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(), num(n) {
16 //create search database and names vector
17 generateDatabaseAndNames(tfile, tempFile, method, kmerSize, gapOpen, gapExtend, match, misMatch);
20 m->errorOut(e, "Knn", "Knn");
24 /**************************************************************************************************/
28 if (database != NULL) { delete database; }
31 m->errorOut(e, "Knn", "~Knn");
35 /**************************************************************************************************/
36 string Knn::getTaxonomy(Sequence* seq) {
40 //use database to find closest seq
41 vector<int> closest = database->findClosestSequences(seq, num);
43 if (m->control_pressed) { return tax; }
45 vector<string> closestNames;
46 for (int i = 0; i < closest.size(); i++) {
47 //find that sequences taxonomy in map
48 it = taxonomy.find(names[closest[i]]);
50 //is this sequence in the taxonomy file
51 if (it == taxonomy.end()) { //error not in file
52 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();
53 }else{ closestNames.push_back(it->first); }
56 if (closestNames.size() == 0) {
57 m->mothurOut("Error: All the matches for sequence " + seq->getName() + " have been eliminated. " + seq->getName() + " will be disregarded."); m->mothurOutEndLine();
60 tax = findCommonTaxonomy(closestNames);
61 if (tax == "") { m->mothurOut("There are no common levels for sequence " + seq->getName() + ". " + seq->getName() + " will be disregarded."); m->mothurOutEndLine(); tax = "bad seq"; }
68 m->errorOut(e, "Knn", "getTaxonomy");
72 /**************************************************************************************************/
73 string Knn::findCommonTaxonomy(vector<string> closest) {
75 vector< vector<string> > taxons; //taxon[0] = vector of taxonomy info for closest[0].
76 //so if closest[0] taxonomy is Bacteria;Alphaproteobacteria;Rhizobiales;Azorhizobium_et_rel.;Methylobacterium_et_rel.;Bosea;
77 //taxon[0][0] = Bacteria, taxon[0][1] = Alphaproteobacteria....
79 taxons.resize(closest.size());
82 for (int i = 0; i < closest.size(); i++) {
83 if (m->control_pressed) { return "control"; }
85 string tax = taxonomy[closest[i]]; //we know its there since we checked in getTaxonomy
87 taxons[i] = parseTax(tax);
89 //figure out who has the shortest taxonomy info. so you can start comparing there
90 if (taxons[i].size() < smallest) {
91 smallest = taxons[i].size();
95 //start at the highest level all the closest seqs have
97 for (int i = (smallest-1); i >= 0; i--) {
98 if (m->control_pressed) { return "control"; }
100 string thistax = taxons[0][i];
102 for (int j = 1; j < taxons.size(); j++) {
103 if (taxons[j][i] != thistax) { break; }
107 if (num == (taxons.size()-1)) { //they all match at this level
108 for (int k = 0; k <= i; k++) {
109 common += taxons[0][k] + ';';
117 catch(exception& e) {
118 m->errorOut(e, "Knn", "findCommonTaxonomy");
122 /**************************************************************************************************/