5 * Created by westcott on 11/4/09.
6 * Copyright 2009 Schloss Lab. All rights reserved.
12 /**************************************************************************************************/
13 Knn::Knn(string tfile, string tempFile, string method, int kmerSize, float gapOpen, float gapExtend, float match, float misMatch, int n)
14 : Classify(tfile, tempFile, method, kmerSize, gapOpen, gapExtend, match, misMatch), num(n) {}
15 /**************************************************************************************************/
16 string Knn::getTaxonomy(Sequence* seq) {
20 //use database to find closest seq
22 vector<int> closest = database->findClosestSequences(seq, num);
24 vector<string> closestNames;
25 for (int i = 0; i < closest.size(); i++) {
26 //find that sequences taxonomy in map
27 it = taxonomy.find(names[closest[i]]);
29 //is this sequence in the taxonomy file
30 if (it == taxonomy.end()) { //error not in file
31 mothurOut("Error: sequence " + names[closest[i]] + " is not in the taxonomy file. It will be eliminated as a match to sequence " + seq->getName() + "."); mothurOutEndLine();
32 }else{ closestNames.push_back(it->first); }
35 if (closestNames.size() == 0) {
36 mothurOut("Error: All the matches for sequence " + seq->getName() + " have been eliminated. " + seq->getName() + " will be disregarded."); mothurOutEndLine();
39 tax = findCommonTaxonomy(closestNames);
40 if (tax == "") { mothurOut("There are no common levels for sequence " + seq->getName() + ". " + seq->getName() + " will be disregarded."); mothurOutEndLine(); tax = "bad seq"; }
47 errorOut(e, "Knn", "getTaxonomy");
51 /**************************************************************************************************/
52 string Knn::findCommonTaxonomy(vector<string> closest) {
54 vector< vector<string> > taxons; //taxon[0] = vector of taxonomy info for closest[0].
55 //so if closest[0] taxonomy is Bacteria;Alphaproteobacteria;Rhizobiales;Azorhizobium_et_rel.;Methylobacterium_et_rel.;Bosea;
56 //taxon[0][0] = Bacteria, taxon[0][1] = Alphaproteobacteria....
58 taxons.resize(closest.size());
61 for (int i = 0; i < closest.size(); i++) {
63 string tax = taxonomy[closest[i]]; //we know its there since we checked in getTaxonomy
65 taxons[i] = parseTax(tax);
67 //figure out who has the shortest taxonomy info. so you can start comparing there
68 if (taxons[i].size() < smallest) {
69 smallest = taxons[i].size();
73 //start at the highest level all the closest seqs have
75 for (int i = (smallest-1); i >= 0; i--) {
77 string thistax = taxons[0][i];
79 for (int j = 1; j < taxons.size(); j++) {
80 if (taxons[j][i] != thistax) { break; }
84 if (num == (taxons.size()-1)) { //they all match at this level
85 for (int k = 0; k <= i; k++) {
86 common += taxons[0][k] + ';';
95 errorOut(e, "Knn", "findCommonTaxonomy");
99 /**************************************************************************************************/