pipelinepdscommand.cpp

   1 /*
   2  *  pipelinepdscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 10/5/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "pipelinepdscommand.h"
  11 #include "sffinfocommand.h"
  12 #include "commandoptionparser.hpp"
  13
  14 //**********************************************************************************************************************
  15 vector<string> PipelineCommand::getValidParameters(){
  16         try {
  17                 string Array[] =  {"sff","oligos","align","chimera","classify","taxonomy","pipeline","processors","outputdir","inputdir"};
  18                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  19                 return myArray;
  20         }
  21         catch(exception& e) {
  22                 m->errorOut(e, "PipelineCommand", "getValidParameters");
  23                 exit(1);
  24         }
  25 }
  26 //**********************************************************************************************************************
  27 vector<string> PipelineCommand::getRequiredParameters(){
  28         try {
  29                 vector<string> myArray;
  30                 return myArray;
  31         }
  32         catch(exception& e) {
  33                 m->errorOut(e, "PipelineCommand", "getRequiredParameters");
  34                 exit(1);
  35         }
  36 }
  37 //**********************************************************************************************************************
  38 vector<string> PipelineCommand::getRequiredFiles(){
  39         try {
  40                 vector<string> myArray;
  41                 return myArray;
  42         }
  43         catch(exception& e) {
  44                 m->errorOut(e, "PipelineCommand", "getRequiredFiles");
  45                 exit(1);
  46         }
  47 }
  48 //**********************************************************************************************************************
  49 PipelineCommand::PipelineCommand(string option) {
  50         try {
  51                 cFactory = CommandFactory::getInstance();
  52                 abort = false;
  53
  54                 //allow user to run help
  55                 if(option == "help") { help(); abort = true; }
  56
  57                 else {
  58
  59                         //valid paramters for this command
  60                         string AlignArray[] =  {"sff","oligos","align","chimera","classify","taxonomy","pipeline","processors","outputdir","inputdir"};
  61                         vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  62
  63                         OptionParser parser(option);
  64                         map<string, string> parameters = parser.getParameters();
  65
  66                         ValidParameters validParameter;
  67                         map<string, string>::iterator it;
  68
  69                         //check to make sure all parameters are valid for command
  70                         for (it = parameters.begin(); it != parameters.end(); it++) {
  71                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  72                         }
  73
  74                         //if the user changes the input directory command factory will send this info to us in the output parameter
  75                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  76                         if (inputDir == "not found"){   inputDir = "";          }
  77                         else {
  78                                 string path;
  79                                 it = parameters.find("sff");
  80                                 //user has given a template file
  81                                 if(it != parameters.end()){
  82                                         path = m->hasPath(it->second);
  83                                         //if the user has not given a path then, add inputdir. else leave path alone.
  84                                         if (path == "") {       parameters["sff"] = inputDir + it->second;              }
  85                                 }
  86
  87                                 it = parameters.find("oligos");
  88                                 //user has given a template file
  89                                 if(it != parameters.end()){
  90                                         path = m->hasPath(it->second);
  91                                         //if the user has not given a path then, add inputdir. else leave path alone.
  92                                         if (path == "") {       parameters["oligos"] = inputDir + it->second;           }
  93                                 }
  94
  95                                 it = parameters.find("align");
  96                                 //user has given a template file
  97                                 if(it != parameters.end()){
  98                                         path = m->hasPath(it->second);
  99                                         //if the user has not given a path then, add inputdir. else leave path alone.
 100                                         if (path == "") {       parameters["align"] = inputDir + it->second;            }
 101                                 }
 102
 103                                 it = parameters.find("chimera");
 104                                 //user has given a template file
 105                                 if(it != parameters.end()){
 106                                         path = m->hasPath(it->second);
 107                                         //if the user has not given a path then, add inputdir. else leave path alone.
 108                                         if (path == "") {       parameters["chimera"] = inputDir + it->second;          }
 109                                 }
 110
 111                                 it = parameters.find("classify");
 112                                 //user has given a template file
 113                                 if(it != parameters.end()){
 114                                         path = m->hasPath(it->second);
 115                                         //if the user has not given a path then, add inputdir. else leave path alone.
 116                                         if (path == "") {       parameters["classify"] = inputDir + it->second;         }
 117                                 }
 118
 119                                 it = parameters.find("taxonomy");
 120                                 //user has given a template file
 121                                 if(it != parameters.end()){
 122                                         path = m->hasPath(it->second);
 123                                         //if the user has not given a path then, add inputdir. else leave path alone.
 124                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
 125                                 }
 126
 127                                 it = parameters.find("pipeline");
 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["pipeline"] = inputDir + it->second;         }
 133                                 }
 134                         }
 135
 136                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 137
 138                         pipeFilename = validParameter.validFile(parameters, "pipeline", true);
 139                         if (pipeFilename == "not found") { pipeFilename = "";  }
 140                         else if (pipeFilename == "not open") { pipeFilename = ""; abort = true; }
 141
 142                         string temp = validParameter.validFile(parameters, "processors", false);        if (temp == "not found"){       temp = "1";                             }
 143                         convert(temp, processors);
 144
 145                         if (pipeFilename != "") {
 146                                 abort = readUsersPipeline();
 147                         }else{
 148                                 sffFile = validParameter.validFile(parameters, "sff", true);
 149                                 if (sffFile == "not found") { m->mothurOut("sff is a required parameter for the pipeline command."); m->mothurOutEndLine(); abort = true;  }
 150                                 else if (sffFile == "not open") { sffFile = ""; abort = true; }
 151
 152                                 oligosFile = validParameter.validFile(parameters, "oligos", true);
 153                                 if (oligosFile == "not found") { m->mothurOut("oligos is a required parameter for the pipeline command."); m->mothurOutEndLine(); abort = true;  }
 154                                 else if (oligosFile == "not open") { oligosFile = ""; abort = true; }
 155
 156                                 alignFile = validParameter.validFile(parameters, "align", true);
 157                                 if (alignFile == "not found") { m->mothurOut("align is a required parameter for the pipeline command. Please provide the template to align with."); m->mothurOutEndLine(); abort = true;  }
 158                                 else if (alignFile == "not open") { alignFile = ""; abort = true; }
 159
 160                                 chimeraFile = validParameter.validFile(parameters, "chimera", true);
 161                                 if (chimeraFile == "not found") { m->mothurOut("chimera is a required parameter for the pipeline command. Please provide the template to check for chimeras with."); m->mothurOutEndLine(); abort = true;  }
 162                                 else if (chimeraFile == "not open") { chimeraFile = ""; abort = true; }
 163
 164                                 classifyFile = validParameter.validFile(parameters, "classify", true);
 165                                 if (classifyFile == "not found") { m->mothurOut("classify is a required parameter for the pipeline command. Please provide the template to use with the classifier."); m->mothurOutEndLine(); abort = true;  }
 166                                 else if (classifyFile == "not open") { classifyFile = ""; abort = true; }
 167
 168                                 taxonomyFile = validParameter.validFile(parameters, "taxonomy", true);
 169                                 if (taxonomyFile == "not found") { m->mothurOut("taxonomy is a required parameter for the pipeline command."); m->mothurOutEndLine(); abort = true;  }
 170                                 else if (taxonomyFile == "not open") { taxonomyFile = ""; abort = true; }
 171                         }
 172                 }
 173
 174         }
 175         catch(exception& e) {
 176                 m->errorOut(e, "PipelineCommand", "PipelineCommand");
 177                 exit(1);
 178         }
 179 }
 180 //**********************************************************************************************************************
 181
 182 void PipelineCommand::help(){
 183         try {
 184                  m->mothurOut("The pipeline command is designed to guide you through your analysis using mothur.\n");
 185                  m->mothurOut("The pipeline command parameters are pipeline, sff, oligos, align, chimera, classify, taxonomy and processors.\n");
 186                  m->mothurOut("The sff parameter allows you to enter your sff file. It is required.\n");
 187                  m->mothurOut("The oligos parameter allows you to enter your oligos file. It is required.\n");
 188                  m->mothurOut("The align parameter allows you to enter a template to use with the aligner. It is required.\n");
 189                  m->mothurOut("The chimera parameter allows you to enter a template to use for chimera detection. It is required.\n");
 190                  m->mothurOut("The classify parameter allows you to enter a template to use for classification. It is required.\n");
 191                  m->mothurOut("The taxonomy parameter allows you to enter a taxonomy file for the classify template to use for classification. It is required.\n");
 192                  m->mothurOut("The processors parameter allows you to specify the number of processors to use. The default is 1.\n");
 193                  m->mothurOut("The pipeline parameter allows you to enter your own pipeline file. This file should look like a mothur batchfile, but where you would be using a mothur generated file, you can use mothurmade instead.\n");
 194                  m->mothurOut("First column contains the command name, and the second column contains the parameter options or 'defaults', meaning use defaults. You may leave out file options.\n");
 195                  m->mothurOut("Example: trim.seqs(processors=8, allfiles=T, maxambig=0, maxhomop=8, flip=T, bdiffs=1, pdiffs=2, qwindowaverage=35, qwindowsize=50, fasta=may1.v13.fasta, oligos=may1.v13.oligos, qfile=may1.v13.qual)\n");
 196                  m->mothurOut("then, you could enter unique.seqs(fasta=mothurmade), and mothur would use the .trim.fasta file from the trim.seqs command. \n");
 197                  m->mothurOut("then you could enter align.seqs(candidate=mothurmade, template=silva.v13.align, processors=8). , and mothur would use the .trim.unique.fasta file from the unique.seqs command. \n");
 198                  m->mothurOut("If no pipeline file is given then mothur will use Pat's pipeline. \n\n");
 199                  m->mothurOut("Here is a list of the commands used in Pat's pipeline.\n");
 200                  m->mothurOut("All paralellized commands will use the processors you entered.\n");
 201                  m->mothurOut("The sffinfo command takes your sff file and extracts the fasta and quality files.\n");
 202                  m->mothurOut("The trim.seqs command uses your oligos file and the quality and fasta files generated by sffinfo.\n");
 203                  m->mothurOut("The trim.seqs command sets the following parameters: allfiles=T, maxambig=0, maxhomop=8, flip=T, bdiffs=1, pdiffs=2, qwindowaverage=35, qwindowsize=50.\n");
 204                  m->mothurOut("The unique.seqs command uses the trimmed fasta file and removes redundant sequences, don't worry the names file generated by unique.seqs will be used in the pipeline to make sure they are included.\n");
 205                  m->mothurOut("The align.seqs command aligns the unique sequences using the aligners default options. \n");
 206                  m->mothurOut("The screen.seqs command screens the sequences using optimize=end-minlength. \n");
 207                  m->mothurOut("The pipeline uses chimera.slayer to detect chimeras using the default options. \n");
 208                  m->mothurOut("The pipeline removes all sequences determined to be chimeric by chimera.slayer. \n");
 209                  m->mothurOut("The filter.seqs command filters the sequences using vertical=T, trump=. \n");
 210                  m->mothurOut("The unique.seqs command uses the filtered fasta file and name file to remove sequences that have become redundant after filtering.\n");
 211                  m->mothurOut("The pre.cluster command clusters sequences that have no more than 2 differences.\n");
 212                  m->mothurOut("The dist.seqs command is used to generate a column and phylip formatted distance matrix using cutoff=0.20 for column.\n");
 213                  m->mothurOut("The pipeline uses cluster with method=average, hard=T. \n");
 214                  m->mothurOut("The classify.seqs command is used to classify the sequences using the bayesian method with a cutoff of 80.\n");
 215                  m->mothurOut("The phylotype command is used to cluster the sequences based on their classification.\n");
 216                  m->mothurOut("The clearcut command is used to generate a tree using neighbor=T. \n");
 217                  m->mothurOut("The summary.single and summary.shared commands are run on the otu files from cluster and phylotype commands. \n");
 218                  m->mothurOut("The summary.shared command uses calc=sharednseqs-sharedsobs-sharedchao-sharedace-anderberg-jclass-jest-kulczynski-kulczynskicody-lennon-ochiai-sorclass-sorest-whittaker-braycurtis-jabund-morisitahorn-sorabund-thetan-thetayc. \n");
 219                  m->mothurOut("The summary.single command uses calc=nseqs-sobs-coverage-bergerparker-chao-ace-jack-bootstrap-boneh-efron-shen-solow-shannon-npshannon-invsimpson-qstat-simpsoneven-shannoneven-heip-smithwilson. \n");
 220                  m->mothurOut("The classify.otu command is used to get the concensus taxonomy for otu files from cluster and phylotype commands. \n");
 221                  m->mothurOut("The phylo.diversity command run on the tree generated by clearcut with rarefy=T, iters=100. \n");
 222                  m->mothurOut("The unifrac commands are also run on the tree generated by clearcut with random=F, distance=T. \n");
 223                  m->mothurOut("\n\n");
 224         }
 225         catch(exception& e) {
 226                 m->errorOut(e, "PipelineCommand", "help");
 227                 exit(1);
 228         }
 229 }
 230
 231 //**********************************************************************************************************************
 232
 233 PipelineCommand::~PipelineCommand(){}
 234
 235 //**********************************************************************************************************************
 236
 237 int PipelineCommand::execute(){
 238         try {
 239                 if (abort == true) { return 0; }
 240
 241                 int start = time(NULL);
 242
 243                 if (pipeFilename == "") {
 244                         createPatsPipeline();
 245
 246                         //run Pats pipeline
 247                         for (int i = 0; i < commands.size(); i++) {
 248                                 m->mothurOutEndLine(); m->mothurOut("mothur > " + commands[i]); m->mothurOutEndLine();
 249
 250                                 if (m->control_pressed) { return 0; }
 251
 252                                 CommandOptionParser parser(commands[i]);
 253                                 string commandName = parser.getCommandString();
 254                                 string options = parser.getOptionString();
 255
 256                                 #ifdef USE_MPI
 257                                         int pid;
 258                                         MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 259
 260                                         if ((cFactory->MPIEnabled(commandName)) || (pid == 0)) {
 261                                 #endif
 262
 263                                 //executes valid command
 264                                 Command* command = cFactory->getCommand(commandName, options, "pipe");
 265                                 command->execute();
 266
 267                                 //add output files to list
 268                                 map<string, vector<string> > thisCommandsFile = command->getOutputFiles();
 269                                 map<string, vector<string> >::iterator itMade;
 270                                 for (itMade = thisCommandsFile.begin(); itMade != thisCommandsFile.end(); itMade++) {
 271                                         vector<string> temp = itMade->second;
 272                                         for (int j = 0; j < temp.size(); j++) { outputNames.push_back(temp[j]); }
 273                                 }
 274
 275                                 #ifdef USE_MPI
 276                                         }
 277                                 #endif
 278                         }
 279
 280                 }else {  runUsersPipeline(); }
 281
 282                 if (m->control_pressed) { return 0; }
 283
 284                 m->mothurOut("It took " + toString(time(NULL) - start) + " secs to run the pipeline analysis."); m->mothurOutEndLine(); m->mothurOutEndLine();
 285
 286                 m->mothurOutEndLine();
 287                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 288                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 289                 m->mothurOutEndLine();
 290
 291                 return 0;
 292         }
 293         catch(exception& e) {
 294                 m->errorOut(e, "PipelineCommand", "execute");
 295                 exit(1);
 296         }
 297 }
 298 //**********************************************************************************************************************
 299
 300 bool PipelineCommand::readUsersPipeline(){
 301         try {
 302
 303                 ifstream in;
 304                 m->openInputFile(pipeFilename, in);
 305
 306                 string nextCommand = "";
 307
 308                 map<string, vector<string> > mothurMadeFiles;
 309
 310                 while(!in.eof()) {
 311                         nextCommand = m->getline(in); m->gobble(in);
 312
 313                         if (nextCommand[0] != '#') {
 314                                 bool error = false;
 315
 316                                 string commandName, options;
 317                                 error = parseCommand(nextCommand, commandName, options);
 318
 319                                 if (error) { in.close(); return error; }
 320                                 if (commandName == "pipeline.pds") { m->mothurOut("Cannot run the pipeline.pds command from inside the pipeline.pds command."); m->mothurOutEndLine(); in.close(); return true; }
 321
 322                                 error = checkForValidAndRequiredParameters(commandName, options, mothurMadeFiles);
 323
 324                                 if (error) { in.close(); return error; }
 325                         }
 326                 }
 327
 328                 in.close();
 329
 330                 return false;
 331         }
 332         catch(exception& e) {
 333                 m->errorOut(e, "PipelineCommand", "readUsersPipeline");
 334                 exit(1);
 335         }
 336 }
 337 //**********************************************************************************************************************
 338
 339 bool PipelineCommand::parseCommand(string nextCommand, string& name, string& options){
 340         try {
 341                 CommandOptionParser parser(nextCommand);
 342                 name = parser.getCommandString();
 343                 options = parser.getOptionString();
 344
 345                 if (name == "") { return true; } //name == "" if () are not right
 346
 347                 return false;
 348         }
 349         catch(exception& e) {
 350                 m->errorOut(e, "PipelineCommand", "parseCommand");
 351                 exit(1);
 352         }
 353 }
 354 //**********************************************************************************************************************
 355
 356 bool PipelineCommand::checkForValidAndRequiredParameters(string name, string options, map<string, vector<string> >& mothurMadeFiles){
 357         try {
 358
 359                 if (name == "system") { return false; }
 360
 361                 //get shell of the command so we can check to make sure its valid without running it
 362                 Command* command = cFactory->getCommand(name);
 363
 364                 //check to make sure all parameters are valid for command
 365                 vector<string> validParameters = command->getValidParameters();
 366
 367                 OptionParser parser(options);
 368                 map<string, string> parameters = parser.getParameters();
 369
 370                 ValidParameters validParameter;
 371                 map<string, string>::iterator it;
 372                 map<string, vector<string> >::iterator itMade;
 373
 374                 for (it = parameters.begin(); it != parameters.end(); it++) {
 375                         if (validParameter.isValidParameter(it->first, validParameters, it->second) != true) {  return true;  } // not valid
 376                         if (it->second == "mothurmade") {
 377                                 itMade = mothurMadeFiles.find(it->first);
 378
 379                                 if (itMade == mothurMadeFiles.end()) {
 380                                         if ((name == "align.seqs") && (it->first == "candidate")) {} //do nothing about candidate
 381                                         else {
 382                                                 m->mothurOut("You have the " + it->first + " listed as a mothurmade file for the " + name + " command, but it seems mothur will not make that file in your current pipeline, please correct."); m->mothurOutEndLine();
 383                                                 return true;
 384                                         }
 385                                 }
 386                         }
 387                 }
 388
 389                 //is the command missing any required
 390                 vector<string> requiredParameters = command->getRequiredParameters();
 391
 392                 //check for or
 393                 bool hasOr = false;
 394                 int numFound = 0;
 395                 if (requiredParameters.size() > 2) {
 396                         if (requiredParameters[(requiredParameters.size()-1)] == "or") { hasOr = true; }
 397                 }
 398
 399                 for (int i = 0; i < requiredParameters.size(); i++) {
 400                         it = parameters.find(requiredParameters[i]);
 401
 402                         if (it != parameters.end()) { numFound++; }
 403                         else {
 404                                 if (!hasOr) { m->mothurOut(name + " requires the " + requiredParameters[i] + " parameter, please correct."); m->mothurOutEndLine(); }
 405                         }
 406                 }
 407
 408                 // if all are needed and not all are found
 409                 if ((!hasOr) && (numFound != requiredParameters.size())) { return true; }
 410                 //if one is needed and none are found
 411                 else if ((hasOr) && (numFound == 0)) { return true; }
 412
 413                 //update MothurMade
 414                 map<string, vector<string> > thisCommandsFile = command->getOutputFiles();
 415                 for (itMade = thisCommandsFile.begin(); itMade != thisCommandsFile.end(); itMade++) {
 416                         mothurMadeFiles[itMade->first] = itMade->second; //adds any new types
 417                 }
 418
 419                 return false;
 420         }
 421         catch(exception& e) {
 422                 m->errorOut(e, "PipelineCommand", "checkForValidAndRequiredParameters");
 423                 exit(1);
 424         }
 425 }
 426 //**********************************************************************************************************************
 427 int PipelineCommand::runUsersPipeline(){
 428         try {
 429                 ifstream in;
 430                 m->openInputFile(pipeFilename, in);
 431
 432                 string nextCommand = "";
 433
 434                 map<string, vector<string> > mothurMadeFiles;
 435
 436                 while(!in.eof()) {
 437                         nextCommand = m->getline(in);  m->gobble(in);
 438
 439                         if (nextCommand[0] != '#') {
 440                                 CommandOptionParser parser(nextCommand);
 441                                 string commandName = parser.getCommandString();
 442                                 string options = parser.getOptionString();
 443
 444                                 if (options != "") {
 445                                         bool error = fillInMothurMade(options, mothurMadeFiles);
 446                                         if (error) { in.close(); return 0; }
 447                                 }
 448
 449                                 m->mothurOutEndLine(); m->mothurOut("mothur > " + commandName + "(" + options + ")"); m->mothurOutEndLine();
 450
 451                                 if (m->control_pressed) { return 0; }
 452
 453                                 #ifdef USE_MPI
 454                                         int pid;
 455                                         MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 456
 457                                         if ((cFactory->MPIEnabled(commandName)) || (pid == 0)) {
 458                                 #endif
 459
 460                                 //executes valid command
 461                                 Command* command = cFactory->getCommand(commandName, options, "pipe");
 462                                 command->execute();
 463
 464                                 //add output files to list
 465                                 map<string, vector<string> > thisCommandsFile = command->getOutputFiles();
 466                                 map<string, vector<string> >::iterator itMade;
 467                                 map<string, vector<string> >::iterator it;
 468                                 for (itMade = thisCommandsFile.begin(); itMade != thisCommandsFile.end(); itMade++) {
 469
 470                                         vector<string> temp = itMade->second;
 471                                         for (int k = 0; k < temp.size(); k++) { outputNames.push_back(temp[k]); }  //
 472
 473                                         //update Mothur Made for each file
 474                                         it = mothurMadeFiles.find(itMade->first);
 475
 476                                         if (it == mothurMadeFiles.end()) { //new type
 477
 478                                                 mothurMadeFiles[itMade->first] = temp;
 479
 480                                         }else{ //update existing type
 481                                                 vector<string> oldFileNames = it->second;
 482                                                 //look at new files, see if an old version of the file exists, if so update, else just add.
 483                                                 //for example you may have abrecovery.fasta and amazon.fasta as old files and you created a new amazon.trim.fasta.
 484
 485                                                 for (int k = 0; k < temp.size(); k++) {
 486
 487                                                         //get base name
 488                                                         string root = m->getSimpleName(temp[k]);
 489                                                         string individual = "";
 490                                                         for(int i=0;i<root.length();i++){
 491                                                                 if(root[i] == '.'){
 492                                                                         root = individual;
 493                                                                         break;
 494                                                                 }else{
 495                                                                         individual += root[i];
 496                                                                 }
 497                                                         }
 498
 499                                                         //look for that base name in oldfiles
 500                                                         int spot = -1;
 501                                                         for (int l = 0; l < oldFileNames.size(); l++) {
 502                                                                 int pos = oldFileNames[l].find(root);
 503                                                                 if (pos != string::npos) {
 504                                                                         spot = l;
 505                                                                         break;
 506                                                                 }
 507                                                         }
 508
 509                                                         //if you found it update it, else add it
 510                                                         if (spot != -1) {
 511                                                                 mothurMadeFiles[it->first][spot] = temp[k];
 512                                                         }else{
 513                                                                 mothurMadeFiles[it->first].push_back(temp[k]);
 514                                                         }
 515                                                 }
 516                                         }
 517                                 }
 518
 519                                 #ifdef USE_MPI
 520                                         }
 521                                 #endif
 522                         }
 523                 }
 524
 525                 in.close();
 526         }
 527         catch(exception& e) {
 528                 m->errorOut(e, "PipelineCommand", "runUsersPipeline");
 529                 exit(1);
 530         }
 531 }
 532 //**********************************************************************************************************************
 533 bool PipelineCommand::fillInMothurMade(string& options, map<string, vector<string> >& mothurMadeFiles){
 534         try {
 535                 OptionParser parser(options);
 536                 map<string, string> parameters = parser.getParameters();
 537                 map<string, string>::iterator it;
 538                 map<string, vector<string> >::iterator itMade;
 539
 540                 options = "";
 541
 542                 //fill in mothurmade filenames
 543                 for (it = parameters.begin(); it != parameters.end(); it++) {
 544                         string paraType = it->first;
 545                         string tempOption = it->second;
 546
 547                         if (tempOption == "mothurmade") {
 548
 549                                 if (it->first == "candidate") { paraType = "fasta"; }
 550
 551                                 itMade = mothurMadeFiles.find(paraType);
 552
 553                                 if (itMade == mothurMadeFiles.end()) {
 554                                         m->mothurOut("Looking for a mothurmade " + paraType + " file, but it seems mothur has not made that file type in your current pipeline, please correct."); m->mothurOutEndLine();
 555                                         return true;
 556                                 }else{
 557                                         vector<string> temp = itMade->second;
 558
 559                                         if (temp.size() > 1) {
 560                                                 //ask user which file to use
 561                                                 m->mothurOut("More than one file has been created for the " + paraType + " parameter. "); m->mothurOutEndLine();
 562                                                 for (int i = 0; i < temp.size(); i++) {
 563                                                         m->mothurOut(toString(i) + " - " + temp[i]); m->mothurOutEndLine();
 564                                                 }
 565
 566                                                 m->mothurOut("Please select the number of the file you would like to use: ");
 567                                                 int num = 0;
 568                                                 cin >> num;
 569                                                 m->mothurOutJustToLog(toString(num)); m->mothurOutEndLine();
 570
 571                                                 if ((num < 0) || (num > (temp.size()-1))) { m->mothurOut("Not a valid response, quitting."); m->mothurOutEndLine(); return true; }
 572                                                 else {
 573                                                         tempOption = temp[num];
 574                                                 }
 575
 576                                                 //clears buffer so next command doesn't have error
 577                                                 string s;
 578                                                 getline(cin, s);
 579
 580                                                 vector<string> newTemp;
 581                                                 for (int i = 0; i < temp.size(); i++) {
 582                                                         if (i == num) { newTemp.push_back(temp[i]); }
 583                                                         else {
 584                                                                 m->mothurOut("Would you like to remove " + temp[i] + " as an option for " + paraType + ", (y/n): "); m->mothurOutEndLine();
 585                                                                 string response;
 586                                                                 cin >> response;
 587                                                                 m->mothurOutJustToLog(response); m->mothurOutEndLine();
 588
 589                                                                 if (response == "n") {  newTemp.push_back(temp[i]); }
 590
 591                                                                 //clears buffer so next command doesn't have error
 592                                                                 string s;
 593                                                                 getline(cin, s);
 594                                                         }
 595                                                 }
 596
 597                                                 mothurMadeFiles[paraType] = newTemp;
 598
 599
 600                                         }else if (temp.size() == 0){
 601                                                 m->mothurOut("Sorry, we seem to think you created a " + paraType + " file, but it seems mothur doesn't have a filename."); m->mothurOutEndLine();
 602                                                 return true;
 603                                         }else{
 604                                                 tempOption = temp[0];
 605                                         }
 606                                 }
 607                         }
 608
 609                         options += it->first + "=" + tempOption + ", ";
 610                 }
 611
 612                 //rip off extra comma
 613                 options = options.substr(0, (options.length()-2));
 614
 615                 return false;
 616         }
 617         catch(exception& e) {
 618                 m->errorOut(e, "PipelineCommand", "fillInMothurMade");
 619                 exit(1);
 620         }
 621 }
 622
 623 //**********************************************************************************************************************
 624 void PipelineCommand::createPatsPipeline(){
 625         try {
 626
 627                 //sff.info command
 628                 string thisCommand = "sffinfo(sff=" + sffFile + ")";
 629                 commands.push_back(thisCommand);
 630
 631                 //trim.seqs command
 632                 string fastaFile = m->getRootName(m->getSimpleName(sffFile)) + "fasta";
 633                 string qualFile = m->getRootName(m->getSimpleName(sffFile)) + "qual";
 634                 thisCommand = "trim.seqs(processors=" + toString(processors) + ", fasta=" + fastaFile + ", allfiles=T, maxambig=0, maxhomop=8, flip=T, bdiffs=1, pdiffs=2, qwindowaverage=35, qwindowsize=50, oligos=" + oligosFile + ", qfile=" + qualFile + ")";
 635                 commands.push_back(thisCommand);
 636
 637                 //unique.seqs
 638                 string groupFile = m->getRootName(m->getSimpleName(fastaFile)) + "groups";
 639                 qualFile =  m->getRootName(m->getSimpleName(fastaFile)) + "trim.qual";
 640                 fastaFile =  m->getRootName(m->getSimpleName(fastaFile)) + "trim.fasta";
 641                 thisCommand = "unique.seqs(fasta=" + fastaFile + ")";
 642                 commands.push_back(thisCommand);
 643
 644                 //align.seqs
 645                 string nameFile = m->getRootName(m->getSimpleName(fastaFile)) + "names";
 646                 fastaFile = m->getRootName(m->getSimpleName(fastaFile)) + "unique" + m->getExtension(fastaFile);
 647                 thisCommand = "align.seqs(processors=" + toString(processors) + ", candidate=" + fastaFile + ", template=" + alignFile + ")";
 648                 commands.push_back(thisCommand);
 649
 650                 //screen.seqs
 651                 fastaFile = m->getRootName(m->getSimpleName(fastaFile)) + "align";
 652                 thisCommand = "screen.seqs(processors=" + toString(processors) + ", fasta=" + fastaFile + ", name=" + nameFile + ", group=" + groupFile + ", optimize=end-minlength)";
 653                 commands.push_back(thisCommand);
 654
 655                 //chimera.slayer
 656                 fastaFile = m->getRootName(m->getSimpleName(fastaFile)) + "good" + m->getExtension(fastaFile);
 657                 nameFile = m->getRootName(m->getSimpleName(nameFile)) + "good" + m->getExtension(nameFile);
 658                 groupFile = m->getRootName(m->getSimpleName(groupFile)) + "good" + m->getExtension(groupFile);
 659                 thisCommand = "chimera.slayer(processors=" + toString(processors) + ", fasta=" + fastaFile + ", template=" + chimeraFile + ")";
 660                 commands.push_back(thisCommand);
 661
 662                 //remove.seqs
 663                 string accnosFile = m->getRootName(m->getSimpleName(fastaFile))  + "slayer.accnos";
 664                 thisCommand = "remove.seqs(fasta=" + fastaFile + ", name=" + nameFile + ", group=" + groupFile + ", accnos=" + accnosFile + ", dups=T)";
 665                 commands.push_back(thisCommand);
 666
 667                 //filter.seqs
 668                 nameFile = m->getRootName(m->getSimpleName(nameFile)) + "pick" + m->getExtension(nameFile);
 669                 groupFile = m->getRootName(m->getSimpleName(groupFile)) + "pick" + m->getExtension(groupFile);
 670                 fastaFile = m->getRootName(m->getSimpleName(fastaFile)) + "pick" + m->getExtension(fastaFile);
 671                 thisCommand = "filter.seqs(processors=" + toString(processors) + ", fasta=" + fastaFile + ", vertical=T, trump=.)";
 672                 commands.push_back(thisCommand);
 673
 674                 //unique.seqs
 675                 fastaFile =  m->getRootName(m->getSimpleName(fastaFile)) + "filter.fasta";
 676                 thisCommand = "unique.seqs(fasta=" + fastaFile + ", name=" + nameFile + ")";
 677                 commands.push_back(thisCommand);
 678
 679                 //pre.cluster
 680                 nameFile = m->getRootName(m->getSimpleName(fastaFile)) + "names";
 681                 fastaFile = m->getRootName(m->getSimpleName(fastaFile)) + "unique" + m->getExtension(fastaFile);
 682                 thisCommand = "pre.cluster(fasta=" + fastaFile + ", name=" + nameFile + ", diffs=2)";
 683                 commands.push_back(thisCommand);
 684
 685                 //dist.seqs
 686                 nameFile = m->getRootName(m->getSimpleName(fastaFile)) + "precluster.names";
 687                 fastaFile = m->getRootName(m->getSimpleName(fastaFile)) + "precluster" + m->getExtension(fastaFile);
 688                 thisCommand = "dist.seqs(processors=" + toString(processors) + ", fasta=" + fastaFile + ", cutoff=0.20)";
 689                 commands.push_back(thisCommand);
 690
 691                 //dist.seqs
 692                 string columnFile = m->getRootName(m->getSimpleName(fastaFile)) + "dist";
 693                 thisCommand = "dist.seqs(processors=" + toString(processors) + ", fasta=" + fastaFile + ", output=lt)";
 694                 commands.push_back(thisCommand);
 695
 696                 //read.dist
 697                 string phylipFile = m->getRootName(m->getSimpleName(fastaFile)) + "phylip.dist";
 698                 thisCommand = "read.dist(column=" + columnFile + ", name=" + nameFile + ")";
 699                 commands.push_back(thisCommand);
 700
 701                 //cluster
 702                 thisCommand = "cluster(method=average, hard=T)";
 703                 commands.push_back(thisCommand);
 704
 705                 string listFile = m->getRootName(m->getSimpleName(columnFile)) + "an.list";
 706                 string rabundFile = m->getRootName(m->getSimpleName(columnFile)) + "an.rabund";
 707
 708                 //degap.seqs
 709                 thisCommand = "degap.seqs(fasta=" + fastaFile + ")";
 710                 commands.push_back(thisCommand);
 711
 712                 //classify.seqs
 713                 fastaFile = m->getRootName(m->getSimpleName(fastaFile)) + "ng.fasta";
 714                 thisCommand = "classify.seqs(processors=" + toString(processors) + ", fasta=" + fastaFile + ", name=" + nameFile + ", template=" + classifyFile + ", taxonomy=" + taxonomyFile + ", cutoff=80)";
 715                 commands.push_back(thisCommand);
 716
 717                 string RippedTaxName = m->getRootName(m->getSimpleName(taxonomyFile));
 718                 RippedTaxName = m->getExtension(RippedTaxName.substr(0, RippedTaxName.length()-1));
 719                 if (RippedTaxName[0] == '.') { RippedTaxName = RippedTaxName.substr(1, RippedTaxName.length()); }
 720                 RippedTaxName +=  ".";
 721
 722                 string fastaTaxFile = m->getRootName(m->getSimpleName(fastaFile)) + RippedTaxName + "taxonomy";
 723                 string taxSummaryFile = m->getRootName(m->getSimpleName(fastaFile)) + RippedTaxName + "tax.summary";
 724
 725                 //phylotype
 726                 thisCommand = "phylotype(taxonomy=" + fastaTaxFile + ", name=" + nameFile + ")";
 727                 commands.push_back(thisCommand);
 728
 729                 string phyloListFile = m->getRootName(m->getSimpleName(fastaTaxFile)) + "tx.list";
 730                 string phyloRabundFile = m->getRootName(m->getSimpleName(fastaTaxFile)) + "tx.rabund";
 731
 732                 //clearcut
 733                 thisCommand = "clearcut(phylip=" + phylipFile + ", neighbor=T)";
 734                 commands.push_back(thisCommand);
 735
 736                 string treeFile = m->getRootName(m->getSimpleName(phylipFile)) + "tre";
 737
 738                 //read.otu
 739                 thisCommand = "read.otu(list=" + listFile + ", group=" + groupFile + ", label=0.03)";
 740                 commands.push_back(thisCommand);
 741
 742                 string sharedFile = m->getRootName(m->getSimpleName(listFile)) + "shared";
 743
 744                 //read.otu
 745                 thisCommand = "read.otu(list=" + phyloListFile + ", group=" + groupFile + ", label=1)";
 746                 commands.push_back(thisCommand);
 747
 748                 string phyloSharedFile = m->getRootName(m->getSimpleName(phyloListFile)) + "shared";
 749
 750                 //read.otu
 751                 thisCommand = "read.otu(shared=" + sharedFile + ")";
 752                 commands.push_back(thisCommand);
 753
 754                 //summary.single
 755                 thisCommand = "summary.single(calc=nseqs-sobs-coverage-bergerparker-chao-ace-jack-bootstrap-boneh-efron-shen-solow-shannon-npshannon-invsimpson-qstat-simpsoneven-shannoneven-heip-smithwilson, size=5000)";
 756                 commands.push_back(thisCommand);
 757
 758                 //summary.shared
 759                 thisCommand = "summary.shared(calc=sharednseqs-sharedsobs-sharedchao-sharedace-anderberg-jclass-jest-kulczynski-kulczynskicody-lennon-ochiai-sorclass-sorest-whittaker-braycurtis-jabund-morisitahorn-sorabund-thetan-thetayc)";
 760                 commands.push_back(thisCommand);
 761
 762                 //read.otu
 763                 thisCommand = "read.otu(rabund=" + rabundFile + ", label=0.03)";
 764                 commands.push_back(thisCommand);
 765
 766                 //summary.single
 767                 thisCommand = "summary.single(calc=nseqs-sobs-coverage-bergerparker-chao-ace-jack-bootstrap-boneh-efron-shen-solow-shannon-npshannon-invsimpson-qstat-simpsoneven-shannoneven-heip-smithwilson, size=5000)";
 768                 commands.push_back(thisCommand);
 769
 770                 //read.otu
 771                 thisCommand = "read.otu(shared=" + phyloSharedFile + ")";
 772                 commands.push_back(thisCommand);
 773
 774                 //summary.single
 775                 thisCommand = "summary.single(calc=nseqs-sobs-coverage-bergerparker-chao-ace-jack-bootstrap-boneh-efron-shen-solow-shannon-npshannon-invsimpson-qstat-simpsoneven-shannoneven-heip-smithwilson, size=5000)";
 776                 commands.push_back(thisCommand);
 777
 778                 //summary.shared
 779                 thisCommand = "summary.shared(calc=sharednseqs-sharedsobs-sharedchao-sharedace-anderberg-jclass-jest-kulczynski-kulczynskicody-lennon-ochiai-sorclass-sorest-whittaker-braycurtis-jabund-morisitahorn-sorabund-thetan-thetayc)";
 780                 commands.push_back(thisCommand);
 781
 782                 //read.otu
 783                 thisCommand = "read.otu(rabund=" + phyloRabundFile + ", label=1)";
 784                 commands.push_back(thisCommand);
 785
 786                 //summary.single
 787                 thisCommand = "summary.single(calc=nseqs-sobs-coverage-bergerparker-chao-ace-jack-bootstrap-boneh-efron-shen-solow-shannon-npshannon-invsimpson-qstat-simpsoneven-shannoneven-heip-smithwilson, size=5000)";
 788                 commands.push_back(thisCommand);
 789
 790                 //classify.otu
 791                 thisCommand = "classify.otu(taxonomy=" + fastaTaxFile + ", name=" + nameFile + ", list=" + listFile + ", cutoff=51, label=0.03)";
 792                 commands.push_back(thisCommand);
 793
 794                 //classify.otu
 795                 thisCommand = "classify.otu(taxonomy=" + fastaTaxFile + ", name=" + nameFile + ", list=" + phyloListFile + ", cutoff=51, label=1)";
 796                 commands.push_back(thisCommand);
 797
 798                 //read.tree
 799                 thisCommand = "read.tree(tree=" + treeFile + ", name=" + nameFile + ", group=" + groupFile + ")";
 800                 commands.push_back(thisCommand);
 801
 802                 //phylo.diversity
 803                 thisCommand = "phylo.diversity(iters=100,rarefy=T)";
 804                 commands.push_back(thisCommand);
 805
 806                 //unifrac.weighted
 807                 thisCommand = "unifrac.weighted(random=false, distance=true, groups=all, processors=" + toString(processors) + ")";
 808                 commands.push_back(thisCommand);
 809
 810                 //unifrac.unweighted
 811                 thisCommand = "unifrac.unweighted(random=false, distance=true, processors=" + toString(processors) + ")";
 812                 commands.push_back(thisCommand);
 813
 814
 815         }
 816         catch(exception& e) {
 817                 m->errorOut(e, "PipelineCommand", "createPatsPipeline");
 818                 exit(1);
 819         }
 820 }
 821
 822 //**********************************************************************************************************************