File: kNearestNeighbor.java

package info (click to toggle)
metastudent 2.0.1-10
  • links: PTS, VCS
  • area: main
  • in suites: forky, sid, trixie
  • size: 95,644 kB
  • sloc: java: 3,287; perl: 2,089; python: 1,421; ruby: 242; sh: 39; makefile: 19
file content (161 lines) | stat: -rwxr-xr-x 4,892 bytes parent folder | download | duplicates (6)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
package pp2.prediction.knn;

import java.util.ArrayList;
import java.util.Collections;
import java.util.HashMap;
import java.util.LinkedList;

/**
 * 
 * @author Thomas Hopf
 *
 */
public class kNearestNeighbor {

	/*
	 * constants for different node scoring methods
	 */
	public static final int EVAL_UNWEIGHTED = 1;	// this is unweighted kNN
	public static final int EVAL_LOG_SCORING = 2;
	public static final int EVAL_LOG_LOG_SCORING = 3;
	//public static final int SEQID_SCORING = 4;

	/**
	 *  limits the range of E-values such that log values are meaningful (i.e., -log(value) is defined and non-negative)
	 */
	private double limitRange(double value)
	{
		return Math.max(1E-308, Math.min(value, 1));
	}
	
	/**
	 * 
	 * @param blastHits file like "go_test_output". must already contain fully added GO paths!
	 * @param scoringMethod determines type of sequence weighting, specified by one of the constants EVAL_*
	 * @param useEValueThreshold determines whether conventional kNN or a fixed E-Value threshold is used to include BLAST hits
	 * @param eValueThreshold if useEvalueThreshold == true, then all hits with an E-Value below this parameter are included
	 * @param k if conventional kNN is used, this determines the fixed number of neighbors to use, if there are enough
	 */
	public ArrayList<GoNode> predictFunction(BlastResultList blastHits, int scoringMethod, boolean useEValueThreshold, float eValueThreshold, int k)
	{
		// all GO term nodes
		HashMap<String, GoNode> nodes = new HashMap<String, GoNode>();
		
		// reliability values of each BLAST hit (e.g. E-Value per hit)
		LinkedList<Double> normalizationValues = new LinkedList<Double>();
		int i = 0;
		
		/*
		 * iterate over all hits in the BLAST file
		 */
		for(BlastHit h: blastHits.getHits())
		{
			// only consider first k hits if doing conventional kNN
			if(!useEValueThreshold && i>=k)
				break;
			
			// if doing "E-Value threshold"-kNN, skip current hit (for safety: not assuming the blast hit list is sorted by E-value)
			if(useEValueThreshold && h.getEValue() > eValueThreshold)
				continue;
			
			// iterate over all go terms of the current hit
			for(String goTerm: h.getGoTerms())
			{
				// check if we already had that GO term, otherwise create a new GO term node.
				// in both cases add the index and the reliability score (e.g. E-value) of the blast hit that supports the node
				if(nodes.containsKey(goTerm)){
					nodes.get(goTerm).addHit(i, h.getEValue());
				} else {
					nodes.put(goTerm, new GoNode(goTerm, i, h.getEValue()));
				}
			}
			
			// add current hit to scoring function normalization factor list
			normalizationValues.add(h.getEValue());
			
			i++;
		}
		
		/*
		 * calculate normalization factor
		 */
		double normalizationFactor = 0;
		for(double value: normalizationValues)
		{
			value = limitRange(value);
			
			switch(scoringMethod) {
				case EVAL_UNWEIGHTED:
					normalizationFactor += 1;
					break;
				case EVAL_LOG_SCORING:
					normalizationFactor += -Math.log(value);
					break;
				case EVAL_LOG_LOG_SCORING:
					normalizationFactor += Math.log((-Math.log(value)));
					break;
				default: break;
			}
		}

		// slight hack for safety: ensure normalization factor is nonzero
		if(normalizationFactor <= 0.01)
			normalizationFactor = 1;
		
		//System.out.println("normalization factor: " + normalizationFactor);
		
		
		/*
		 * iterate over all nodes to calculate their individual score
		 */
		for(GoNode node: nodes.values())
		{
			double score = 0;
			
			for(double value: node.getReliabilities())
			{
				value = limitRange(value);

				switch(scoringMethod) {
					case EVAL_UNWEIGHTED:
						score += 1;
						break;
					case EVAL_LOG_SCORING:
						score += -Math.log(value);
						break;
					case EVAL_LOG_LOG_SCORING:
						score += Math.log((-Math.log(value)));
						break;
					default: break;
				}	
			}
			node.setScore(score / normalizationFactor);
			
			//System.out.println(node.getGoTerm() + ":" + node.getReliabilities().size() + "->" + node.getScore());
		}
		
		ArrayList<GoNode> resultNodeList = new ArrayList<GoNode>(nodes.values());
		Collections.sort(resultNodeList);
		return resultNodeList;
	}
	
	
	/**
	 * @param args
	 */
	public static void main(String[] args) throws Exception {
		// nimmt an, dass go_test_output aufsteigend nach E-values sortiert ist, und bereits die vollen GO-pfade enthält 
		BlastResultList brl = new BlastResultList("examples/go_test_output");
		
		kNearestNeighbor knn = new kNearestNeighbor();
		
		System.out.println("#GOterm\t\tnum seq\tscore");
		for(GoNode g: knn.predictFunction(brl, kNearestNeighbor.EVAL_LOG_LOG_SCORING, true, 1, 9))
			System.out.println(g.getGoTerm() + "\t" + g.getReliabilities().size() + "\t" + g.getScore());

		// TODO: die result-liste kann noch nodes mit score 0 enthalten!
		// TODO: kommandozeilenparameter, output file schreiben etc.
		
	}

}