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::getValidParameters(){
  15         try {
  16                 string Array[] =  {"name","group","groups","fasta","outputdir","inputdir"};
  17                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  18                 return myArray;
  19         }
  20         catch(exception& e) {
  21                 m->errorOut(e, "SplitGroupCommand", "getValidParameters");
  22                 exit(1);
  23         }
  24 }
  25 //**********************************************************************************************************************
  26 vector<string> SplitGroupCommand::getRequiredParameters(){
  27         try {
  28                 string Array[] =  {"fasta","group"};
  29                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  30                 return myArray;
  31         }
  32         catch(exception& e) {
  33                 m->errorOut(e, "SplitGroupCommand", "getRequiredParameters");
  34                 exit(1);
  35         }
  36 }
  37 //**********************************************************************************************************************
  38 vector<string> SplitGroupCommand::getRequiredFiles(){
  39         try {
  40                 vector<string> myArray;
  41                 return myArray;
  42         }
  43         catch(exception& e) {
  44                 m->errorOut(e, "SplitGroupCommand", "getRequiredFiles");
  45                 exit(1);
  46         }
  47 }
  48 //**********************************************************************************************************************
  49 SplitGroupCommand::SplitGroupCommand(){
  50         try {
  51                 abort = true; calledHelp = true;
  52                 vector<string> tempOutNames;
  53                 outputTypes["fasta"] = tempOutNames;
  54                 outputTypes["name"] = tempOutNames;
  55         }
  56         catch(exception& e) {
  57                 m->errorOut(e, "SplitGroupCommand", "SplitGroupCommand");
  58                 exit(1);
  59         }
  60 }
  61 //**********************************************************************************************************************
  62 SplitGroupCommand::SplitGroupCommand(string option)  {
  63         try {
  64                 abort = false; calledHelp = false;
  65
  66                 //allow user to run help
  67                 if(option == "help") { help(); abort = true; calledHelp = true; }
  68
  69                 else {
  70                         //valid paramters for this command
  71                         string Array[] =  {"name","group","groups","fasta","outputdir","inputdir"};
  72                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  73
  74                         OptionParser parser(option);
  75                         map<string, string> parameters = parser.getParameters();
  76
  77                         ValidParameters validParameter;
  78                         map<string, string>::iterator it;
  79
  80                         //check to make sure all parameters are valid for command
  81                         for (it = parameters.begin(); it != parameters.end(); it++) {
  82                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  83                         }
  84
  85                         //initialize outputTypes
  86                         vector<string> tempOutNames;
  87                         outputTypes["fasta"] = tempOutNames;
  88                         outputTypes["name"] = tempOutNames;
  89
  90                         //if the user changes the input directory command factory will send this info to us in the output parameter
  91                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  92                         if (inputDir == "not found"){   inputDir = "";          }
  93                         else {
  94                                 string path;
  95                                 it = parameters.find("group");
  96                                 //user has given a template file
  97                                 if(it != parameters.end()){
  98                                         path = m->hasPath(it->second);
  99                                         //if the user has not given a path then, add inputdir. else leave path alone.
 100                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 101                                 }
 102
 103                                 it = parameters.find("fasta");
 104                                 //user has given a template file
 105                                 if(it != parameters.end()){
 106                                         path = m->hasPath(it->second);
 107                                         //if the user has not given a path then, add inputdir. else leave path alone.
 108                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 109                                 }
 110
 111                                 it = parameters.find("name");
 112                                 //user has given a template file
 113                                 if(it != parameters.end()){
 114                                         path = m->hasPath(it->second);
 115                                         //if the user has not given a path then, add inputdir. else leave path alone.
 116                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 117                                 }
 118
 119                         }
 120
 121
 122                         namefile = validParameter.validFile(parameters, "name", true);
 123                         if (namefile == "not open") { abort = true; }
 124                         else if (namefile == "not found") { namefile = ""; }
 125
 126                         fastafile = validParameter.validFile(parameters, "fasta", true);
 127                         if (fastafile == "not open") { abort = true; }
 128                         else if (fastafile == "not found") { fastafile = ""; m->mothurOut("fasta is a required parameter for the split.group command. "); m->mothurOutEndLine(); abort = true;  }
 129
 130                         groupfile = validParameter.validFile(parameters, "group", true);
 131                         if (groupfile == "not open") {  groupfile = ""; abort = true; }
 132                         else if (groupfile == "not found") { groupfile = ""; m->mothurOut("group is a required parameter for the split.group command. "); m->mothurOutEndLine(); abort = true; }
 133
 134                         groups = validParameter.validFile(parameters, "groups", false);
 135                         if (groups == "not found") { groups = ""; }
 136                         else { m->splitAtDash(groups, Groups);  }
 137
 138                         //if the user changes the output directory command factory will send this info to us in the output parameter
 139                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(groupfile);      }
 140                 }
 141
 142         }
 143         catch(exception& e) {
 144                 m->errorOut(e, "SplitGroupCommand", "SplitAbundCommand");
 145                 exit(1);
 146         }
 147 }
 148 //**********************************************************************************************************************
 149 void SplitGroupCommand::help(){
 150         try {
 151                 m->mothurOut("The split.group command reads a group file, and parses your fasta and names files by groups. \n");
 152                 m->mothurOut("The split.group command parameters are fasta, name, group and groups.\n");
 153                 m->mothurOut("The fasta and group parameters are required.\n");
 154                 m->mothurOut("The groups parameter allows you to select groups to create files for.  \n");
 155                 m->mothurOut("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");
 156                 m->mothurOut("If you want .fasta and .names files for all groups, set groups=all.  \n");
 157                 m->mothurOut("The split.group command should be used in the following format: split.group(fasta=yourFasta, group=yourGroupFile).\n");
 158                 m->mothurOut("Example: split.group(fasta=abrecovery.fasta, group=abrecovery.groups).\n");
 159                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
 160
 161         }
 162         catch(exception& e) {
 163                 m->errorOut(e, "SplitGroupCommand", "help");
 164                 exit(1);
 165         }
 166 }
 167 //**********************************************************************************************************************
 168 SplitGroupCommand::~SplitGroupCommand(){ }
 169 //**********************************************************************************************************************
 170 int SplitGroupCommand::execute(){
 171         try {
 172
 173                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 174
 175                 groupMap = new GroupMap(groupfile);
 176                 groupMap->readMap();
 177
 178                 SharedUtil util;  util.setGroups(Groups, groupMap->namesOfGroups);
 179
 180                 if (namefile != "") {  readNames();  }
 181                 splitFasta();
 182
 183                 delete groupMap;
 184
 185                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str()); } return 0; }
 186
 187                 string current = "";
 188                 itTypes = outputTypes.find("fasta");
 189                 if (itTypes != outputTypes.end()) {
 190                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 191                 }
 192
 193                 itTypes = outputTypes.find("name");
 194                 if (itTypes != outputTypes.end()) {
 195                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 196                 }
 197
 198                 m->mothurOutEndLine();
 199                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 200                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 201                 m->mothurOutEndLine();
 202
 203                 return 0;
 204         }
 205         catch(exception& e) {
 206                 m->errorOut(e, "SplitGroupCommand", "execute");
 207                 exit(1);
 208         }
 209 }
 210 /**********************************************************************************************************************/
 211 int SplitGroupCommand::readNames() {
 212         try {
 213                 //open input file
 214                 ifstream in;
 215                 m->openInputFile(namefile, in);
 216
 217                 while (!in.eof()) {
 218                         if (m->control_pressed) { break; }
 219
 220                         string firstCol, secondCol;
 221                         in >> firstCol >> secondCol; m->gobble(in);
 222
 223                         vector<string> names;
 224                         m->splitAtComma(secondCol, names);
 225
 226                         nameMap[firstCol] = names;
 227                 }
 228                 in.close();
 229
 230                 return 0;
 231
 232         }
 233         catch(exception& e) {
 234                 m->errorOut(e, "SplitGroupCommand", "readNames");
 235                 exit(1);
 236         }
 237 }
 238
 239 /**********************************************************************************************************************/
 240 int SplitGroupCommand::splitFasta() {
 241         try {
 242
 243                 string filerootFasta =  outputDir + m->getRootName(m->getSimpleName(fastafile));
 244                 string filerootName =  outputDir + m->getRootName(m->getSimpleName(namefile));
 245                 ofstream* temp;
 246                 ofstream* temp2;
 247                 map<string, ofstream*> filehandles;
 248                 map<string, ofstream*>::iterator it3;
 249
 250                 for (int i=0; i<Groups.size(); i++) {
 251                         temp = new ofstream;
 252                         filehandles[Groups[i]+"fasta"] = temp;
 253                         m->openOutputFile(filerootFasta + Groups[i] + ".fasta", *(filehandles[Groups[i]+"fasta"]));
 254                         outputNames.push_back(filerootFasta + Groups[i] + ".fasta"); outputTypes["fasta"].push_back(filerootFasta + Groups[i] + ".fasta");
 255
 256                         if (namefile != "") {
 257                                 temp2 = new ofstream;
 258                                 filehandles[Groups[i]+"name"] = temp2;
 259                                 m->openOutputFile(filerootName + Groups[i] + ".names", *(filehandles[Groups[i]+"name"]));
 260                                 outputNames.push_back(filerootName + Groups[i] + ".names"); outputTypes["name"].push_back(filerootFasta + Groups[i] + ".names");
 261                         }
 262                 }
 263
 264                 //open input file
 265                 ifstream in;
 266                 m->openInputFile(fastafile, in);
 267
 268                 while (!in.eof()) {
 269                         if (m->control_pressed) { break; }
 270
 271                         Sequence seq(in); m->gobble(in);
 272
 273                         if (seq.getName() != "") {
 274
 275                                 itNames = nameMap.find(seq.getName());
 276
 277                                 if (itNames == nameMap.end()) {
 278                                         if (namefile != "") {  m->mothurOut(seq.getName() + " is not in your namesfile, ignoring."); m->mothurOutEndLine();  }
 279                                         else { //no names file given
 280                                                 string group = groupMap->getGroup(seq.getName());
 281
 282                                                 if (m->inUsersGroups(group, Groups)) { //only add if this is in a group we want
 283                                                         seq.printSequence(*(filehandles[group+"fasta"]));
 284                                                 }else if(group == "not found") {
 285                                                         m->mothurOut(seq.getName() + " is not in your groupfile. Ignoring."); m->mothurOutEndLine();
 286                                                 }
 287                                         }
 288                                 }else{
 289                                         vector<string> names = itNames->second;
 290                                         map<string, string> group2Names;
 291                                         map<string, string>::iterator it;
 292
 293                                         for (int i = 0; i < names.size(); i++) {  //build strings for new group names file, will select rep below
 294                                                 string group = groupMap->getGroup(names[i]);
 295
 296                                                 if (m->inUsersGroups(group, Groups)) { //only add if this is in a group we want
 297                                                         it = group2Names.find(group);
 298                                                         if (it == group2Names.end()) {
 299                                                                 group2Names[group] = names[i] + ",";
 300                                                         }else{
 301                                                                 group2Names[group] += names[i] + ",";
 302                                                         }
 303                                                 }else if(group == "not found") {
 304                                                         m->mothurOut(names[i] + " is not in your groupfile. Ignoring."); m->mothurOutEndLine();
 305                                                 }
 306                                         }
 307
 308                                         for (map<string, string>::iterator itGroups = group2Names.begin(); itGroups != group2Names.end(); itGroups++) {
 309                                                 //edit names string, by grabbing the first guy as the rep and removing the last comma
 310                                                 string newNames = itGroups->second;
 311                                                 newNames = newNames.substr(0, newNames.length()-1);
 312                                                 string repName = "";
 313
 314                                                 int pos = newNames.find_first_of(',');
 315                                                 if (pos == newNames.npos) { //only one sequence so it represents itself
 316                                                         repName = newNames;
 317                                                 }else{
 318                                                         repName = newNames.substr(0, pos);
 319                                                 }
 320
 321                                                 *(filehandles[itGroups->first+"name"]) << repName << '\t' << newNames << endl;
 322                                                 *(filehandles[itGroups->first+"fasta"]) << ">" << repName << endl << seq.getAligned() << endl;
 323                                         }
 324                                 }
 325                         }
 326                 }
 327
 328                 in.close();
 329
 330                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 331                         (*(filehandles[it3->first])).close();
 332                         delete it3->second;
 333                 }
 334
 335                 vector<string> newOutputNames;
 336                 //remove blank files
 337                 for (int i = 0; i < outputNames.size(); i++) {
 338                         if (m->isBlank(outputNames[i])) {
 339                                 remove(outputNames[i].c_str());
 340                         }else { newOutputNames.push_back(outputNames[i]); }
 341                 }
 342                 outputNames = newOutputNames;
 343
 344                 return 0;
 345
 346         }
 347         catch(exception& e) {
 348                 m->errorOut(e, "SplitGroupCommand", "splitFasta");
 349                 exit(1);
 350         }
 351 }
 352 /**********************************************************************************************************************/
 353