1 | package de.ugoe.cs.autoquest.tasktrees.alignment.algorithms; |
---|
2 | |
---|
3 | import java.util.ArrayList; |
---|
4 | import java.util.Iterator; |
---|
5 | import java.util.LinkedList; |
---|
6 | |
---|
7 | import de.ugoe.cs.autoquest.tasktrees.alignment.matrix.SubstitutionMatrix; |
---|
8 | import de.ugoe.cs.autoquest.tasktrees.alignment.algorithms.Constants; |
---|
9 | |
---|
10 | |
---|
11 | public class NeedlemanWunsch implements AlignmentAlgorithm { |
---|
12 | |
---|
13 | /** |
---|
14 | * The first input |
---|
15 | */ |
---|
16 | private int[] input1; |
---|
17 | |
---|
18 | /** |
---|
19 | * The second input String |
---|
20 | */ |
---|
21 | private int[] input2; |
---|
22 | |
---|
23 | /** |
---|
24 | * The lengths of the input |
---|
25 | */ |
---|
26 | private int length1, length2; |
---|
27 | |
---|
28 | /** |
---|
29 | * The score matrix. The true scores should be divided by the normalization |
---|
30 | * factor. |
---|
31 | */ |
---|
32 | private MatrixEntry[][] matrix; |
---|
33 | |
---|
34 | private ArrayList<NumberSequence> alignment; |
---|
35 | |
---|
36 | /** |
---|
37 | * Substitution matrix to calculate scores |
---|
38 | */ |
---|
39 | private SubstitutionMatrix submat; |
---|
40 | |
---|
41 | |
---|
42 | /** |
---|
43 | * Compute the similarity score of substitution The position of the first |
---|
44 | * character is 1. A position of 0 represents a gap. |
---|
45 | * |
---|
46 | * @param i |
---|
47 | * Position of the character in str1 |
---|
48 | * @param j |
---|
49 | * Position of the character in str2 |
---|
50 | * @return Cost of substitution of the character in str1 by the one in str2 |
---|
51 | */ |
---|
52 | private double similarity(int i, int j) { |
---|
53 | return submat.getScore(input1[i - 1], input2[j - 1]); |
---|
54 | } |
---|
55 | |
---|
56 | /** |
---|
57 | * Build the score matrix using dynamic programming. |
---|
58 | */ |
---|
59 | private void buildMatrix() { |
---|
60 | if (submat.getGapPenalty() >= 0) { |
---|
61 | throw new Error("Indel score must be negative"); |
---|
62 | } |
---|
63 | |
---|
64 | // it's a gap |
---|
65 | matrix[0][0].setScore(0); |
---|
66 | matrix[0][0].setPrevious(null); // starting point |
---|
67 | |
---|
68 | // the first column |
---|
69 | for (int j = 1; j <= length2; j++) { |
---|
70 | matrix[0][j].setScore(j*submat.getGapPenalty()); |
---|
71 | matrix[0][j].setPrevious(matrix[0][j - 1]); |
---|
72 | matrix[0][j].setYvalue(input2[j-1]); |
---|
73 | matrix[0][j].setXvalue(Constants.GAP_SYMBOL); |
---|
74 | } |
---|
75 | // the first row |
---|
76 | |
---|
77 | for (int j = 1; j <= length1; j++) { |
---|
78 | matrix[j][0].setScore(j*submat.getGapPenalty()); |
---|
79 | matrix[j][0].setPrevious(matrix[j - 1][0]); |
---|
80 | matrix[j][0].setXvalue(input1[j-1]); |
---|
81 | matrix[j][0].setYvalue(Constants.GAP_SYMBOL); |
---|
82 | } |
---|
83 | |
---|
84 | for (int i = 1; i <= length1; i++) { |
---|
85 | |
---|
86 | for (int j = 1; j <= length2; j++) { |
---|
87 | double diagScore = matrix[i - 1][j - 1].getScore() |
---|
88 | + similarity(i, j); |
---|
89 | double upScore = matrix[i][j - 1].getScore() |
---|
90 | + submat.getGapPenalty(); |
---|
91 | double leftScore = matrix[i - 1][j].getScore() |
---|
92 | + submat.getGapPenalty(); |
---|
93 | |
---|
94 | matrix[i][j].setScore(Math.max(diagScore, |
---|
95 | Math.max(upScore, leftScore))); |
---|
96 | |
---|
97 | // find the directions that give the maximum scores. |
---|
98 | // TODO: Multiple directions are ignored, we choose the first |
---|
99 | // maximum score |
---|
100 | // True if we had a match |
---|
101 | if (diagScore == matrix[i][j].getScore()) { |
---|
102 | matrix[i][j].setPrevious(matrix[i - 1][j - 1]); |
---|
103 | matrix[i][j].setXvalue(input1[i - 1]); |
---|
104 | matrix[i][j].setYvalue(input2[j - 1]); |
---|
105 | } |
---|
106 | // true if we took an event from sequence x and not from y |
---|
107 | if (leftScore == matrix[i][j].getScore()) { |
---|
108 | matrix[i][j].setXvalue(input1[i - 1]); |
---|
109 | matrix[i][j].setYvalue(Constants.GAP_SYMBOL); |
---|
110 | matrix[i][j].setPrevious(matrix[i - 1][j]); |
---|
111 | } |
---|
112 | // true if we took an event from sequence y and not from x |
---|
113 | if (upScore == matrix[i][j].getScore()) { |
---|
114 | matrix[i][j].setXvalue(Constants.GAP_SYMBOL); |
---|
115 | matrix[i][j].setYvalue(input2[j - 1]); |
---|
116 | matrix[i][j].setPrevious(matrix[i][j - 1]); |
---|
117 | } |
---|
118 | } |
---|
119 | } |
---|
120 | } |
---|
121 | |
---|
122 | /** |
---|
123 | * Get the maximum value in the score matrix. |
---|
124 | */ |
---|
125 | public double getMaxScore() { |
---|
126 | double maxScore = 0; |
---|
127 | |
---|
128 | // skip the first row and column |
---|
129 | for (int i = 1; i <= length1; i++) { |
---|
130 | for (int j = 1; j <= length2; j++) { |
---|
131 | if (matrix[i][j].getScore() > maxScore) { |
---|
132 | maxScore = matrix[i][j].getScore(); |
---|
133 | } |
---|
134 | } |
---|
135 | } |
---|
136 | |
---|
137 | return maxScore; |
---|
138 | } |
---|
139 | |
---|
140 | /* |
---|
141 | * (non-Javadoc) |
---|
142 | * |
---|
143 | * @see |
---|
144 | * de.ugoe.cs.autoquest.tasktrees.alignment.algorithms.AlignmentAlgorithm |
---|
145 | * #getAlignmentScore() |
---|
146 | */ |
---|
147 | @Override |
---|
148 | public double getAlignmentScore() { |
---|
149 | return getMaxScore(); |
---|
150 | } |
---|
151 | |
---|
152 | public void traceback() { |
---|
153 | MatrixEntry tmp = matrix[length1][length2]; |
---|
154 | LinkedList<Integer> aligned1 = new LinkedList<Integer>(); |
---|
155 | LinkedList<Integer> aligned2 = new LinkedList<Integer>(); |
---|
156 | while (tmp.getPrevious() != null) { |
---|
157 | |
---|
158 | aligned1.add(new Integer(tmp.getXvalue())); |
---|
159 | aligned2.add(new Integer(tmp.getYvalue())); |
---|
160 | |
---|
161 | tmp = tmp.getPrevious(); |
---|
162 | } |
---|
163 | |
---|
164 | // reverse order of the alignment |
---|
165 | int reversed1[] = new int[aligned1.size()]; |
---|
166 | int reversed2[] = new int[aligned2.size()]; |
---|
167 | |
---|
168 | int count = 0; |
---|
169 | for (Iterator<Integer> it = aligned1.iterator(); it.hasNext();) { |
---|
170 | count++; |
---|
171 | reversed1[reversed1.length - count] = it.next(); |
---|
172 | |
---|
173 | } |
---|
174 | count = 0; |
---|
175 | for (Iterator<Integer> it = aligned2.iterator(); it.hasNext();) { |
---|
176 | count++; |
---|
177 | reversed2[reversed2.length - count] = it.next(); |
---|
178 | } |
---|
179 | |
---|
180 | NumberSequence ns1 = new NumberSequence(reversed1.length); |
---|
181 | NumberSequence ns2 = new NumberSequence(reversed2.length); |
---|
182 | ns1.setSequence(reversed1); |
---|
183 | ns2.setSequence(reversed2); |
---|
184 | |
---|
185 | alignment.add(ns1); |
---|
186 | alignment.add(ns2); |
---|
187 | } |
---|
188 | |
---|
189 | |
---|
190 | /** |
---|
191 | * print the dynmaic programming matrix |
---|
192 | */ |
---|
193 | public void printDPMatrix() { |
---|
194 | System.out.print(" "); |
---|
195 | for (int i = 1; i <= length1; i++) |
---|
196 | System.out.format("%5d", input1[i - 1]); |
---|
197 | System.out.println(); |
---|
198 | for (int j = 0; j <= length2; j++) { |
---|
199 | if (j > 0) |
---|
200 | System.out.format("%5d ", input2[j - 1]); |
---|
201 | else { |
---|
202 | System.out.print(" "); |
---|
203 | } |
---|
204 | for (int i = 0; i <= length1; i++) { |
---|
205 | System.out.format("%4.1f ", matrix[i][j].getScore()); |
---|
206 | } |
---|
207 | System.out.println(); |
---|
208 | } |
---|
209 | } |
---|
210 | |
---|
211 | public void printAlignment() { |
---|
212 | int[] tmp1 = alignment.get(0).getSequence(); |
---|
213 | int[] tmp2 = alignment.get(1).getSequence(); |
---|
214 | for (int i=0; i< tmp1.length;i++) { |
---|
215 | if(tmp1[i] == Constants.GAP_SYMBOL) { |
---|
216 | System.out.print(" ___"); |
---|
217 | } |
---|
218 | else if(tmp1[i] == Constants.UNMATCHED_SYMBOL) { |
---|
219 | System.out.print(" ..."); |
---|
220 | } |
---|
221 | else { |
---|
222 | System.out.format("%5d", tmp1[i]); |
---|
223 | } |
---|
224 | |
---|
225 | } |
---|
226 | System.out.println(); |
---|
227 | for (int i=0; i< tmp2.length;i++) { |
---|
228 | if(tmp2[i] == Constants.GAP_SYMBOL) { |
---|
229 | System.out.print(" ___"); |
---|
230 | } |
---|
231 | else if(tmp2[i] == Constants.UNMATCHED_SYMBOL) { |
---|
232 | System.out.print(" ..."); |
---|
233 | } |
---|
234 | else { |
---|
235 | System.out.format("%5d", tmp2[i]); |
---|
236 | } |
---|
237 | |
---|
238 | } |
---|
239 | System.out.println(); |
---|
240 | |
---|
241 | |
---|
242 | |
---|
243 | } |
---|
244 | |
---|
245 | /* |
---|
246 | * (non-Javadoc) |
---|
247 | * |
---|
248 | * @see |
---|
249 | * de.ugoe.cs.autoquest.tasktrees.alignment.algorithms.AlignmentAlgorithm |
---|
250 | * #getAlignment() |
---|
251 | */ |
---|
252 | @Override |
---|
253 | public ArrayList<NumberSequence> getAlignment() { |
---|
254 | return alignment; |
---|
255 | } |
---|
256 | |
---|
257 | public void setAlignment(ArrayList<NumberSequence> alignment) { |
---|
258 | this.alignment = alignment; |
---|
259 | } |
---|
260 | |
---|
261 | @Override |
---|
262 | public ArrayList<Match> getMatches() { |
---|
263 | // TODO Auto-generated method stub |
---|
264 | return null; |
---|
265 | } |
---|
266 | |
---|
267 | @Override |
---|
268 | public void align(NumberSequence input1, NumberSequence input2, SubstitutionMatrix submat, |
---|
269 | float threshold) { |
---|
270 | this.input1 = input1.getSequence(); |
---|
271 | this.input2 = input2.getSequence(); |
---|
272 | length1 = input1.size(); |
---|
273 | length2 = input2.size(); |
---|
274 | this.submat = submat; |
---|
275 | |
---|
276 | // System.out.println("Starting SmithWaterman algorithm with a " |
---|
277 | // + submat.getClass() + " Substitution Matrix: " + |
---|
278 | // submat.getClass().getCanonicalName()); |
---|
279 | |
---|
280 | matrix = new MatrixEntry[length1 + 1][length2 + 1]; |
---|
281 | alignment = new ArrayList<NumberSequence>(); |
---|
282 | |
---|
283 | for (int i = 0; i < length1+1; i++) { |
---|
284 | for (int j = 0; j < length2+1; j++) { |
---|
285 | matrix[i][j] = new MatrixEntry(); |
---|
286 | } |
---|
287 | } |
---|
288 | |
---|
289 | buildMatrix(); |
---|
290 | traceback(); |
---|
291 | |
---|
292 | } |
---|
293 | |
---|
294 | |
---|
295 | |
---|
296 | } |
---|