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 if (m->control_pressed) { return tax; }
26 vector<string> closestNames;
27 for (int i = 0; i < closest.size(); i++) {
28 //find that sequences taxonomy in map
29 it = taxonomy.find(names[closest[i]]);
31 //is this sequence in the taxonomy file
32 if (it == taxonomy.end()) { //error not in file
33 m->mothurOut("Error: sequence " + names[closest[i]] + " is not in the taxonomy file. It will be eliminated as a match to sequence " + seq->getName() + "."); m->mothurOutEndLine();
34 }else{ closestNames.push_back(it->first); }
37 if (closestNames.size() == 0) {
38 m->mothurOut("Error: All the matches for sequence " + seq->getName() + " have been eliminated. " + seq->getName() + " will be disregarded."); m->mothurOutEndLine();
41 tax = findCommonTaxonomy(closestNames);
42 if (tax == "") { m->mothurOut("There are no common levels for sequence " + seq->getName() + ". " + seq->getName() + " will be disregarded."); m->mothurOutEndLine(); tax = "bad seq"; }
49 m->errorOut(e, "Knn", "getTaxonomy");
53 /**************************************************************************************************/
54 string Knn::findCommonTaxonomy(vector<string> closest) {
56 vector< vector<string> > taxons; //taxon[0] = vector of taxonomy info for closest[0].
57 //so if closest[0] taxonomy is Bacteria;Alphaproteobacteria;Rhizobiales;Azorhizobium_et_rel.;Methylobacterium_et_rel.;Bosea;
58 //taxon[0][0] = Bacteria, taxon[0][1] = Alphaproteobacteria....
60 taxons.resize(closest.size());
63 for (int i = 0; i < closest.size(); i++) {
64 if (m->control_pressed) { return "control"; }
66 string tax = taxonomy[closest[i]]; //we know its there since we checked in getTaxonomy
68 taxons[i] = parseTax(tax);
70 //figure out who has the shortest taxonomy info. so you can start comparing there
71 if (taxons[i].size() < smallest) {
72 smallest = taxons[i].size();
76 //start at the highest level all the closest seqs have
78 for (int i = (smallest-1); i >= 0; i--) {
79 if (m->control_pressed) { return "control"; }
81 string thistax = taxons[0][i];
83 for (int j = 1; j < taxons.size(); j++) {
84 if (taxons[j][i] != thistax) { break; }
88 if (num == (taxons.size()-1)) { //they all match at this level
89 for (int k = 0; k <= i; k++) {
90 common += taxons[0][k] + ';';
99 m->errorOut(e, "Knn", "findCommonTaxonomy");
103 /**************************************************************************************************/