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