getsharedotucommand.cpp

   1 /*
   2  *  getsharedotucommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 9/22/09.
   6  *  Copyright 2009 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "getsharedotucommand.h"
  11
  12 //**********************************************************************************************************************
  13
  14 GetSharedOTUCommand::GetSharedOTUCommand(string option){
  15         try {
  16
  17                 globaldata = GlobalData::getInstance();
  18                 abort = false;
  19                 allLines = 1;
  20                 labels.clear();
  21
  22                 //allow user to run help
  23                 if(option == "help") { help(); abort = true; }
  24
  25                 else {
  26                         //valid paramters for this command
  27                         string Array[] =  {"label","groups","fasta","list","group","output"};
  28                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  29
  30                         OptionParser parser(option);
  31                         map<string,string> parameters = parser.getParameters();
  32
  33                         ValidParameters validParameter;
  34
  35                         //check to make sure all parameters are valid for command
  36                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  37                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  38                         }
  39
  40                         //check for required parameters
  41                         listfile = validParameter.validFile(parameters, "list", true);
  42                         if (listfile == "not open") { abort = true; }
  43                         else if (listfile == "not found") { listfile = ""; }
  44                         else {  globaldata->setListFile(listfile);  globaldata->setFormat("list");      }
  45
  46                         groupfile = validParameter.validFile(parameters, "group", true);
  47                         if (groupfile == "not open") { abort = true; }
  48                         else if (groupfile == "not found") { groupfile = ""; }
  49
  50                         if ((listfile == "") || (groupfile == "")) { mothurOut("The list and group parameters are required."); mothurOutEndLine(); abort = true; }
  51
  52                         //check for optional parameter and set defaults
  53                         // ...at some point should added some additional type checking...
  54                         label = validParameter.validFile(parameters, "label", false);
  55                         if (label == "not found") { label = ""; }
  56                         else {
  57                                 if(label != "all") {  splitAtDash(label, labels);  allLines = 0;  }
  58                                 else { allLines = 1;  }
  59                         }
  60
  61                         output = validParameter.validFile(parameters, "output", false);
  62                         if (output == "not found") { output = ""; }
  63
  64                         groups = validParameter.validFile(parameters, "groups", false);
  65                         if (groups == "not found") { groups = ""; }
  66                         else {
  67                                 splitAtDash(groups, Groups);
  68                                 globaldata->Groups = Groups;
  69                         }
  70
  71                         fastafile = validParameter.validFile(parameters, "fasta", true);
  72                         if (fastafile == "not open") { abort = true; }
  73                         else if (fastafile == "not found") {  fastafile = "";  }
  74
  75                 }
  76
  77         }
  78         catch(exception& e) {
  79                 errorOut(e, "GetSharedOTUCommand", "GetSharedOTUCommand");
  80                 exit(1);
  81         }
  82 }
  83 //**********************************************************************************************************************
  84
  85 void GetSharedOTUCommand::help(){
  86         try {
  87                 mothurOut("The get.sharedotu command parameters are list, group, label, groups, output and fasta.  The list and group parameters are required.\n");
  88                 mothurOut("The label parameter allows you to select what distance levels you would like output files for, and are separated by dashes.\n");
  89                 mothurOut("The groups parameter allows you to select groups you would like to know the shared info for, and are separated by dashes.\n");
  90                 mothurOut("The fasta parameter allows you to input a fasta file and outputs a fasta file for each distance level containing only the sequences that are in OTUs shared by the groups specified.\n");
  91                 mothurOut("The output parameter allows you to output the list of names without the group and bin number added. \n");
  92                 mothurOut("With this option you can use the names file as an input in get.seqs and remove.seqs commands. To do this enter output=accnos. \n");
  93                 mothurOut("The get.sharedotu command outputs a .names file for each distance level containing a list of sequences in the OTUs shared by the groups specified.\n");
  94                 mothurOut("The get.sharedotu command should be in the following format: get.sabund(label=yourLabels, groups=yourGroups, fasta=yourFastafile, output=yourOutput).\n");
  95                 mothurOut("Example get.sharedotu(list=amazon.fn.list, label=unique-0.01, group=forest-pasture, fasta=amazon.fasta, output=accnos).\n");
  96                 mothurOut("The output to the screen is the distance and the number of otus at that distance for the groups you specified.\n");
  97                 mothurOut("The default value for label is all labels in your inputfile. The default for groups is all groups in your file.\n");
  98                 mothurOut("Note: No spaces between parameter labels (i.e. label), '=' and parameters (i.e.yourLabel).\n\n");
  99         }
 100         catch(exception& e) {
 101                 errorOut(e, "GetSharedOTUCommand", "help");
 102                 exit(1);
 103         }
 104 }
 105
 106 //**********************************************************************************************************************
 107
 108 GetSharedOTUCommand::~GetSharedOTUCommand(){}
 109
 110 //**********************************************************************************************************************
 111
 112 int GetSharedOTUCommand::execute(){
 113         try {
 114
 115                 if (abort == true) { return 0; }
 116
 117                 groupMap = new GroupMap(groupfile);
 118                 groupMap->readMap();
 119                 globaldata->gGroupmap = groupMap;
 120
 121                 if (Groups.size() == 0) {
 122                         Groups = groupMap->namesOfGroups;
 123                 }
 124
 125                 //put groups in map to find easier
 126                 for(int i = 0; i < Groups.size(); i++) {
 127                         groupFinder[Groups[i]] = Groups[i];
 128                 }
 129
 130                 if (fastafile != "") {
 131                         ifstream inFasta;
 132                         openInputFile(fastafile, inFasta);
 133
 134                         while(!inFasta.eof()) {
 135                                 Sequence seq(inFasta);
 136                                 seqs.push_back(seq);
 137                         }
 138                         inFasta.close();
 139                 }
 140
 141                 ListVector* lastlist = NULL;
 142                 string lastLabel = "";
 143
 144                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 145                 set<string> processedLabels;
 146                 set<string> userLabels = labels;
 147
 148                 ifstream in;
 149                 openInputFile(listfile, in);
 150
 151                 //as long as you are not at the end of the file or done wih the lines you want
 152                 while((!in.eof()) && ((allLines == 1) || (userLabels.size() != 0))) {
 153
 154                         list = new ListVector(in);
 155
 156                         if(allLines == 1 || labels.count(list->getLabel()) == 1){
 157                                 mothurOut(list->getLabel());
 158                                 process(list);
 159
 160                                 processedLabels.insert(list->getLabel());
 161                                 userLabels.erase(list->getLabel());
 162                         }
 163
 164                         if ((anyLabelsToProcess(list->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 165                                         string saveLabel = list->getLabel();
 166
 167                                         mothurOut(lastlist->getLabel());
 168                                         process(lastlist);
 169
 170                                         processedLabels.insert(lastlist->getLabel());
 171                                         userLabels.erase(lastlist->getLabel());
 172
 173                                         //restore real lastlabel to save below
 174                                         list->setLabel(saveLabel);
 175                         }
 176
 177                         lastLabel = list->getLabel();
 178
 179                         if (lastlist != NULL) {         delete lastlist;        }
 180                         lastlist = list;
 181                 }
 182
 183                 in.close();
 184
 185                 //output error messages about any remaining user labels
 186                 set<string>::iterator it;
 187                 bool needToRun = false;
 188                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 189                         mothurOut("Your file does not include the label " + *it);
 190                         if (processedLabels.count(lastLabel) != 1) {
 191                                 mothurOut(". I will use " + lastLabel + "."); mothurOutEndLine();
 192                                 needToRun = true;
 193                         }else {
 194                                 mothurOut(". Please refer to " + lastLabel + "."); mothurOutEndLine();
 195                         }
 196                 }
 197
 198                 //run last label if you need to
 199                 if (needToRun == true)  {
 200                                 mothurOut(lastlist->getLabel());
 201                                 process(lastlist);
 202
 203                                 processedLabels.insert(lastlist->getLabel());
 204                                 userLabels.erase(lastlist->getLabel());
 205                 }
 206
 207
 208                 //reset groups parameter
 209                 globaldata->Groups.clear();
 210
 211                 if (lastlist != NULL) {         delete lastlist;        }
 212                 return 0;
 213         }
 214
 215         catch(exception& e) {
 216                 errorOut(e, "GetSharedOTUCommand", "execute");
 217                 exit(1);
 218         }
 219 }
 220 /***********************************************************/
 221 void GetSharedOTUCommand::process(ListVector* shared) {
 222         try {
 223
 224                 map<string, string> fastaMap;
 225
 226                 ofstream outNames;
 227                 string outputFileNames = getRootName(listfile) + shared->getLabel() + ".names";
 228                 openOutputFile(outputFileNames, outNames);
 229
 230                 bool wroteSomething = false;
 231                 int num = 0;
 232
 233                 //go through each bin, find out if shared
 234                 for (int i = 0; i < shared->getNumBins(); i++) {
 235
 236                         bool sharedByAll = true;
 237
 238                         map<string, int> atLeastOne;
 239                         for (int m = 0; m < Groups.size(); m++) {
 240                                 atLeastOne[Groups[m]] = 0;
 241                         }
 242
 243                         vector<string> namesOfSeqsInThisBin;
 244
 245                         string names = shared->get(i);
 246                         while ((names.find_first_of(',') != -1) && sharedByAll) {
 247                                 string name = names.substr(0,names.find_first_of(','));
 248                                 names = names.substr(names.find_first_of(',')+1, names.length());
 249
 250                                 //find group
 251                                 string seqGroup = groupMap->getGroup(name);
 252                                 if (output != "accnos") {
 253                                         namesOfSeqsInThisBin.push_back((name + "|" + seqGroup + "|" + toString(i+1)));
 254                                 }else {  namesOfSeqsInThisBin.push_back(name);  }
 255
 256                                 if (seqGroup == "not found") { mothurOut(name + " is not in your groupfile. Please correct."); mothurOutEndLine(); exit(1);  }
 257
 258                                 //is this seq in one of hte groups we care about
 259                                 it = groupFinder.find(seqGroup);
 260                                 if (it == groupFinder.end()) {  sharedByAll = false;  } //you have a sequence from a group you don't want
 261                                 else {  atLeastOne[seqGroup]++;  }
 262                         }
 263
 264                         //get last name
 265                         //find group
 266                         if (sharedByAll) {
 267                                 string seqGroup = groupMap->getGroup(names);
 268                                 if (output != "accnos") {
 269                                         namesOfSeqsInThisBin.push_back((names + "|" + seqGroup + "|" + toString(i+1)));
 270                                 }else {  namesOfSeqsInThisBin.push_back(names); }
 271
 272                                 if (seqGroup == "not found") { mothurOut(names + " is not in your groupfile. Please correct."); mothurOutEndLine(); exit(1);  }
 273
 274                                 //is this seq in one of hte groups we care about
 275                                 it = groupFinder.find(seqGroup);
 276                                 if (it == groupFinder.end()) {  sharedByAll = false;  } //you have a sequence from a group you don't want
 277                                 else {  atLeastOne[seqGroup]++;  }
 278                         }
 279
 280                         //make sure you have at least one seq from each group you want
 281                         map<string, int>::iterator it2;
 282                         for (it2 = atLeastOne.begin(); it2 != atLeastOne.end(); it2++) {
 283                                 if (it2->second == 0) {  sharedByAll = false;   }
 284                         }
 285
 286                         //if shared, save names of seqs in that bin
 287                         if (sharedByAll) {
 288
 289                                 wroteSomething = true;
 290                                 num++;
 291
 292                                 //output list of names
 293                                 for (int j = 0; j < namesOfSeqsInThisBin.size(); j++) {
 294                                         outNames << namesOfSeqsInThisBin[j] << endl;
 295
 296                                         if (fastafile != "") {
 297                                                 if (output != "accnos") {
 298                                                         string seqName = namesOfSeqsInThisBin[j].substr(0,namesOfSeqsInThisBin[j].find_last_of('|'));
 299                                                         seqName = seqName.substr(0,seqName.find_last_of('|'));
 300                                                         fastaMap[seqName] = namesOfSeqsInThisBin[j];  //fastaMap needs to contain just the seq name for output later
 301                                                 }else {
 302                                                         fastaMap[namesOfSeqsInThisBin[j]] = namesOfSeqsInThisBin[j];
 303                                                 }
 304                                         }
 305                                 }
 306                         }
 307
 308
 309
 310                 }
 311
 312                 outNames.close();
 313
 314                 if (!wroteSomething) {
 315                         remove(outputFileNames.c_str());
 316                         string outputString = "\t" + toString(num) + " - No otus shared by groups";
 317
 318                         string groupString = "";
 319                         for (int h = 0; h < Groups.size(); h++) {
 320                                 groupString += "  " + Groups[h];
 321                         }
 322
 323                         outputString += groupString + ".";
 324                         mothurOut(outputString); mothurOutEndLine();
 325                 }else { mothurOut("\t" + toString(num)); mothurOutEndLine(); }
 326
 327                 //if fasta file provided output new fasta file
 328                 if ((fastafile != "") && wroteSomething) {
 329                         string outputFileFasta = getRootName(fastafile) + shared->getLabel() + ".fasta";
 330                         ofstream outFasta;
 331                         openOutputFile(outputFileFasta, outFasta);
 332
 333                         for (int k = 0; k < seqs.size(); k++) {
 334                                 //if this is a sequence we want, output it
 335                                 it = fastaMap.find(seqs[k].getName());
 336                                 if (it != fastaMap.end()) {
 337
 338                                         if (output != "accnos") {
 339                                                 outFasta << ">" << it->second << endl;
 340                                         }else {
 341                                                 outFasta << ">" << it->first << endl;
 342                                         }
 343
 344                                         outFasta << seqs[k].getAligned() << endl;
 345                                 }
 346                         }
 347
 348                         outFasta.close();
 349                 }
 350
 351
 352         }
 353         catch(exception& e) {
 354                 errorOut(e, "GetSharedOTUCommand", "process");
 355                 exit(1);
 356         }
 357 }
 358
 359 //**********************************************************************************************************************