clusterfragmentscommand.cpp

   1 /*
   2  *  ryanscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 9/23/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "clusterfragmentscommand.h"
  11 #include "needlemanoverlap.hpp"
  12
  13 //**********************************************************************************************************************
  14 //sort by unaligned
  15 inline bool comparePriority(seqRNode first, seqRNode second) {
  16         bool better = false;
  17
  18         if (first.length > second.length) {
  19                 better = true;
  20         }else if (first.length == second.length) {
  21                 if (first.numIdentical > second.numIdentical) {
  22                         better = true;
  23                 }
  24         }
  25
  26         return better;
  27 }
  28 //**********************************************************************************************************************
  29 vector<string> ClusterFragmentsCommand::getValidParameters(){
  30         try {
  31                 string AlignArray[] =  {"fasta","name","diffs","percent","outputdir","inputdir"};
  32                 vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
  33                 return myArray;
  34         }
  35         catch(exception& e) {
  36                 m->errorOut(e, "ClusterFragmentsCommand", "getValidParameters");
  37                 exit(1);
  38         }
  39 }
  40 //**********************************************************************************************************************
  41 ClusterFragmentsCommand::ClusterFragmentsCommand(){
  42         try {
  43                 abort = true; calledHelp = true;
  44                 vector<string> tempOutNames;
  45                 outputTypes["fasta"] = tempOutNames;
  46                 outputTypes["name"] = tempOutNames;
  47         }
  48         catch(exception& e) {
  49                 m->errorOut(e, "ClusterFragmentsCommand", "ClusterFragmentsCommand");
  50                 exit(1);
  51         }
  52 }
  53 //**********************************************************************************************************************
  54 vector<string> ClusterFragmentsCommand::getRequiredParameters(){
  55         try {
  56                 string Array[] =  {"fasta"};
  57                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  58                 return myArray;
  59         }
  60         catch(exception& e) {
  61                 m->errorOut(e, "ClusterFragmentsCommand", "getRequiredParameters");
  62                 exit(1);
  63         }
  64 }
  65 //**********************************************************************************************************************
  66 vector<string> ClusterFragmentsCommand::getRequiredFiles(){
  67         try {
  68                 vector<string> myArray;
  69                 return myArray;
  70         }
  71         catch(exception& e) {
  72                 m->errorOut(e, "ClusterFragmentsCommand", "getRequiredFiles");
  73                 exit(1);
  74         }
  75 }
  76 //**********************************************************************************************************************
  77 ClusterFragmentsCommand::ClusterFragmentsCommand(string option) {
  78         try {
  79                 abort = false; calledHelp = false;
  80
  81                 //allow user to run help
  82                 if(option == "help") { help(); abort = true; calledHelp = true; }
  83
  84                 else {
  85                         //valid paramters for this command
  86                         string Array[] =  {"fasta","name","diffs","percent","outputdir","inputdir"};
  87                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  88
  89                         OptionParser parser(option);
  90                         map<string, string> parameters = parser.getParameters();
  91
  92                         ValidParameters validParameter;
  93                         map<string, string>::iterator it;
  94
  95                         //check to make sure all parameters are valid for command
  96                         for (map<string, string>::iterator it2 = parameters.begin(); it2 != parameters.end(); it2++) {
  97                                 if (validParameter.isValidParameter(it2->first, myArray, it2->second) != true) {  abort = true;  }
  98                         }
  99
 100                         //initialize outputTypes
 101                         vector<string> tempOutNames;
 102                         outputTypes["fasta"] = tempOutNames;
 103                         outputTypes["name"] = tempOutNames;
 104
 105                         //if the user changes the input directory command factory will send this info to us in the output parameter
 106                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 107                         if (inputDir == "not found"){   inputDir = "";          }
 108                         else {
 109                                 string path;
 110                                 it = parameters.find("fasta");
 111                                 //user has given a template file
 112                                 if(it != parameters.end()){
 113                                         path = m->hasPath(it->second);
 114                                         //if the user has not given a path then, add inputdir. else leave path alone.
 115                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 116                                 }
 117
 118                                 it = parameters.find("name");
 119                                 //user has given a template file
 120                                 if(it != parameters.end()){
 121                                         path = m->hasPath(it->second);
 122                                         //if the user has not given a path then, add inputdir. else leave path alone.
 123                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 124                                 }
 125                         }
 126
 127                         //check for required parameters
 128                         fastafile = validParameter.validFile(parameters, "fasta", true);
 129                         if (fastafile == "not found") { m->mothurOut("fasta is a required parameter for the cluster.fragments command."); m->mothurOutEndLine(); abort = true; }
 130                         else if (fastafile == "not open") { abort = true; }
 131
 132                         //if the user changes the output directory command factory will send this info to us in the output parameter
 133                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(fastafile);      }
 134
 135                         //check for optional parameter and set defaults
 136                         // ...at some point should added some additional type checking...
 137                         namefile = validParameter.validFile(parameters, "name", true);
 138                         if (namefile == "not found") { namefile =  "";  }
 139                         else if (namefile == "not open") { abort = true; }
 140                         else {  readNameFile();  }
 141
 142                         string temp;
 143                         temp = validParameter.validFile(parameters, "diffs", false);            if (temp == "not found"){       temp = "0";                             }
 144                         convert(temp, diffs);
 145
 146                         temp = validParameter.validFile(parameters, "percent", false);          if (temp == "not found"){       temp = "0";                             }
 147                         convert(temp, percent);
 148
 149                 }
 150
 151         }
 152         catch(exception& e) {
 153                 m->errorOut(e, "ClusterFragmentsCommand", "ClusterFragmentsCommand");
 154                 exit(1);
 155         }
 156 }
 157
 158 //**********************************************************************************************************************
 159 ClusterFragmentsCommand::~ClusterFragmentsCommand(){}
 160 //**********************************************************************************************************************
 161 void ClusterFragmentsCommand::help(){
 162         try {
 163                 m->mothurOut("The cluster.fragments command groups sequences that are part of a larger sequence.\n");
 164                 m->mothurOut("The cluster.fragments command outputs a new fasta and name file.\n");
 165                 m->mothurOut("The cluster.fragments command parameters are fasta, name, diffs and percent. The fasta parameter is required. \n");
 166                 m->mothurOut("The names parameter allows you to give a list of seqs that are identical. This file is 2 columns, first column is name or representative sequence, second column is a list of its identical sequences separated by commas.\n");
 167                 m->mothurOut("The diffs parameter allows you to set the number of differences allowed, default=0. \n");
 168                 m->mothurOut("The percent parameter allows you to set percentage of differences allowed, default=0. percent=2 means if the number of difference is less than or equal to two percent of the length of the fragment, then cluster.\n");
 169                 m->mothurOut("You may use diffs and percent at the same time to say something like: If the number or differences is greater than 1 or more than 2% of the fragment length, don't merge. \n");
 170                 m->mothurOut("The cluster.fragments command should be in the following format: \n");
 171                 m->mothurOut("cluster.fragments(fasta=yourFastaFile, names=yourNamesFile) \n");
 172                 m->mothurOut("Example cluster.fragments(fasta=amazon.fasta).\n");
 173                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
 174         }
 175         catch(exception& e) {
 176                 m->errorOut(e, "ClusterFragmentsCommand", "help");
 177                 exit(1);
 178         }
 179 }
 180 //**********************************************************************************************************************
 181 int ClusterFragmentsCommand::execute(){
 182         try {
 183
 184                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 185
 186                 int start = time(NULL);
 187
 188                 //reads fasta file and return number of seqs
 189                 int numSeqs = readFASTA(); //fills alignSeqs and makes all seqs active
 190
 191                 if (m->control_pressed) { return 0; }
 192
 193                 if (numSeqs == 0) { m->mothurOut("Error reading fasta file...please correct."); m->mothurOutEndLine(); return 0;  }
 194
 195                 //sort seqs by length of unaligned sequence
 196                 sort(alignSeqs.begin(), alignSeqs.end(), comparePriority);
 197
 198                 int count = 0;
 199
 200                 //think about running through twice...
 201                 for (int i = 0; i < numSeqs; i++) {
 202
 203                         if (alignSeqs[i].active) {  //this sequence has not been merged yet
 204
 205                                 string iBases = alignSeqs[i].seq.getUnaligned();
 206
 207                                 //try to merge it with all smaller seqs
 208                                 for (int j = i+1; j < numSeqs; j++) {
 209
 210                                         if (m->control_pressed) { return 0; }
 211
 212                                         if (alignSeqs[j].active) {  //this sequence has not been merged yet
 213
 214                                                 string jBases = alignSeqs[j].seq.getUnaligned();
 215
 216                                                 if (isFragment(iBases, jBases)) {
 217                                                         //merge
 218                                                         alignSeqs[i].names += ',' + alignSeqs[j].names;
 219                                                         alignSeqs[i].numIdentical += alignSeqs[j].numIdentical;
 220
 221                                                         alignSeqs[j].active = 0;
 222                                                         alignSeqs[j].numIdentical = 0;
 223                                                         count++;
 224                                                 }
 225                                         }//end if j active
 226                                 }//end if i != j
 227
 228                                 //remove from active list
 229                                 alignSeqs[i].active = 0;
 230
 231                         }//end if active i
 232                         if(i % 100 == 0)        { m->mothurOut(toString(i) + "\t" + toString(numSeqs - count) + "\t" + toString(count)); m->mothurOutEndLine(); }
 233                 }
 234
 235                 if(numSeqs % 100 != 0)  { m->mothurOut(toString(numSeqs) + "\t" + toString(numSeqs - count) + "\t" + toString(count)); m->mothurOutEndLine();   }
 236
 237
 238                 string fileroot = outputDir + m->getRootName(m->getSimpleName(fastafile));
 239
 240                 string newFastaFile = fileroot + "fragclust.fasta";
 241                 string newNamesFile = fileroot + "names";
 242
 243                 if (m->control_pressed) { return 0; }
 244
 245                 m->mothurOutEndLine();
 246                 m->mothurOut("Total number of sequences before cluster.fragments was " + toString(alignSeqs.size()) + "."); m->mothurOutEndLine();
 247                 m->mothurOut("cluster.fragments removed " + toString(count) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine();
 248
 249                 printData(newFastaFile, newNamesFile);
 250
 251                 m->mothurOut("It took " + toString(time(NULL) - start) + " secs to cluster " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
 252
 253                 if (m->control_pressed) { remove(newFastaFile.c_str()); remove(newNamesFile.c_str()); return 0; }
 254
 255                 m->mothurOutEndLine();
 256                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 257                 m->mothurOut(newFastaFile); m->mothurOutEndLine();
 258                 m->mothurOut(newNamesFile); m->mothurOutEndLine();
 259                 outputNames.push_back(newFastaFile);  outputNames.push_back(newNamesFile); outputTypes["fasta"].push_back(newFastaFile); outputTypes["name"].push_back(newNamesFile);
 260                 m->mothurOutEndLine();
 261
 262                 //set fasta file as new current fastafile
 263                 string current = "";
 264                 itTypes = outputTypes.find("fasta");
 265                 if (itTypes != outputTypes.end()) {
 266                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 267                 }
 268
 269                 itTypes = outputTypes.find("name");
 270                 if (itTypes != outputTypes.end()) {
 271                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 272                 }
 273
 274                 return 0;
 275
 276         }
 277         catch(exception& e) {
 278                 m->errorOut(e, "ClusterFragmentsCommand", "execute");
 279                 exit(1);
 280         }
 281 }
 282 //***************************************************************************************************************
 283 bool ClusterFragmentsCommand::isFragment(string seq1, string seq2){
 284         try {
 285                 bool fragment = false;
 286
 287                 //exact match
 288                 int pos = seq1.find(seq2);
 289                 if (pos != string::npos) { return true; }
 290                 //no match, no diffs wanted
 291                 else if ((diffs == 0) && (percent == 0)) { return false; }
 292                 else { //try aligning and see if you can find it
 293
 294                         //find number of acceptable differences for this sequence fragment
 295                         int totalDiffs;
 296                         if (diffs == 0) { //you didnt set diffs you want a percentage
 297                                 totalDiffs = floor((seq2.length() * (percent / 100.0)));
 298                         }else if (percent == 0) { //you didn't set percent you want diffs
 299                                 totalDiffs = diffs;
 300                         }else if ((percent != 0) && (diffs != 0)) { //you want both, set total diffs to smaller of 2
 301                                 totalDiffs = diffs;
 302                                 int percentDiff = floor((seq2.length() * (percent / 100.0)));
 303                                 if (percentDiff < totalDiffs) { totalDiffs = percentDiff; }
 304                         }
 305
 306                         Alignment* alignment = new NeedlemanOverlap(-1.0, 1.0, -1.0, (seq1.length()+totalDiffs+1));
 307
 308                         //use needleman to align
 309                         alignment->align(seq2, seq1);
 310                         string tempSeq2 = alignment->getSeqAAln();
 311                         string temp = alignment->getSeqBAln();
 312
 313                         delete alignment;
 314
 315                         //chop gap ends
 316                         int startPos = 0;
 317                         int endPos = tempSeq2.length()-1;
 318                         for (int i = 0; i < tempSeq2.length(); i++) {  if (isalpha(tempSeq2[i])) { startPos = i; break; } }
 319                         for (int i = tempSeq2.length()-1; i >= 0; i--) {  if (isalpha(tempSeq2[i])) { endPos = i; break; } }
 320
 321                         //count number of diffs
 322                         int numDiffs = 0;
 323                         for (int i = startPos; i <= endPos; i++) {
 324                                 if (tempSeq2[i] != temp[i]) { numDiffs++; }
 325                         }
 326
 327                         if (numDiffs <= totalDiffs) { fragment = true; }
 328
 329                 }
 330
 331                 return fragment;
 332
 333         }
 334         catch(exception& e) {
 335                 m->errorOut(e, "ClusterFragmentsCommand", "isFragment");
 336                 exit(1);
 337         }
 338 }
 339 /**************************************************************************************************/
 340 int ClusterFragmentsCommand::readFASTA(){
 341         try {
 342
 343                 ifstream inFasta;
 344                 m->openInputFile(fastafile, inFasta);
 345
 346                 while (!inFasta.eof()) {
 347
 348                         if (m->control_pressed) { inFasta.close(); return 0; }
 349
 350                         Sequence seq(inFasta);  m->gobble(inFasta);
 351
 352                         if (seq.getName() != "") {  //can get "" if commented line is at end of fasta file
 353                                 if (namefile != "") {
 354                                         itSize = sizes.find(seq.getName());
 355
 356                                         if (itSize == sizes.end()) { m->mothurOut(seq.getName() + " is not in your names file, please correct."); m->mothurOutEndLine(); exit(1); }
 357                                         else{
 358                                                 seqRNode tempNode(itSize->second, seq, names[seq.getName()], seq.getUnaligned().length());
 359                                                 alignSeqs.push_back(tempNode);
 360                                         }
 361                                 }else { //no names file, you are identical to yourself
 362                                         seqRNode tempNode(1, seq, seq.getName(), seq.getUnaligned().length());
 363                                         alignSeqs.push_back(tempNode);
 364                                 }
 365                         }
 366                 }
 367
 368                 inFasta.close();
 369                 return alignSeqs.size();
 370         }
 371
 372         catch(exception& e) {
 373                 m->errorOut(e, "ClusterFragmentsCommand", "readFASTA");
 374                 exit(1);
 375         }
 376 }
 377 /**************************************************************************************************/
 378 void ClusterFragmentsCommand::printData(string newfasta, string newname){
 379         try {
 380                 ofstream outFasta;
 381                 ofstream outNames;
 382
 383                 m->openOutputFile(newfasta, outFasta);
 384                 m->openOutputFile(newname, outNames);
 385
 386                 for (int i = 0; i < alignSeqs.size(); i++) {
 387                         if (alignSeqs[i].numIdentical != 0) {
 388                                 alignSeqs[i].seq.printSequence(outFasta);
 389                                 outNames << alignSeqs[i].seq.getName() << '\t' << alignSeqs[i].names << endl;
 390                         }
 391                 }
 392
 393                 outFasta.close();
 394                 outNames.close();
 395         }
 396         catch(exception& e) {
 397                 m->errorOut(e, "ClusterFragmentsCommand", "printData");
 398                 exit(1);
 399         }
 400 }
 401 /**************************************************************************************************/
 402
 403 void ClusterFragmentsCommand::readNameFile(){
 404         try {
 405                 ifstream in;
 406                 m->openInputFile(namefile, in);
 407                 string firstCol, secondCol;
 408
 409                 while (!in.eof()) {
 410                         in >> firstCol >> secondCol; m->gobble(in);
 411                         names[firstCol] = secondCol;
 412                         int size = 1;
 413
 414                         for(int i=0;i<secondCol.size();i++){
 415                                 if(secondCol[i] == ','){        size++; }
 416                         }
 417                         sizes[firstCol] = size;
 418                 }
 419                 in.close();
 420         }
 421         catch(exception& e) {
 422                 m->errorOut(e, "ClusterFragmentsCommand", "readNameFile");
 423                 exit(1);
 424         }
 425 }
 426
 427 /**************************************************************************************************/
 428