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
  13 //**********************************************************************************************************************
  14 vector<string> SplitGroupCommand::setParameters(){
  15         try {
  16                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfasta);
  17                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
  18                 CommandParameter pgroup("group", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pgroup);
  19                 CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
  20                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  21                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  22
  23                 vector<string> myArray;
  24                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  25                 return myArray;
  26         }
  27         catch(exception& e) {
  28                 m->errorOut(e, "SplitGroupCommand", "setParameters");
  29                 exit(1);
  30         }
  31 }
  32 //**********************************************************************************************************************
  33 string SplitGroupCommand::getHelpString(){
  34         try {
  35                 string helpString = "";
  36                 helpString += "The split.group command reads a group file, and parses your fasta and names files by groups. \n";
  37                 helpString += "The split.group command parameters are fasta, name, group and groups.\n";
  38                 helpString += "The fasta and group parameters are required.\n";
  39                 helpString += "The groups parameter allows you to select groups to create files for.  \n";
  40                 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";
  41                 helpString += "If you want .fasta and .names files for all groups, set groups=all.  \n";
  42                 helpString += "The split.group command should be used in the following format: split.group(fasta=yourFasta, group=yourGroupFile).\n";
  43                 helpString += "Example: split.group(fasta=abrecovery.fasta, group=abrecovery.groups).\n";
  44                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
  45                 return helpString;
  46         }
  47         catch(exception& e) {
  48                 m->errorOut(e, "SplitGroupCommand", "getHelpString");
  49                 exit(1);
  50         }
  51 }
  52 //**********************************************************************************************************************
  53 SplitGroupCommand::SplitGroupCommand(){
  54         try {
  55                 abort = true; calledHelp = true;
  56                 setParameters();
  57                 vector<string> tempOutNames;
  58                 outputTypes["fasta"] = tempOutNames;
  59                 outputTypes["name"] = tempOutNames;
  60         }
  61         catch(exception& e) {
  62                 m->errorOut(e, "SplitGroupCommand", "SplitGroupCommand");
  63                 exit(1);
  64         }
  65 }
  66 //**********************************************************************************************************************
  67 SplitGroupCommand::SplitGroupCommand(string option)  {
  68         try {
  69                 abort = false; calledHelp = false;
  70
  71                 //allow user to run help
  72                 if(option == "help") { help(); abort = true; calledHelp = true; }
  73                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  74
  75                 else {
  76                         vector<string> myArray = setParameters();
  77
  78                         OptionParser parser(option);
  79                         map<string, string> parameters = parser.getParameters();
  80
  81                         ValidParameters validParameter;
  82                         map<string, string>::iterator it;
  83
  84                         //check to make sure all parameters are valid for command
  85                         for (it = parameters.begin(); it != parameters.end(); it++) {
  86                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  87                         }
  88
  89                         //initialize outputTypes
  90                         vector<string> tempOutNames;
  91                         outputTypes["fasta"] = tempOutNames;
  92                         outputTypes["name"] = tempOutNames;
  93
  94                         //if the user changes the input directory command factory will send this info to us in the output parameter
  95                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  96                         if (inputDir == "not found"){   inputDir = "";          }
  97                         else {
  98                                 string path;
  99                                 it = parameters.find("group");
 100                                 //user has given a template file
 101                                 if(it != parameters.end()){
 102                                         path = m->hasPath(it->second);
 103                                         //if the user has not given a path then, add inputdir. else leave path alone.
 104                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 105                                 }
 106
 107                                 it = parameters.find("fasta");
 108                                 //user has given a template file
 109                                 if(it != parameters.end()){
 110                                         path = m->hasPath(it->second);
 111                                         //if the user has not given a path then, add inputdir. else leave path alone.
 112                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 113                                 }
 114
 115                                 it = parameters.find("name");
 116                                 //user has given a template file
 117                                 if(it != parameters.end()){
 118                                         path = m->hasPath(it->second);
 119                                         //if the user has not given a path then, add inputdir. else leave path alone.
 120                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 121                                 }
 122
 123                         }
 124
 125
 126                         namefile = validParameter.validFile(parameters, "name", true);
 127                         if (namefile == "not open") { abort = true; }
 128                         else if (namefile == "not found") { namefile = ""; }
 129
 130                         fastafile = validParameter.validFile(parameters, "fasta", true);
 131                         if (fastafile == "not open") { abort = true; }
 132                         else if (fastafile == "not found") {
 133                                 fastafile = m->getFastaFile();
 134                                 if (fastafile != "") { m->mothurOut("Using " + fastafile + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 135                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 136                         }
 137
 138                         groupfile = validParameter.validFile(parameters, "group", true);
 139                         if (groupfile == "not open") {  groupfile = ""; abort = true; }
 140                         else if (groupfile == "not found") {
 141                                 groupfile = m->getGroupFile();
 142                                 if (groupfile != "") { m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 143                                 else {  m->mothurOut("You have no current groupfile and the group parameter is required."); m->mothurOutEndLine(); abort = true; }
 144                         }
 145
 146                         groups = validParameter.validFile(parameters, "groups", false);
 147                         if (groups == "not found") { groups = ""; }
 148                         else { m->splitAtDash(groups, Groups);  }
 149
 150                         //if the user changes the output directory command factory will send this info to us in the output parameter
 151                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(groupfile);      }
 152                 }
 153
 154         }
 155         catch(exception& e) {
 156                 m->errorOut(e, "SplitGroupCommand", "SplitAbundCommand");
 157                 exit(1);
 158         }
 159 }
 160 //**********************************************************************************************************************
 161 int SplitGroupCommand::execute(){
 162         try {
 163
 164                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 165
 166                 groupMap = new GroupMap(groupfile);
 167                 groupMap->readMap();
 168
 169                 SharedUtil util;  util.setGroups(Groups, groupMap->namesOfGroups);
 170
 171                 if (namefile != "") {  readNames();  }
 172                 splitFasta();
 173
 174                 delete groupMap;
 175
 176                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str()); } return 0; }
 177
 178                 string current = "";
 179                 itTypes = outputTypes.find("fasta");
 180                 if (itTypes != outputTypes.end()) {
 181                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 182                 }
 183
 184                 itTypes = outputTypes.find("name");
 185                 if (itTypes != outputTypes.end()) {
 186                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 187                 }
 188
 189                 m->mothurOutEndLine();
 190                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 191                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 192                 m->mothurOutEndLine();
 193
 194                 return 0;
 195         }
 196         catch(exception& e) {
 197                 m->errorOut(e, "SplitGroupCommand", "execute");
 198                 exit(1);
 199         }
 200 }
 201 /**********************************************************************************************************************/
 202 int SplitGroupCommand::readNames() {
 203         try {
 204                 //open input file
 205                 ifstream in;
 206                 m->openInputFile(namefile, in);
 207
 208                 while (!in.eof()) {
 209                         if (m->control_pressed) { break; }
 210
 211                         string firstCol, secondCol;
 212                         in >> firstCol >> secondCol; m->gobble(in);
 213
 214                         vector<string> names;
 215                         m->splitAtComma(secondCol, names);
 216
 217                         nameMap[firstCol] = names;
 218                 }
 219                 in.close();
 220
 221                 return 0;
 222
 223         }
 224         catch(exception& e) {
 225                 m->errorOut(e, "SplitGroupCommand", "readNames");
 226                 exit(1);
 227         }
 228 }
 229
 230 /**********************************************************************************************************************/
 231 int SplitGroupCommand::splitFasta() {
 232         try {
 233
 234                 string filerootFasta =  outputDir + m->getRootName(m->getSimpleName(fastafile));
 235                 string filerootName =  outputDir + m->getRootName(m->getSimpleName(namefile));
 236                 ofstream* temp;
 237                 ofstream* temp2;
 238                 map<string, ofstream*> filehandles;
 239                 map<string, ofstream*>::iterator it3;
 240
 241                 for (int i=0; i<Groups.size(); i++) {
 242                         temp = new ofstream;
 243                         filehandles[Groups[i]+"fasta"] = temp;
 244                         m->openOutputFile(filerootFasta + Groups[i] + ".fasta", *(filehandles[Groups[i]+"fasta"]));
 245                         outputNames.push_back(filerootFasta + Groups[i] + ".fasta"); outputTypes["fasta"].push_back(filerootFasta + Groups[i] + ".fasta");
 246
 247                         if (namefile != "") {
 248                                 temp2 = new ofstream;
 249                                 filehandles[Groups[i]+"name"] = temp2;
 250                                 m->openOutputFile(filerootName + Groups[i] + ".names", *(filehandles[Groups[i]+"name"]));
 251                                 outputNames.push_back(filerootName + Groups[i] + ".names"); outputTypes["name"].push_back(filerootFasta + Groups[i] + ".names");
 252                         }
 253                 }
 254
 255                 //open input file
 256                 ifstream in;
 257                 m->openInputFile(fastafile, in);
 258
 259                 while (!in.eof()) {
 260                         if (m->control_pressed) { break; }
 261
 262                         Sequence seq(in); m->gobble(in);
 263
 264                         if (seq.getName() != "") {
 265
 266                                 itNames = nameMap.find(seq.getName());
 267
 268                                 if (itNames == nameMap.end()) {
 269                                         if (namefile != "") {  m->mothurOut(seq.getName() + " is not in your namesfile, ignoring."); m->mothurOutEndLine();  }
 270                                         else { //no names file given
 271                                                 string group = groupMap->getGroup(seq.getName());
 272
 273                                                 if (m->inUsersGroups(group, Groups)) { //only add if this is in a group we want
 274                                                         seq.printSequence(*(filehandles[group+"fasta"]));
 275                                                 }else if(group == "not found") {
 276                                                         m->mothurOut(seq.getName() + " is not in your groupfile. Ignoring."); m->mothurOutEndLine();
 277                                                 }
 278                                         }
 279                                 }else{
 280                                         vector<string> names = itNames->second;
 281                                         map<string, string> group2Names;
 282                                         map<string, string>::iterator it;
 283
 284                                         for (int i = 0; i < names.size(); i++) {  //build strings for new group names file, will select rep below
 285                                                 string group = groupMap->getGroup(names[i]);
 286
 287                                                 if (m->inUsersGroups(group, Groups)) { //only add if this is in a group we want
 288                                                         it = group2Names.find(group);
 289                                                         if (it == group2Names.end()) {
 290                                                                 group2Names[group] = names[i] + ",";
 291                                                         }else{
 292                                                                 group2Names[group] += names[i] + ",";
 293                                                         }
 294                                                 }else if(group == "not found") {
 295                                                         m->mothurOut(names[i] + " is not in your groupfile. Ignoring."); m->mothurOutEndLine();
 296                                                 }
 297                                         }
 298
 299                                         for (map<string, string>::iterator itGroups = group2Names.begin(); itGroups != group2Names.end(); itGroups++) {
 300                                                 //edit names string, by grabbing the first guy as the rep and removing the last comma
 301                                                 string newNames = itGroups->second;
 302                                                 newNames = newNames.substr(0, newNames.length()-1);
 303                                                 string repName = "";
 304
 305                                                 int pos = newNames.find_first_of(',');
 306                                                 if (pos == newNames.npos) { //only one sequence so it represents itself
 307                                                         repName = newNames;
 308                                                 }else{
 309                                                         repName = newNames.substr(0, pos);
 310                                                 }
 311
 312                                                 *(filehandles[itGroups->first+"name"]) << repName << '\t' << newNames << endl;
 313                                                 *(filehandles[itGroups->first+"fasta"]) << ">" << repName << endl << seq.getAligned() << endl;
 314                                         }
 315                                 }
 316                         }
 317                 }
 318
 319                 in.close();
 320
 321                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 322                         (*(filehandles[it3->first])).close();
 323                         delete it3->second;
 324                 }
 325
 326                 vector<string> newOutputNames;
 327                 //remove blank files
 328                 for (int i = 0; i < outputNames.size(); i++) {
 329                         if (m->isBlank(outputNames[i])) {
 330                                 remove(outputNames[i].c_str());
 331                         }else { newOutputNames.push_back(outputNames[i]); }
 332                 }
 333                 outputNames = newOutputNames;
 334
 335                 return 0;
 336
 337         }
 338         catch(exception& e) {
 339                 m->errorOut(e, "SplitGroupCommand", "splitFasta");
 340                 exit(1);
 341         }
 342 }
 343 /**********************************************************************************************************************/
 344