5 * Created by westcott on 11/3/09.
6 * Copyright 2009 Schloss Lab. All rights reserved.
12 #include "phylosummary.h"
13 #include "referencedb.h"
14 /**************************************************************************************************/
15 Bayesian::Bayesian(string tfile, string tempFile, string method, int ksize, int cutoff, int i, int tid, bool f, bool sh) :
16 Classify(), kmerSize(ksize), confidenceThreshold(cutoff), iters(i) {
18 ReferenceDB* rdb = ReferenceDB::getInstance();
23 string baseName = tempFile;
25 if (baseName == "saved") { baseName = rdb->getSavedReference(); }
27 string baseTName = tfile;
28 if (baseTName == "saved") { baseTName = rdb->getSavedTaxonomy(); }
30 /************calculate the probablity that each word will be in a specific taxonomy*************/
31 string tfileroot = m->getFullPathName(baseTName.substr(0,baseTName.find_last_of(".")+1));
32 string tempfileroot = m->getRootName(m->getSimpleName(baseName));
33 string phyloTreeName = tfileroot + "tree.train";
34 string phyloTreeSumName = tfileroot + "tree.sum";
35 string probFileName = tfileroot + tempfileroot + char('0'+ kmerSize) + "mer.prob";
36 string probFileName2 = tfileroot + tempfileroot + char('0'+ kmerSize) + "mer.numNonZero";
41 ifstream phyloTreeTest(phyloTreeName.c_str());
42 ifstream probFileTest2(probFileName2.c_str());
43 ifstream probFileTest(probFileName.c_str());
44 ifstream probFileTest3(phyloTreeSumName.c_str());
46 int start = time(NULL);
48 //if they are there make sure they were created after this release date
49 bool FilesGood = false;
50 if(probFileTest && probFileTest2 && phyloTreeTest && probFileTest3){
51 FilesGood = checkReleaseDate(probFileTest, probFileTest2, phyloTreeTest, probFileTest3);
54 //if you want to save, but you dont need to calculate then just read
55 if (rdb->save && probFileTest && probFileTest2 && phyloTreeTest && probFileTest3 && FilesGood && (tempFile != "saved")) {
57 m->openInputFile(tempFile, saveIn);
59 while (!saveIn.eof()) {
60 Sequence temp(saveIn);
63 rdb->referenceSeqs.push_back(temp);
68 if(probFileTest && probFileTest2 && phyloTreeTest && probFileTest3 && FilesGood){
69 if (tempFile == "saved") { m->mothurOutEndLine(); m->mothurOut("Using sequences from " + rdb->getSavedReference() + " that are saved in memory."); m->mothurOutEndLine(); }
71 m->mothurOut("Reading template taxonomy... "); cout.flush();
73 phyloTree = new PhyloTree(phyloTreeTest, phyloTreeName);
75 m->mothurOut("DONE."); m->mothurOutEndLine();
77 genusNodes = phyloTree->getGenusNodes();
78 genusTotals = phyloTree->getGenusTotals();
80 if (tfile == "saved") {
81 m->mothurOutEndLine(); m->mothurOut("Using probabilties from " + rdb->getSavedTaxonomy() + " that are saved in memory... "); cout.flush();;
82 wordGenusProb = rdb->wordGenusProb;
83 WordPairDiffArr = rdb->WordPairDiffArr;
85 m->mothurOut("Reading template probabilities... "); cout.flush();
86 readProbFile(probFileTest, probFileTest2, probFileName, probFileName2);
90 if (rdb->save) { rdb->wordGenusProb = wordGenusProb; rdb->WordPairDiffArr = WordPairDiffArr; }
93 //create search database and names vector
94 generateDatabaseAndNames(tfile, tempFile, method, ksize, 0.0, 0.0, 0.0, 0.0);
96 //prevents errors caused by creating shortcut files if you had an error in the sanity check.
97 if (m->control_pressed) { m->mothurRemove(phyloTreeName); m->mothurRemove(probFileName); m->mothurRemove(probFileName2); }
99 genusNodes = phyloTree->getGenusNodes();
100 genusTotals = phyloTree->getGenusTotals();
102 m->mothurOut("Calculating template taxonomy tree... "); cout.flush();
104 phyloTree->printTreeNodes(phyloTreeName);
106 m->mothurOut("DONE."); m->mothurOutEndLine();
108 m->mothurOut("Calculating template probabilities... "); cout.flush();
110 numKmers = database->getMaxKmer() + 1;
112 //initialze probabilities
113 wordGenusProb.resize(numKmers);
114 WordPairDiffArr.resize(numKmers);
116 for (int j = 0; j < wordGenusProb.size(); j++) { wordGenusProb[j].resize(genusNodes.size()); }
122 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
129 m->openOutputFile(probFileName, out);
131 //output mothur version
132 out << "#" << m->getVersion() << endl;
134 out << numKmers << endl;
136 m->openOutputFile(probFileName2, out2);
138 //output mothur version
139 out2 << "#" << m->getVersion() << endl;
147 for (int i = 0; i < numKmers; i++) {
148 if (m->control_pressed) { break; }
151 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
156 if (shortcuts) { out << i << '\t'; }
162 vector<int> seqsWithWordi = database->getSequencesWithKmer(i);
164 //for each sequence with that word
165 vector<int> count; count.resize(genusNodes.size(), 0);
166 for (int j = 0; j < seqsWithWordi.size(); j++) {
167 int temp = phyloTree->getGenusIndex(names[seqsWithWordi[j]]);
168 count[temp]++; //increment count of seq in this genus who have this word
171 //probabilityInTemplate = (# of seqs with that word in template + 0.50) / (total number of seqs in template + 1);
172 float probabilityInTemplate = (seqsWithWordi.size() + 0.50) / (float) (names.size() + 1);
173 diffPair tempProb(log(probabilityInTemplate), 0.0);
174 WordPairDiffArr[i] = tempProb;
177 for (int k = 0; k < genusNodes.size(); k++) {
178 //probabilityInThisTaxonomy = (# of seqs with that word in this taxonomy + probabilityInTemplate) / (total number of seqs in this taxonomy + 1);
181 wordGenusProb[i][k] = log((count[k] + probabilityInTemplate) / (float) (genusTotals[k] + 1));
186 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
191 if (shortcuts) { out << k << '\t' << wordGenusProb[i][k] << '\t' ; }
202 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
209 out2 << probabilityInTemplate << '\t' << numNotZero << '\t' << log(probabilityInTemplate) << endl;
218 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
231 //read in new phylotree with less info. - its faster
232 ifstream phyloTreeTest(phyloTreeName.c_str());
235 phyloTree = new PhyloTree(phyloTreeTest, phyloTreeName);
238 if (rdb->save) { rdb->wordGenusProb = wordGenusProb; rdb->WordPairDiffArr = WordPairDiffArr; }
242 generateWordPairDiffArr();
245 if (rdb->save) { rdb->wordGenusProb = wordGenusProb; rdb->WordPairDiffArr = WordPairDiffArr; }
247 m->mothurOut("DONE."); m->mothurOutEndLine();
248 m->mothurOut("It took " + toString(time(NULL) - start) + " seconds get probabilities. "); m->mothurOutEndLine();
250 catch(exception& e) {
251 m->errorOut(e, "Bayesian", "Bayesian");
255 /**************************************************************************************************/
256 Bayesian::~Bayesian() {
260 if (database != NULL) { delete database; }
262 catch(exception& e) {
263 m->errorOut(e, "Bayesian", "~Bayesian");
268 /**************************************************************************************************/
269 string Bayesian::getTaxonomy(Sequence* seq) {
275 //get words contained in query
276 //getKmerString returns a string where the index in the string is hte kmer number
277 //and the character at that index can be converted to be the number of times that kmer was seen
278 string queryKmerString = kmer.getKmerString(seq->getUnaligned());
280 vector<int> queryKmers;
281 for (int i = 0; i < queryKmerString.length()-1; i++) { // the -1 is to ignore any kmer with an N in it
282 if (queryKmerString[i] != '!') { //this kmer is in the query
283 queryKmers.push_back(i);
287 //if user wants to test reverse compliment and its reversed use that instead
289 if (isReversed(queryKmers)) {
291 seq->reverseComplement();
292 queryKmerString = kmer.getKmerString(seq->getUnaligned());
294 for (int i = 0; i < queryKmerString.length()-1; i++) { // the -1 is to ignore any kmer with an N in it
295 if (queryKmerString[i] != '!') { //this kmer is in the query
296 queryKmers.push_back(i);
302 if (queryKmers.size() == 0) { m->mothurOut(seq->getName() + "is bad."); m->mothurOutEndLine(); simpleTax = "unknown;"; return "unknown;"; }
305 int index = getMostProbableTaxonomy(queryKmers);
307 if (m->control_pressed) { return tax; }
309 //bootstrap - to set confidenceScore
310 int numToSelect = queryKmers.size() / 8;
312 tax = bootstrapResults(queryKmers, index, numToSelect);
316 catch(exception& e) {
317 m->errorOut(e, "Bayesian", "getTaxonomy");
321 /**************************************************************************************************/
322 string Bayesian::bootstrapResults(vector<int> kmers, int tax, int numToSelect) {
325 map<int, int> confidenceScores;
327 //initialize confidences to 0
329 TaxNode seq = phyloTree->get(tax);
330 confidenceScores[tax] = 0;
332 while (seq.level != 0) { //while you are not at the root
333 seqIndex = seq.parent;
334 confidenceScores[seqIndex] = 0;
335 seq = phyloTree->get(seq.parent);
338 map<int, int>::iterator itBoot;
339 map<int, int>::iterator itBoot2;
340 map<int, int>::iterator itConvert;
342 for (int i = 0; i < iters; i++) {
343 if (m->control_pressed) { return "control"; }
346 for (int j = 0; j < numToSelect; j++) {
347 int index = int(rand() % kmers.size());
350 temp.push_back(kmers[index]);
354 int newTax = getMostProbableTaxonomy(temp);
356 TaxNode taxonomyTemp = phyloTree->get(newTax);
358 //add to confidence results
359 while (taxonomyTemp.level != 0) { //while you are not at the root
360 itBoot2 = confidenceScores.find(newTax); //is this a classification we already have a count on
362 if (itBoot2 != confidenceScores.end()) { //this is a classification we need a confidence for
366 newTax = taxonomyTemp.parent;
367 taxonomyTemp = phyloTree->get(newTax);
372 string confidenceTax = "";
375 int seqTaxIndex = tax;
376 TaxNode seqTax = phyloTree->get(tax);
378 while (seqTax.level != 0) { //while you are not at the root
380 itBoot2 = confidenceScores.find(seqTaxIndex); //is this a classification we already have a count on
383 if (itBoot2 != confidenceScores.end()) { //already in confidence scores
384 confidence = itBoot2->second;
387 if (((confidence/(float)iters) * 100) >= confidenceThreshold) {
388 confidenceTax = seqTax.name + "(" + toString(((confidence/(float)iters) * 100)) + ");" + confidenceTax;
389 simpleTax = seqTax.name + ";" + simpleTax;
392 seqTaxIndex = seqTax.parent;
393 seqTax = phyloTree->get(seqTax.parent);
396 if (confidenceTax == "") { confidenceTax = "unknown;"; simpleTax = "unknown;"; }
398 return confidenceTax;
401 catch(exception& e) {
402 m->errorOut(e, "Bayesian", "bootstrapResults");
406 /**************************************************************************************************/
407 int Bayesian::getMostProbableTaxonomy(vector<int> queryKmer) {
409 int indexofGenus = 0;
411 double maxProbability = -1000000.0;
412 //find taxonomy with highest probability that this sequence is from it
415 // cout << genusNodes.size() << endl;
418 for (int k = 0; k < genusNodes.size(); k++) {
419 //for each taxonomy calc its probability
421 double prob = 0.0000;
422 for (int i = 0; i < queryKmer.size(); i++) {
423 prob += wordGenusProb[queryKmer[i]][k];
426 // cout << phyloTree->get(genusNodes[k]).name << '\t' << prob << endl;
428 //is this the taxonomy with the greatest probability?
429 if (prob > maxProbability) {
430 indexofGenus = genusNodes[k];
431 maxProbability = prob;
438 catch(exception& e) {
439 m->errorOut(e, "Bayesian", "getMostProbableTaxonomy");
443 //********************************************************************************************************************
444 //if it is more probable that the reverse compliment kmers are in the template, then we assume the sequence is reversed.
445 bool Bayesian::isReversed(vector<int>& queryKmers){
447 bool reversed = false;
449 float reverseProb = 0;
451 for (int i = 0; i < queryKmers.size(); i++){
452 int kmer = queryKmers[i];
454 prob += WordPairDiffArr[kmer].prob;
455 reverseProb += WordPairDiffArr[kmer].reverseProb;
459 if (reverseProb > prob){ reversed = true; }
463 catch(exception& e) {
464 m->errorOut(e, "Bayesian", "isReversed");
468 //********************************************************************************************************************
469 int Bayesian::generateWordPairDiffArr(){
472 for (int i = 0; i < WordPairDiffArr.size(); i++) {
473 int reversedWord = kmer.getReverseKmerNumber(i);
474 WordPairDiffArr[i].reverseProb = WordPairDiffArr[reversedWord].prob;
478 }catch(exception& e) {
479 m->errorOut(e, "Bayesian", "generateWordPairDiffArr");
483 /*************************************************************************************************
484 map<string, int> Bayesian::parseTaxMap(string newTax) {
487 map<string, int> parsed;
489 newTax = newTax.substr(0, newTax.length()-1); //get rid of last ';'
493 while (newTax.find_first_of(';') != -1) {
494 individual = newTax.substr(0,newTax.find_first_of(';'));
495 newTax = newTax.substr(newTax.find_first_of(';')+1, newTax.length());
496 parsed[individual] = 1;
505 catch(exception& e) {
506 m->errorOut(e, "Bayesian", "parseTax");
510 **************************************************************************************************/
511 void Bayesian::readProbFile(ifstream& in, ifstream& inNum, string inName, string inNumName) {
516 int pid, num, num2, processors;
517 vector<unsigned long long> positions;
518 vector<unsigned long long> positions2;
523 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
524 MPI_Comm_size(MPI_COMM_WORLD, &processors);
527 char inFileName[1024];
528 strcpy(inFileName, inNumName.c_str());
530 char inFileName2[1024];
531 strcpy(inFileName2, inName.c_str());
533 MPI_File_open(MPI_COMM_WORLD, inFileName, MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI); //comm, filename, mode, info, filepointer
534 MPI_File_open(MPI_COMM_WORLD, inFileName2, MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI2); //comm, filename, mode, info, filepointer
537 positions = m->setFilePosEachLine(inNumName, num);
538 positions2 = m->setFilePosEachLine(inName, num2);
540 for(int i = 1; i < processors; i++) {
541 MPI_Send(&num, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
542 MPI_Send(&positions[0], (num+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
544 MPI_Send(&num2, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
545 MPI_Send(&positions2[0], (num2+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
549 MPI_Recv(&num, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
550 positions.resize(num+1);
551 MPI_Recv(&positions[0], (num+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
553 MPI_Recv(&num2, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
554 positions2.resize(num2+1);
555 MPI_Recv(&positions2[0], (num2+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
559 int length = positions2[1] - positions2[0];
560 char* buf5 = new char[length];
562 MPI_File_read_at(inMPI2, positions2[0], buf5, length, MPI_CHAR, &status);
566 length = positions2[2] - positions2[1];
567 char* buf = new char[length];
569 MPI_File_read_at(inMPI2, positions2[1], buf, length, MPI_CHAR, &status);
571 string tempBuf = buf;
572 if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length); }
575 istringstream iss (tempBuf,istringstream::in);
578 //initialze probabilities
579 wordGenusProb.resize(numKmers);
581 for (int j = 0; j < wordGenusProb.size(); j++) { wordGenusProb[j].resize(genusNodes.size()); }
584 vector<int> numbers; numbers.resize(numKmers);
586 vector<float> zeroCountProb; zeroCountProb.resize(numKmers);
587 WordPairDiffArr.resize(numKmers);
590 length = positions[1] - positions[0];
591 char* buf6 = new char[length];
593 MPI_File_read_at(inMPI2, positions[0], buf6, length, MPI_CHAR, &status);
597 for(int i=1;i<num;i++){
599 length = positions[i+1] - positions[i];
600 char* buf4 = new char[length];
602 MPI_File_read_at(inMPI, positions[i], buf4, length, MPI_CHAR, &status);
605 if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length); }
608 istringstream iss (tempBuf,istringstream::in);
610 iss >> zeroCountProb[i] >> numbers[i] >> probTemp;
611 WordPairDiffArr[i].prob = probTemp;
615 MPI_File_close(&inMPI);
617 for(int i=2;i<num2;i++){
619 length = positions2[i+1] - positions2[i];
620 char* buf4 = new char[length];
622 MPI_File_read_at(inMPI2, positions2[i], buf4, length, MPI_CHAR, &status);
625 if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length); }
628 istringstream iss (tempBuf,istringstream::in);
632 //set them all to zero value
633 for (int i = 0; i < genusNodes.size(); i++) {
634 wordGenusProb[kmer][i] = log(zeroCountProb[kmer] / (float) (genusTotals[i]+1));
637 //get probs for nonzero values
638 for (int i = 0; i < numbers[kmer]; i++) {
640 wordGenusProb[kmer][name] = prob;
644 MPI_File_close(&inMPI2);
645 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
648 string line = m->getline(in); m->gobble(in);
650 in >> numKmers; m->gobble(in);
651 //cout << threadID << '\t' << line << '\t' << numKmers << &in << '\t' << &inNum << '\t' << genusNodes.size() << endl;
652 //initialze probabilities
653 wordGenusProb.resize(numKmers);
655 for (int j = 0; j < wordGenusProb.size(); j++) { wordGenusProb[j].resize(genusNodes.size()); }
657 int kmer, name, count; count = 0;
658 vector<int> num; num.resize(numKmers);
660 vector<float> zeroCountProb; zeroCountProb.resize(numKmers);
661 WordPairDiffArr.resize(numKmers);
664 string line2 = m->getline(inNum); m->gobble(inNum);
666 //cout << threadID << '\t' << line2 << '\t' << this << endl;
668 inNum >> zeroCountProb[count] >> num[count] >> probTemp;
669 WordPairDiffArr[count].prob = probTemp;
672 //cout << threadID << '\t' << count << endl;
675 //cout << threadID << '\t' << "here1 " << &wordGenusProb << '\t' << &num << endl; //
676 //cout << threadID << '\t' << &genusTotals << '\t' << endl;
677 //cout << threadID << '\t' << genusNodes.size() << endl;
680 //cout << threadID << '\t' << kmer << endl;
681 //set them all to zero value
682 for (int i = 0; i < genusNodes.size(); i++) {
683 wordGenusProb[kmer][i] = log(zeroCountProb[kmer] / (float) (genusTotals[i]+1));
685 //cout << threadID << '\t' << num[kmer] << "here" << endl;
686 //get probs for nonzero values
687 for (int i = 0; i < num[kmer]; i++) {
689 wordGenusProb[kmer][name] = prob;
695 //cout << threadID << '\t' << "here" << endl;
698 catch(exception& e) {
699 m->errorOut(e, "Bayesian", "readProbFile");
703 /**************************************************************************************************/
704 bool Bayesian::checkReleaseDate(ifstream& file1, ifstream& file2, ifstream& file3, ifstream& file4) {
709 vector<string> lines;
710 lines.push_back(m->getline(file1));
711 lines.push_back(m->getline(file2));
712 lines.push_back(m->getline(file3));
713 lines.push_back(m->getline(file4));
715 //before we added this check
716 if ((lines[0][0] != '#') || (lines[1][0] != '#') || (lines[2][0] != '#') || (lines[3][0] != '#')) { good = false; }
719 for (int i = 0; i < lines.size(); i++) { lines[i] = lines[i].substr(1); }
721 //get mothurs current version
722 string version = m->getVersion();
724 vector<string> versionVector;
725 m->splitAtChar(version, versionVector, '.');
727 //check each files version
728 for (int i = 0; i < lines.size(); i++) {
729 vector<string> linesVector;
730 m->splitAtChar(lines[i], linesVector, '.');
732 if (versionVector.size() != linesVector.size()) { good = false; break; }
734 for (int j = 0; j < versionVector.size(); j++) {
736 convert(versionVector[j], num1);
737 convert(linesVector[j], num2);
739 //if mothurs version is newer than this files version, then we want to remake it
740 if (num1 > num2) { good = false; break; }
744 if (!good) { break; }
748 if (!good) { file1.close(); file2.close(); file3.close(); file4.close(); }
749 else { file1.seekg(0); file2.seekg(0); file3.seekg(0); file4.seekg(0); }
753 catch(exception& e) {
754 m->errorOut(e, "Bayesian", "checkReleaseDate");
758 /**************************************************************************************************/