#include "phylosummary.h"
#include "referencedb.h"
/**************************************************************************************************/
-Bayesian::Bayesian(string tfile, string tempFile, string method, int ksize, int cutoff, int i, int tid) :
-Classify(), kmerSize(ksize), confidenceThreshold(cutoff), iters(i) {
+Bayesian::Bayesian(string tfile, string tempFile, string method, int ksize, int cutoff, int i, int tid, bool f) :
+Classify(), kmerSize(ksize), confidenceThreshold(cutoff), iters(i) {
try {
ReferenceDB* rdb = ReferenceDB::getInstance();
threadID = tid;
+ flip = f;
string baseName = tempFile;
if (baseName == "saved") { baseName = rdb->getSavedReference(); }
if (baseTName == "saved") { baseTName = rdb->getSavedTaxonomy(); }
/************calculate the probablity that each word will be in a specific taxonomy*************/
- string tfileroot = baseTName.substr(0,baseTName.find_last_of(".")+1);
+ string tfileroot = m->getFullPathName(baseTName.substr(0,baseTName.find_last_of(".")+1));
string tempfileroot = m->getRootName(m->getSimpleName(baseName));
string phyloTreeName = tfileroot + "tree.train";
string phyloTreeSumName = tfileroot + "tree.sum";
if (tfile == "saved") {
m->mothurOutEndLine(); m->mothurOut("Using probabilties from " + rdb->getSavedTaxonomy() + " that are saved in memory... "); cout.flush();;
wordGenusProb = rdb->wordGenusProb;
+ WordPairDiffArr = rdb->WordPairDiffArr;
}else {
m->mothurOut("Reading template probabilities... "); cout.flush();
readProbFile(probFileTest, probFileTest2, probFileName, probFileName2);
}
//save probabilities
- if (rdb->save) { rdb->wordGenusProb = wordGenusProb; }
+ if (rdb->save) { rdb->wordGenusProb = wordGenusProb; rdb->WordPairDiffArr = WordPairDiffArr; }
}else{
//create search database and names vector
//initialze probabilities
wordGenusProb.resize(numKmers);
- //cout << numKmers << '\t' << genusNodes.size() << endl;
+ WordPairDiffArr.resize(numKmers);
+
for (int j = 0; j < wordGenusProb.size(); j++) { wordGenusProb[j].resize(genusNodes.size()); }
- //cout << numKmers << '\t' << genusNodes.size() << endl;
- ofstream out;
+ ofstream out;
ofstream out2;
#ifdef USE_MPI
//probabilityInTemplate = (# of seqs with that word in template + 0.50) / (total number of seqs in template + 1);
float probabilityInTemplate = (seqsWithWordi.size() + 0.50) / (float) (names.size() + 1);
-
+ diffPair tempProb(log(probabilityInTemplate), 0.0);
+ WordPairDiffArr[i] = tempProb;
+
int numNotZero = 0;
for (int k = 0; k < genusNodes.size(); k++) {
//probabilityInThisTaxonomy = (# of seqs with that word in this taxonomy + probabilityInTemplate) / (total number of seqs in this taxonomy + 1);
if (pid == 0) {
#endif
- out << k << '\t' << wordGenusProb[i][k] << '\t';
+ out << k << '\t' << wordGenusProb[i][k] << '\t' ;
#ifdef USE_MPI
}
#endif
out << endl;
- out2 << probabilityInTemplate << '\t' << numNotZero << endl;
+ out2 << probabilityInTemplate << '\t' << numNotZero << '\t' << log(probabilityInTemplate) << endl;
#ifdef USE_MPI
}
delete phyloTree;
phyloTree = new PhyloTree(phyloTreeTest, phyloTreeName);
-
+
//save probabilities
- if (rdb->save) { rdb->wordGenusProb = wordGenusProb; }
+ if (rdb->save) { rdb->wordGenusProb = wordGenusProb; rdb->WordPairDiffArr = WordPairDiffArr; }
}
}
-
+
+ generateWordPairDiffArr();
+
+ //save probabilities
+ if (rdb->save) { rdb->wordGenusProb = wordGenusProb; rdb->WordPairDiffArr = WordPairDiffArr; }
+
m->mothurOut("DONE."); m->mothurOutEndLine();
m->mothurOut("It took " + toString(time(NULL) - start) + " seconds get probabilities. "); m->mothurOutEndLine();
}
try {
string tax = "";
Kmer kmer(kmerSize);
+ flipped = false;
//get words contained in query
//getKmerString returns a string where the index in the string is hte kmer number
//and the character at that index can be converted to be the number of times that kmer was seen
-
string queryKmerString = kmer.getKmerString(seq->getUnaligned());
-
+
vector<int> queryKmers;
for (int i = 0; i < queryKmerString.length()-1; i++) { // the -1 is to ignore any kmer with an N in it
if (queryKmerString[i] != '!') { //this kmer is in the query
}
}
- if (queryKmers.size() == 0) { m->mothurOut(seq->getName() + "is bad."); m->mothurOutEndLine(); return "bad seq"; }
+ //if user wants to test reverse compliment and its reversed use that instead
+ if (flip) {
+ if (isReversed(queryKmers)) {
+ flipped = true;
+ seq->reverseComplement();
+ queryKmerString = kmer.getKmerString(seq->getUnaligned());
+ queryKmers.clear();
+ for (int i = 0; i < queryKmerString.length()-1; i++) { // the -1 is to ignore any kmer with an N in it
+ if (queryKmerString[i] != '!') { //this kmer is in the query
+ queryKmers.push_back(i);
+ }
+ }
+ }
+ }
+
+ if (queryKmers.size() == 0) { m->mothurOut(seq->getName() + "is bad."); m->mothurOutEndLine(); simpleTax = "unknown;"; return "unknown;"; }
int index = getMostProbableTaxonomy(queryKmers);
int numToSelect = queryKmers.size() / 8;
tax = bootstrapResults(queryKmers, index, numToSelect);
-
+
return tax;
}
catch(exception& e) {
seqTax = phyloTree->get(seqTax.parent);
}
- if (confidenceTax == "") { confidenceTax = "unclassified;"; simpleTax = "unclassified;"; }
+ if (confidenceTax == "") { confidenceTax = "unknown;"; simpleTax = "unknown;"; }
+
return confidenceTax;
}
exit(1);
}
}
+//********************************************************************************************************************
+//if it is more probable that the reverse compliment kmers are in the template, then we assume the sequence is reversed.
+bool Bayesian::isReversed(vector<int>& queryKmers){
+ try{
+ bool reversed = false;
+ float prob = 0;
+ float reverseProb = 0;
+
+ for (int i = 0; i < queryKmers.size(); i++){
+ int kmer = queryKmers[i];
+ if (kmer >= 0){
+ prob += WordPairDiffArr[kmer].prob;
+ reverseProb += WordPairDiffArr[kmer].reverseProb;
+ }
+ }
+
+ if (reverseProb > prob){ reversed = true; }
+
+ return reversed;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "Bayesian", "isReversed");
+ exit(1);
+ }
+}
+//********************************************************************************************************************
+int Bayesian::generateWordPairDiffArr(){
+ try{
+ Kmer kmer(kmerSize);
+ for (int i = 0; i < WordPairDiffArr.size(); i++) {
+ int reversedWord = kmer.getReverseKmerNumber(i);
+ WordPairDiffArr[i].reverseProb = WordPairDiffArr[reversedWord].prob;
+ }
+
+ return 0;
+ }catch(exception& e) {
+ m->errorOut(e, "Bayesian", "generateWordPairDiffArr");
+ exit(1);
+ }
+}
/*************************************************************************************************
map<string, int> Bayesian::parseTaxMap(string newTax) {
try{
exit(1);
}
}
-/**************************************************************************************************/
+**************************************************************************************************/
void Bayesian::readProbFile(ifstream& in, ifstream& inNum, string inName, string inNumName) {
try{
int kmer, name;
vector<int> numbers; numbers.resize(numKmers);
float prob;
- vector<float> zeroCountProb; zeroCountProb.resize(numKmers);
+ vector<float> zeroCountProb; zeroCountProb.resize(numKmers);
+ WordPairDiffArr.resize(numKmers);
//read version
length = positions[1] - positions[0];
delete buf4;
istringstream iss (tempBuf,istringstream::in);
- iss >> zeroCountProb[i] >> numbers[i];
+ float probTemp;
+ iss >> zeroCountProb[i] >> numbers[i] >> probTemp;
+ WordPairDiffArr[i].prob = probTemp;
+
}
MPI_File_close(&inMPI);
int kmer, name, count; count = 0;
vector<int> num; num.resize(numKmers);
float prob;
- vector<float> zeroCountProb; zeroCountProb.resize(numKmers);
+ vector<float> zeroCountProb; zeroCountProb.resize(numKmers);
+ WordPairDiffArr.resize(numKmers);
//read version
string line2 = m->getline(inNum); m->gobble(inNum);
+ float probTemp;
//cout << threadID << '\t' << line2 << '\t' << this << endl;
while (inNum) {
- inNum >> zeroCountProb[count] >> num[count];
+ inNum >> zeroCountProb[count] >> num[count] >> probTemp;
+ WordPairDiffArr[count].prob = probTemp;
count++;
m->gobble(inNum);
//cout << threadID << '\t' << count << endl;