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