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