",false,false); parameters.push_back(pgroups);
  23                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
  24                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
  25
  26                 vector<string> myArray;
  27                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  28                 return myArray;
  29         }
  30         catch(exception& e) {
  31                 m->errorOut(e, "SplitGroupCommand", "setParameters");
  32                 exit(1);
  33         }
  34 }
  35 //**********************************************************************************************************************
  36 string SplitGroupCommand::getHelpString(){
  37         try {
  38                 string helpString = "";
  39                 helpString += "The split.groups command reads a group or count file, and parses your fasta and names or count files by groups. \n";
  40                 helpString += "The split.groups command parameters are fasta, name, group, count and groups.\n";
  41                 helpString += "The fasta and group or count parameters are required.\n";
  42                 helpString += "The groups parameter allows you to select groups to create files for.  \n";
  43                 helpString += "For example if you set groups=A-B-C, you will get a .A.fasta, .A.names, .B.fasta, .B.names, .C.fasta, .C.names files.  \n";
  44                 helpString += "If you want .fasta and .names files for all groups, set groups=all.  \n";
  45                 helpString += "The split.groups command should be used in the following format: split.group(fasta=yourFasta, group=yourGroupFile).\n";
  46                 helpString += "Example: split.groups(fasta=abrecovery.fasta, group=abrecovery.groups).\n";
  47                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
  48                 return helpString;
  49         }
  50         catch(exception& e) {
  51                 m->errorOut(e, "SplitGroupCommand", "getHelpString");
  52                 exit(1);
  53         }
  54 }
  55 //**********************************************************************************************************************
  56 string SplitGroupCommand::getOutputPattern(string type) {
  57     try {
  58         string pattern = "";
  59
  60         if (type == "fasta") {  pattern = "[filename],[group],fasta"; }
  61         else if (type == "name") {  pattern = "[filename],[group],names"; }
  62         else if (type == "count") {  pattern = "[filename],[group],count_table"; }
  63         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  64
  65         return pattern;
  66     }
  67     catch(exception& e) {
  68         m->errorOut(e, "SplitGroupCommand", "getOutputPattern");
  69         exit(1);
  70     }
  71 }
  72
  73 //**********************************************************************************************************************
  74 SplitGroupCommand::SplitGroupCommand(){
  75         try {
  76                 abort = true; calledHelp = true;
  77                 setParameters();
  78                 vector<string> tempOutNames;
  79                 outputTypes["fasta"] = tempOutNames;
  80                 outputTypes["name"] = tempOutNames;
  81         outputTypes["count"] = tempOutNames;
  82         }
  83         catch(exception& e) {
  84                 m->errorOut(e, "SplitGroupCommand", "SplitGroupCommand");
  85                 exit(1);
  86         }
  87 }
  88 //**********************************************************************************************************************
  89 SplitGroupCommand::SplitGroupCommand(string option)  {
  90         try {
  91                 abort = false; calledHelp = false;
  92
  93                 //allow user to run help
  94                 if(option == "help") { help(); abort = true; calledHelp = true; }
  95                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  96
  97                 else {
  98                         vector<string> myArray = setParameters();
  99
 100                         OptionParser parser(option);
 101                         map<string, string> parameters = parser.getParameters();
 102
 103                         ValidParameters validParameter;
 104                         map<string, string>::iterator it;
 105
 106                         //check to make sure all parameters are valid for command
 107                         for (it = parameters.begin(); it != parameters.end(); it++) {
 108                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 109                         }
 110
 111                         //initialize outputTypes
 112                         vector<string> tempOutNames;
 113                         outputTypes["fasta"] = tempOutNames;
 114                         outputTypes["name"] = tempOutNames;
 115             outputTypes["count"] = tempOutNames;
 116
 117                         //if the user changes the input directory command factory will send this info to us in the output parameter
 118                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 119                         if (inputDir == "not found"){   inputDir = "";          }
 120                         else {
 121                                 string path;
 122                                 it = parameters.find("group");
 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["group"] = inputDir + it->second;            }
 128                                 }
 129
 130                                 it = parameters.find("fasta");
 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["fasta"] = inputDir + it->second;            }
 136                                 }
 137
 138                                 it = parameters.find("name");
 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["name"] = inputDir + it->second;             }
 144                                 }
 145
 146                 it = parameters.find("count");
 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["count"] = inputDir + it->second;            }
 152                                 }
 153                         }
 154
 155
 156                         namefile = validParameter.validFile(parameters, "name", true);
 157                         if (namefile == "not open") { namefile = ""; abort = true; }
 158                         else if (namefile == "not found") { namefile = ""; }
 159                         else { m->setNameFile(namefile); }
 160
 161                         fastafile = validParameter.validFile(parameters, "fasta", true);
 162                         if (fastafile == "not open") { abort = true; }
 163                         else if (fastafile == "not found") {
 164                                 fastafile = m->getFastaFile();
 165                                 if (fastafile != "") { m->mothurOut("Using " + fastafile + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 166                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 167                         }else { m->setFastaFile(fastafile); }
 168
 169                         groupfile = validParameter.validFile(parameters, "group", true);
 170                         if (groupfile == "not open") {  groupfile = ""; abort = true; }
 171                         else if (groupfile == "not found") { groupfile = "";
 172                         }else {  m->setGroupFile(groupfile); }
 173
 174             countfile = validParameter.validFile(parameters, "count", true);
 175                         if (countfile == "not open") { countfile = ""; abort = true; }
 176                         else if (countfile == "not found") { countfile = ""; }
 177                         else { m->setCountTableFile(countfile); }
 178
 179             if ((countfile != "") && (namefile != "")) { m->mothurOut("You must enter ONLY ONE of the following: count or name."); m->mothurOutEndLine(); abort = true; }
 180
 181             if ((countfile != "") && (groupfile != "")) { m->mothurOut("You must enter ONLY ONE of the following: count or group."); m->mothurOutEndLine(); abort = true; }
 182
 183             if ((countfile == "") && (groupfile == "")) {
 184                 if (namefile == "") { //check for count then group
 185                     countfile = m->getCountTableFile();
 186                                         if (countfile != "") {  m->mothurOut("Using " + countfile + " as input file for the count parameter."); m->mothurOutEndLine(); }
 187                                         else {
 188                                                 groupfile = m->getGroupFile();
 189                         if (groupfile != "") {  m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 190                         else {
 191                             m->mothurOut("You need to provide a count or group file."); m->mothurOutEndLine();
 192                             abort = true;
 193                         }
 194                                         }
 195                 }else { //check for group
 196                     groupfile = m->getGroupFile();
 197                     if (groupfile != "") {  m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 198                     else {
 199                         m->mothurOut("You need to provide a count or group file."); m->mothurOutEndLine();
 200                         abort = true;
 201                     }
 202                 }
 203             }
 204
 205                         groups = validParameter.validFile(parameters, "groups", false);
 206                         if (groups == "not found") { groups = ""; }
 207                         else { m->splitAtDash(groups, Groups);  }
 208
 209                         //if the user changes the output directory command factory will send this info to us in the output parameter
 210                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 211                 if (groupfile != "") { outputDir = m->hasPath(groupfile); }
 212                 else { outputDir = m->hasPath(countfile);  }
 213             }
 214
 215             if (countfile == "") {
 216                 if (namefile == "") {
 217                     vector<string> files; files.push_back(fastafile);
 218                     parser.getNameFile(files);
 219                 }
 220             }
 221                 }
 222
 223         }
 224         catch(exception& e) {
 225                 m->errorOut(e, "SplitGroupCommand", "SplitAbundCommand");
 226                 exit(1);
 227         }
 228 }
 229 //**********************************************************************************************************************
 230 int SplitGroupCommand::execute(){
 231         try {
 232
 233                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 234
 235         if (countfile == "" ) {  runNameGroup();  }
 236         else { runCount();  }
 237
 238                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);        } return 0; }
 239
 240                 string current = "";
 241                 itTypes = outputTypes.find("fasta");
 242                 if (itTypes != outputTypes.end()) {
 243                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 244                 }
 245
 246                 itTypes = outputTypes.find("name");
 247                 if (itTypes != outputTypes.end()) {
 248                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 249                 }
 250
 251         itTypes = outputTypes.find("count");
 252                 if (itTypes != outputTypes.end()) {
 253                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setCountTableFile(current); }
 254                 }
 255
 256                 m->mothurOutEndLine();
 257                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 258                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 259                 m->mothurOutEndLine();
 260
 261                 return 0;
 262         }
 263         catch(exception& e) {
 264                 m->errorOut(e, "SplitGroupCommand", "execute");
 265                 exit(1);
 266         }
 267 }
 268 //**********************************************************************************************************************
 269 int SplitGroupCommand::runNameGroup(){
 270         try {
 271         SequenceParser* parser;
 272                 if (namefile == "") {   parser = new SequenceParser(groupfile, fastafile);                              }
 273                 else                            {       parser = new SequenceParser(groupfile, fastafile, namefile);    }
 274
 275                 if (m->control_pressed) { delete parser; return 0; }
 276
 277                 vector<string> namesGroups = parser->getNamesOfGroups();
 278                 SharedUtil util;  util.setGroups(Groups, namesGroups);
 279
 280                 string fastafileRoot = outputDir + m->getRootName(m->getSimpleName(fastafile));
 281                 string namefileRoot = outputDir + m->getRootName(m->getSimpleName(namefile));
 282
 283                 m->mothurOutEndLine();
 284                 for (int i = 0; i < Groups.size(); i++) {
 285
 286                         m->mothurOut("Processing group: " + Groups[i]); m->mothurOutEndLine();
 287
 288             map<string, string> variables;
 289             variables["[filename]"] = fastafileRoot;
 290             variables["[group]"] = Groups[i];
 291
 292                         string newFasta = getOutputFileName("fasta",variables);
 293             variables["[filename]"] = namefileRoot;
 294                         string newName = getOutputFileName("name",variables);
 295
 296                         parser->getSeqs(Groups[i], newFasta, false);
 297                         outputNames.push_back(newFasta); outputTypes["fasta"].push_back(newFasta);
 298                         if (m->control_pressed) { delete parser; for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]);        } return 0; }
 299
 300                         if (namefile != "") {
 301                                 parser->getNameMap(Groups[i], newName);
 302                                 outputNames.push_back(newName); outputTypes["name"].push_back(newName);
 303                         }
 304
 305                         if (m->control_pressed) { delete parser; for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]);        } return 0; }
 306                 }
 307
 308                 delete parser;
 309
 310         return 0;
 311
 312     }
 313         catch(exception& e) {
 314                 m->errorOut(e, "SplitGroupCommand", "runNameGroup");
 315                 exit(1);
 316         }
 317 }
 318 //**********************************************************************************************************************
 319 int SplitGroupCommand::runCount(){
 320         try {
 321
 322         CountTable ct;
 323         ct.readTable(countfile, true);
 324         if (!ct.hasGroupInfo()) { m->mothurOut("[ERROR]: your count file does not contain group info, cannot split by group.\n"); m->control_pressed = true; }
 325
 326         if (m->control_pressed) { return 0; }
 327
 328         vector<string> namesGroups = ct.getNamesOfGroups();
 329         SharedUtil util;  util.setGroups(Groups, namesGroups);
 330
 331         //fill filehandles with neccessary ofstreams
 332         map<string, ofstream*> ffiles;
 333         map<string, ofstream*> cfiles;
 334         ofstream* temp;
 335         for (int i=0; i<Groups.size(); i++) {
 336             temp = new ofstream;
 337             ffiles[Groups[i]] = temp;
 338             map<string, string> variables;
 339             variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(fastafile));
 340             variables["[group]"] = Groups[i];
 341             string newFasta = getOutputFileName("fasta",variables);
 342             outputNames.push_back(newFasta); outputTypes["fasta"].push_back(newFasta);
 343             m->openOutputFile(newFasta, (*temp));
 344             temp = new ofstream;
 345             cfiles[Groups[i]] = temp;
 346             variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(countfile));
 347             string newCount = getOutputFileName("count",variables);
 348             m->openOutputFile(newCount, (*temp));
 349             outputNames.push_back(newCount); outputTypes["count"].push_back(newCount);
 350             (*temp) << "Representative_Sequence\ttotal\t" << Groups[i] << endl;
 351         }
 352
 353         ifstream in;
 354         m->openInputFile(fastafile, in);
 355
 356         while (!in.eof()) {
 357             Sequence seq(in); m->gobble(in);
 358
 359             if (m->control_pressed) { break; }
 360             if (seq.getName() != "") {
 361                 vector<string> thisSeqsGroups = ct.getGroups(seq.getName());
 362                 for (int i = 0; i < thisSeqsGroups.size(); i++) {
 363                     if (m->inUsersGroups(thisSeqsGroups[i], Groups)) { //if this sequence belongs to a group we want them print
 364                         seq.printSequence(*(ffiles[thisSeqsGroups[i]]));
 365                         int numSeqs = ct.getGroupCount(seq.getName(), Groups[i]);
 366                         (*(cfiles[thisSeqsGroups[i]])) << seq.getName() << '\t' << numSeqs << '\t' << numSeqs << endl;
 367                     }
 368                 }
 369             }
 370         }
 371         in.close();
 372
 373         //close and delete ofstreams
 374         for (int i=0; i<Groups.size(); i++) {
 375             (*ffiles[Groups[i]]).close(); delete ffiles[Groups[i]];
 376             (*cfiles[Groups[i]]).close(); delete cfiles[Groups[i]];
 377         }
 378
 379         return 0;
 380
 381     }
 382         catch(exception& e) {
 383                 m->errorOut(e, "SplitGroupCommand", "runCount");
 384                 exit(1);
 385         }
 386 }
 387 //**********************************************************************************************************************
 388
 389