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                         else { m->setNameFile(namefile); }
 130
 131                         fastafile = validParameter.validFile(parameters, "fasta", true);
 132                         if (fastafile == "not open") { abort = true; }
 133                         else if (fastafile == "not found") {
 134                                 fastafile = m->getFastaFile();
 135                                 if (fastafile != "") { m->mothurOut("Using " + fastafile + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 136                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 137                         }else { m->setFastaFile(fastafile); }
 138
 139                         groupfile = validParameter.validFile(parameters, "group", true);
 140                         if (groupfile == "not open") {  groupfile = ""; abort = true; }
 141                         else if (groupfile == "not found") {
 142                                 groupfile = m->getGroupFile();
 143                                 if (groupfile != "") { m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 144                                 else {  m->mothurOut("You have no current groupfile and the group parameter is required."); m->mothurOutEndLine(); abort = true; }
 145                         }else {  m->setGroupFile(groupfile); }
 146
 147                         groups = validParameter.validFile(parameters, "groups", false);
 148                         if (groups == "not found") { groups = ""; }
 149                         else { m->splitAtDash(groups, Groups);  }
 150
 151                         //if the user changes the output directory command factory will send this info to us in the output parameter
 152                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(groupfile);      }
 153                 }
 154
 155         }
 156         catch(exception& e) {
 157                 m->errorOut(e, "SplitGroupCommand", "SplitAbundCommand");
 158                 exit(1);
 159         }
 160 }
 161 //**********************************************************************************************************************
 162 int SplitGroupCommand::execute(){
 163         try {
 164
 165                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 166
 167                 groupMap = new GroupMap(groupfile);
 168                 groupMap->readMap();
 169
 170                 SharedUtil util;  util.setGroups(Groups, groupMap->namesOfGroups);
 171
 172                 if (namefile != "") {  readNames();  }
 173                 splitFasta();
 174
 175                 delete groupMap;
 176
 177                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);        } return 0; }
 178
 179                 string current = "";
 180                 itTypes = outputTypes.find("fasta");
 181                 if (itTypes != outputTypes.end()) {
 182                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 183                 }
 184
 185                 itTypes = outputTypes.find("name");
 186                 if (itTypes != outputTypes.end()) {
 187                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 188                 }
 189
 190                 m->mothurOutEndLine();
 191                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 192                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 193                 m->mothurOutEndLine();
 194
 195                 return 0;
 196         }
 197         catch(exception& e) {
 198                 m->errorOut(e, "SplitGroupCommand", "execute");
 199                 exit(1);
 200         }
 201 }
 202 /**********************************************************************************************************************/
 203 int SplitGroupCommand::readNames() {
 204         try {
 205                 //open input file
 206                 ifstream in;
 207                 m->openInputFile(namefile, in);
 208
 209                 while (!in.eof()) {
 210                         if (m->control_pressed) { break; }
 211
 212                         string firstCol, secondCol;
 213                         in >> firstCol >> secondCol; m->gobble(in);
 214
 215                         vector<string> names;
 216                         m->splitAtComma(secondCol, names);
 217
 218                         nameMap[firstCol] = names;
 219                 }
 220                 in.close();
 221
 222                 return 0;
 223
 224         }
 225         catch(exception& e) {
 226                 m->errorOut(e, "SplitGroupCommand", "readNames");
 227                 exit(1);
 228         }
 229 }
 230
 231 /**********************************************************************************************************************/
 232 int SplitGroupCommand::splitFasta() {
 233         try {
 234
 235                 string filerootFasta =  outputDir + m->getRootName(m->getSimpleName(fastafile));
 236                 string filerootName =  outputDir + m->getRootName(m->getSimpleName(namefile));
 237                 ofstream* temp;
 238                 ofstream* temp2;
 239                 map<string, ofstream*> filehandles;
 240                 map<string, ofstream*>::iterator it3;
 241
 242                 for (int i=0; i<Groups.size(); i++) {
 243                         temp = new ofstream;
 244                         filehandles[Groups[i]+"fasta"] = temp;
 245                         m->openOutputFile(filerootFasta + Groups[i] + ".fasta", *(filehandles[Groups[i]+"fasta"]));
 246                         outputNames.push_back(filerootFasta + Groups[i] + ".fasta"); outputTypes["fasta"].push_back(filerootFasta + Groups[i] + ".fasta");
 247
 248                         if (namefile != "") {
 249                                 temp2 = new ofstream;
 250                                 filehandles[Groups[i]+"name"] = temp2;
 251                                 m->openOutputFile(filerootName + Groups[i] + ".names", *(filehandles[Groups[i]+"name"]));
 252                                 outputNames.push_back(filerootName + Groups[i] + ".names"); outputTypes["name"].push_back(filerootFasta + Groups[i] + ".names");
 253                         }
 254                 }
 255
 256                 //open input file
 257                 ifstream in;
 258                 m->openInputFile(fastafile, in);
 259
 260                 while (!in.eof()) {
 261                         if (m->control_pressed) { break; }
 262
 263                         Sequence seq(in); m->gobble(in);
 264
 265                         if (seq.getName() != "") {
 266
 267                                 itNames = nameMap.find(seq.getName());
 268
 269                                 if (itNames == nameMap.end()) {
 270                                         if (namefile != "") {  m->mothurOut(seq.getName() + " is not in your namesfile, ignoring."); m->mothurOutEndLine();  }
 271                                         else { //no names file given
 272                                                 string group = groupMap->getGroup(seq.getName());
 273
 274                                                 if (m->inUsersGroups(group, Groups)) { //only add if this is in a group we want
 275                                                         seq.printSequence(*(filehandles[group+"fasta"]));
 276                                                 }else if(group == "not found") {
 277                                                         m->mothurOut(seq.getName() + " is not in your groupfile. Ignoring."); m->mothurOutEndLine();
 278                                                 }
 279                                         }
 280                                 }else{
 281                                         vector<string> names = itNames->second;
 282                                         map<string, string> group2Names;
 283                                         map<string, string>::iterator it;
 284
 285                                         for (int i = 0; i < names.size(); i++) {  //build strings for new group names file, will select rep below
 286                                                 string group = groupMap->getGroup(names[i]);
 287
 288                                                 if (m->inUsersGroups(group, Groups)) { //only add if this is in a group we want
 289                                                         it = group2Names.find(group);
 290                                                         if (it == group2Names.end()) {
 291                                                                 group2Names[group] = names[i] + ",";
 292                                                         }else{
 293                                                                 group2Names[group] += names[i] + ",";
 294                                                         }
 295                                                 }else if(group == "not found") {
 296                                                         m->mothurOut(names[i] + " is not in your groupfile. Ignoring."); m->mothurOutEndLine();
 297                                                 }
 298                                         }
 299
 300                                         for (map<string, string>::iterator itGroups = group2Names.begin(); itGroups != group2Names.end(); itGroups++) {
 301                                                 //edit names string, by grabbing the first guy as the rep and removing the last comma
 302                                                 string newNames = itGroups->second;
 303                                                 newNames = newNames.substr(0, newNames.length()-1);
 304                                                 string repName = "";
 305
 306                                                 int pos = newNames.find_first_of(',');
 307                                                 if (pos == newNames.npos) { //only one sequence so it represents itself
 308                                                         repName = newNames;
 309                                                 }else{
 310                                                         repName = newNames.substr(0, pos);
 311                                                 }
 312
 313                                                 *(filehandles[itGroups->first+"name"]) << repName << '\t' << newNames << endl;
 314                                                 *(filehandles[itGroups->first+"fasta"]) << ">" << repName << endl << seq.getAligned() << endl;
 315                                         }
 316                                 }
 317                         }
 318                 }
 319
 320                 in.close();
 321
 322                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 323                         (*(filehandles[it3->first])).close();
 324                         delete it3->second;
 325                 }
 326
 327                 vector<string> newOutputNames;
 328                 //remove blank files
 329                 for (int i = 0; i < outputNames.size(); i++) {
 330                         if (m->isBlank(outputNames[i])) {
 331                                 m->mothurRemove(outputNames[i]);
 332                         }else { newOutputNames.push_back(outputNames[i]); }
 333                 }
 334                 outputNames = newOutputNames;
 335
 336                 return 0;
 337
 338         }
 339         catch(exception& e) {
 340                 m->errorOut(e, "SplitGroupCommand", "splitFasta");
 341                 exit(1);
 342         }
 343 }
 344 /**********************************************************************************************************************/
 345