phylotypecommand.cpp

   1 /*
   2  *  phylotypecommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 11/20/09.
   6  *  Copyright 2009 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "phylotypecommand.h"
  11 #include "phylotree.h"
  12 #include "listvector.hpp"
  13 #include "rabundvector.hpp"
  14 #include "sabundvector.hpp"
  15
  16 //**********************************************************************************************************************
  17 vector<string> PhylotypeCommand::setParameters(){
  18         try {
  19                 CommandParameter ptaxonomy("taxonomy", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(ptaxonomy);
  20                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
  21                 CommandParameter pcutoff("cutoff", "Number", "", "-1", "", "", "",false,false); parameters.push_back(pcutoff);
  22                 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
  23                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  24                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  25
  26                 vector<string> myArray;
  27                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  28                 return myArray;
  29         }
  30         catch(exception& e) {
  31                 m->errorOut(e, "PhylotypeCommand", "setParameters");
  32                 exit(1);
  33         }
  34 }
  35 //**********************************************************************************************************************
  36 string PhylotypeCommand::getHelpString(){
  37         try {
  38                 string helpString = "";
  39                 helpString += "The phylotype command reads a taxonomy file and outputs a .list, .rabund and .sabund file. \n";
  40                 helpString += "The phylotype command parameter options are taxonomy, cutoff and label. The taxonomy parameter is required.\n";
  41                 helpString += "The cutoff parameter allows you to specify the level you want to stop at.  The default is the highest level in your taxonomy file. \n";
  42                 helpString += "For example: taxonomy = Bacteria;Bacteroidetes-Chlorobi;Bacteroidetes; - cutoff=2, would truncate the taxonomy to Bacteria;Bacteroidetes-Chlorobi; \n";
  43                 helpString += "For the cutoff parameter levels count up from the root of the phylotree. This enables you to look at the grouping down to a specific resolution, say the genus level.\n";
  44                 helpString += "The label parameter allows you to specify which level you would like, and are separated by dashes.  The default all levels in your taxonomy file. \n";
  45                 helpString += "For the label parameter, levels count down from the root to keep the output similiar to mothur's other commands which report information from finer resolution to coarser resolutions.\n";
  46                 helpString += "The phylotype command should be in the following format: \n";
  47                 helpString += "phylotype(taxonomy=yourTaxonomyFile, cutoff=yourCutoff, label=yourLabels) \n";
  48                 helpString += "Eaxample: phylotype(taxonomy=amazon.taxonomy, cutoff=5, label=1-3-5).\n";
  49                 return helpString;
  50         }
  51         catch(exception& e) {
  52                 m->errorOut(e, "PhylotypeCommand", "getHelpString");
  53                 exit(1);
  54         }
  55 }
  56 //**********************************************************************************************************************
  57 string PhylotypeCommand::getOutputFileNameTag(string type, string inputName=""){
  58         try {
  59         string outputFileName = "";
  60                 map<string, vector<string> >::iterator it;
  61
  62         //is this a type this command creates
  63         it = outputTypes.find(type);
  64         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  65         else {
  66             if (type == "list") {  outputFileName =  "list"; }
  67             else if (type == "rabund") {  outputFileName =  "rabund"; }
  68             else if (type == "sabund") {  outputFileName =  "sabund"; }
  69             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  70         }
  71         return outputFileName;
  72         }
  73         catch(exception& e) {
  74                 m->errorOut(e, "PhylotypeCommand", "getOutputFileNameTag");
  75                 exit(1);
  76         }
  77 }
  78 //**********************************************************************************************************************
  79 PhylotypeCommand::PhylotypeCommand(){
  80         try {
  81                 abort = true; calledHelp = true;
  82                 setParameters();
  83                 vector<string> tempOutNames;
  84                 outputTypes["list"] = tempOutNames;
  85                 outputTypes["sabund"] = tempOutNames;
  86                 outputTypes["rabund"] = tempOutNames;
  87         }
  88         catch(exception& e) {
  89                 m->errorOut(e, "PhylotypeCommand", "PhylotypeCommand");
  90                 exit(1);
  91         }
  92 }
  93 /**********************************************************************************************************************/
  94 PhylotypeCommand::PhylotypeCommand(string option)  {
  95         try {
  96                 abort = false; calledHelp = false;
  97
  98                 //allow user to run help
  99                 if(option == "help") { help(); abort = true; calledHelp = true; }
 100                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 101
 102                 else {
 103                         vector<string> myArray = setParameters();
 104
 105                         OptionParser parser(option);
 106                         map<string, string> parameters = parser.getParameters();
 107
 108                         ValidParameters validParameter;
 109                         map<string, string>::iterator it;
 110
 111                         //check to make sure all parameters are valid for command
 112                         for (it = parameters.begin(); it != parameters.end(); it++) {
 113                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 114                         }
 115
 116                         //initialize outputTypes
 117                         vector<string> tempOutNames;
 118                         outputTypes["list"] = tempOutNames;
 119                         outputTypes["sabund"] = tempOutNames;
 120                         outputTypes["rabund"] = tempOutNames;
 121
 122                         //if the user changes the input directory command factory will send this info to us in the output parameter
 123                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 124                         if (inputDir == "not found"){   inputDir = "";          }
 125                         else {
 126                                 string path;
 127                                 it = parameters.find("taxonomy");
 128                                 //user has given a template file
 129                                 if(it != parameters.end()){
 130                                         path = m->hasPath(it->second);
 131                                         //if the user has not given a path then, add inputdir. else leave path alone.
 132                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
 133                                 }
 134
 135                                 it = parameters.find("name");
 136                                 //user has given a template file
 137                                 if(it != parameters.end()){
 138                                         path = m->hasPath(it->second);
 139                                         //if the user has not given a path then, add inputdir. else leave path alone.
 140                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 141                                 }
 142                         }
 143
 144                         taxonomyFileName = validParameter.validFile(parameters, "taxonomy", true);
 145                         if (taxonomyFileName == "not found") {
 146                                 taxonomyFileName = m->getTaxonomyFile();
 147                                 if (taxonomyFileName != "") {  m->mothurOut("Using " + taxonomyFileName + " as input file for the taxonomy parameter."); m->mothurOutEndLine(); }
 148                                 else {
 149                                         m->mothurOut("No valid current files. taxonomy is a required parameter."); m->mothurOutEndLine();
 150                                         abort = true;
 151                                 }
 152                         }else if (taxonomyFileName == "not open") { taxonomyFileName = ""; abort = true; }
 153                         else { m->setTaxonomyFile(taxonomyFileName); }
 154
 155                         namefile = validParameter.validFile(parameters, "name", true);
 156                         if (namefile == "not open") { namefile = ""; abort = true; }
 157                         else if (namefile == "not found") { namefile = ""; }
 158                         else { readNamesFile(); m->setNameFile(namefile); }
 159
 160                         //if the user changes the output directory command factory will send this info to us in the output parameter
 161                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 162                                 outputDir = "";
 163                                 outputDir += m->hasPath(taxonomyFileName); //if user entered a file with a path then preserve it
 164                         }
 165
 166                         string temp = validParameter.validFile(parameters, "cutoff", false);
 167                         if (temp == "not found") { temp = "-1"; }
 168                         m->mothurConvert(temp, cutoff);
 169
 170                         label = validParameter.validFile(parameters, "label", false);
 171                         if (label == "not found") { label = ""; allLines = 1; }
 172                         else {
 173                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 174                                 else { allLines = 1;  }
 175                         }
 176
 177                         if (namefile == "") {
 178                                 vector<string> files; files.push_back(taxonomyFileName);
 179                                 parser.getNameFile(files);
 180                         }
 181
 182                 }
 183         }
 184         catch(exception& e) {
 185                 m->errorOut(e, "PhylotypeCommand", "PhylotypeCommand");
 186                 exit(1);
 187         }
 188 }
 189 /**********************************************************************************************************************/
 190
 191 int PhylotypeCommand::execute(){
 192         try {
 193
 194                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 195
 196                 //reads in taxonomy file and makes all the taxonomies the same length
 197                 //by appending the last taxon to a given taxonomy as many times as needed to
 198                 //make it as long as the longest taxonomy in the file
 199                 TaxEqualizer* taxEqual = new TaxEqualizer(taxonomyFileName, cutoff, outputDir);
 200
 201                 if (m->control_pressed) { delete taxEqual; return 0; }
 202
 203                 string equalizedTaxFile = taxEqual->getEqualizedTaxFile();
 204
 205                 delete taxEqual;
 206
 207                 //build taxonomy tree from equalized file
 208                 PhyloTree* tree = new PhyloTree(equalizedTaxFile);
 209                 vector<int> leaves = tree->getGenusNodes();
 210
 211                 //store leaf nodes in current map
 212                 for (int i = 0; i < leaves.size(); i++)         {       currentNodes[leaves[i]] = leaves[i];    }
 213
 214                 bool done = false;
 215                 if (tree->get(leaves[0]).parent == -1) {  m->mothurOut("Empty Tree"); m->mothurOutEndLine();    done = true;    }
 216
 217                 if (m->control_pressed) { delete tree; return 0; }
 218
 219                 string fileroot = outputDir + m->getRootName(m->getSimpleName(taxonomyFileName));
 220
 221                 ofstream outList;
 222                 string outputListFile = fileroot + "tx." + getOutputFileNameTag("list");
 223                 m->openOutputFile(outputListFile, outList);
 224                 ofstream outSabund;
 225                 string outputSabundFile = fileroot + "tx." + getOutputFileNameTag("sabund");
 226                 m->openOutputFile(outputSabundFile, outSabund);
 227                 ofstream outRabund;
 228                 string outputRabundFile = fileroot + "tx." + getOutputFileNameTag("rabund");
 229                 m->openOutputFile(outputRabundFile, outRabund);
 230
 231                 outputNames.push_back(outputListFile); outputTypes["list"].push_back(outputListFile);
 232                 outputNames.push_back(outputSabundFile); outputTypes["sabund"].push_back(outputSabundFile);
 233                 outputNames.push_back(outputRabundFile); outputTypes["rabund"].push_back(outputRabundFile);
 234
 235                 int count = 1;
 236                 //start at leaves of tree and work towards root, processing the labels the user wants
 237                 while((!done) && ((allLines == 1) || (labels.size() != 0))) {
 238
 239                         string level = toString(count);
 240                         count++;
 241
 242                         if (m->control_pressed) {
 243                                 outRabund.close(); outSabund.close(); outList.close();
 244                                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]);  }
 245                                 delete tree; return 0;
 246                         }
 247
 248                         //is this a level the user want output for
 249                         if(allLines == 1 || labels.count(level) == 1){
 250
 251                                 //output level
 252                                 m->mothurOut(level); m->mothurOutEndLine();
 253
 254                                 ListVector list;
 255                                 list.setLabel(level);
 256
 257                                 //go through nodes and build listvector
 258                                 for (itCurrent = currentNodes.begin(); itCurrent != currentNodes.end(); itCurrent++) {
 259
 260                                         //get parents
 261                                         TaxNode node = tree->get(itCurrent->first);
 262                                         parentNodes[node.parent] = node.parent;
 263
 264                                         vector<string> names = node.accessions;
 265
 266                                         //make the names compatable with listvector
 267                                         string name = "";
 268                                         for (int i = 0; i < names.size(); i++) {
 269
 270                         if (names[i] != "unknown") {
 271                             if (namefile != "") {
 272                                 map<string, string>::iterator itNames = namemap.find(names[i]);  //make sure this name is in namefile
 273
 274                                 if (itNames != namemap.end()) {  name += namemap[names[i]] + ",";   } //you found it in namefile
 275                                 else { m->mothurOut("[ERROR]: " + names[i] + " is not in your namefile, please correct."); m->mothurOutEndLine(); m->control_pressed = true;  }
 276
 277                             }else{   name += names[i] + ",";    }
 278                         }
 279                                         }
 280
 281                     if (m->control_pressed) { break; }
 282
 283                                         name = name.substr(0, name.length()-1);  //rip off extra ','
 284                                         //add bin to list vector
 285                                         if (name != "") { list.push_back(name); } //caused by unknown
 286                                 }
 287
 288                                 //print listvector
 289                                 list.print(outList);
 290                                 //print rabund
 291                                 list.getRAbundVector().print(outRabund);
 292                                 //print sabund
 293                                 list.getSAbundVector().print(outSabund);
 294
 295                                 labels.erase(level);
 296
 297                         }else {
 298
 299                                 //just get parents
 300                                 for (itCurrent = currentNodes.begin(); itCurrent != currentNodes.end(); itCurrent++) {
 301                                         int parent = tree->get(itCurrent->first).parent;
 302                                         parentNodes[parent] = parent;
 303                                 }
 304                         }
 305
 306                         //move up a level
 307                         currentNodes = parentNodes;
 308                         parentNodes.clear();
 309
 310                         //have we reached the rootnode
 311                         if (tree->get(currentNodes.begin()->first).parent == -1)  {  done = true;  }
 312                 }
 313
 314                 outList.close();
 315                 outSabund.close();
 316                 outRabund.close();
 317
 318                 delete tree;
 319
 320                 if (m->control_pressed) {
 321                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]);  }
 322                         return 0;
 323                 }
 324
 325                 //set list file as new current listfile
 326                 string current = "";
 327                 itTypes = outputTypes.find("list");
 328                 if (itTypes != outputTypes.end()) {
 329                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 330                 }
 331
 332                 //set rabund file as new current rabundfile
 333                 itTypes = outputTypes.find("rabund");
 334                 if (itTypes != outputTypes.end()) {
 335                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setRabundFile(current); }
 336                 }
 337
 338                 //set sabund file as new current sabundfile
 339                 itTypes = outputTypes.find("sabund");
 340                 if (itTypes != outputTypes.end()) {
 341                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSabundFile(current); }
 342                 }
 343
 344                 m->mothurOutEndLine();
 345                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 346                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 347                 m->mothurOutEndLine();
 348
 349                 return 0;
 350         }
 351
 352         catch(exception& e) {
 353                 m->errorOut(e, "PhylotypeCommand", "execute");
 354                 exit(1);
 355         }
 356 }
 357 /*****************************************************************/
 358 int PhylotypeCommand::readNamesFile() {
 359         try {
 360
 361                 ifstream in;
 362                 m->openInputFile(namefile, in);
 363
 364                 string first, second;
 365                 map<string, string>::iterator itNames;
 366
 367                 while(!in.eof()) {
 368                         in >> first >> second; m->gobble(in);
 369
 370                         itNames = namemap.find(first);
 371                         if (itNames == namemap.end()) {
 372                                 namemap[first] = second;
 373                         }else {  m->mothurOut(first + " has already been seen in namefile, disregarding names file."); m->mothurOutEndLine(); in.close(); namemap.clear(); namefile = ""; return 1; }
 374                 }
 375                 in.close();
 376
 377                 return 0;
 378         }
 379         catch(exception& e) {
 380                 m->errorOut(e, "PhylotypeCommand", "readNamesFile");
 381                 exit(1);
 382         }
 383 }
 384
 385 /**********************************************************************************************************************/