|
0
|
1 package middlewares;
|
|
|
2 import java.io.File;
|
|
|
3 import java.io.FileNotFoundException;
|
|
|
4 import java.util.Scanner;
|
|
|
5
|
|
|
6 import datastructures.AVLTree;
|
|
|
7 import datastructures.AnnotationLine;
|
|
|
8 import datastructures.TargetLine;
|
|
|
9
|
|
|
10 /**
|
|
|
11 *
|
|
|
12 * Generate a AVL-Tree for fast extraction of genes (logarithmic time).
|
|
|
13 *
|
|
|
14 * @author Ali Abdallah
|
|
|
15 * @version 0.4.5, 14.07.2011
|
|
|
16 * @since jdk 1.6.0
|
|
|
17 *
|
|
|
18 */
|
|
|
19
|
|
|
20 public class GeneExtractor {
|
|
|
21
|
|
|
22 /**
|
|
|
23 * The path of the genome annotation file.
|
|
|
24 */
|
|
|
25 String genomeAnnotation;
|
|
|
26
|
|
|
27 /**
|
|
|
28 * The avl tree representing the genome annotation file.
|
|
|
29 */
|
|
|
30 AVLTree genesTree;
|
|
|
31
|
|
|
32 /**
|
|
|
33 * The scanner scanning the genome file.
|
|
|
34 */
|
|
|
35 Scanner s;
|
|
|
36
|
|
|
37 /**
|
|
|
38 * Constructs the avl-tree based on the genome annotation file.
|
|
|
39 * @param genomeAnnotation the genome annotation file.
|
|
|
40 */
|
|
|
41 public GeneExtractor(String genomeAnnotation) {
|
|
|
42 genesTree = new AVLTree();
|
|
|
43 this.genomeAnnotation = genomeAnnotation;
|
|
|
44 try {
|
|
|
45 s = new Scanner(new File(genomeAnnotation));
|
|
|
46 } catch (FileNotFoundException e) {
|
|
|
47 e.printStackTrace();
|
|
|
48 }
|
|
|
49 while (s.hasNextLine()) {
|
|
|
50 genesTree.insert(new AnnotationLine(s.nextLine()));
|
|
|
51 }
|
|
|
52 s.close();
|
|
|
53 }
|
|
|
54
|
|
|
55 /**
|
|
|
56 * Search the tree for a gene overlapping the specified target.
|
|
|
57 *
|
|
|
58 * @param tl the target line of the current target.
|
|
|
59 * @return the gene overlapping the specified target, if it exists and
|
|
|
60 * "unknown" otherwise.
|
|
|
61 */
|
|
|
62 public String extractGene(TargetLine tl) {
|
|
|
63 AnnotationLine a =
|
|
|
64 (AnnotationLine) genesTree.find(new AnnotationLine
|
|
|
65 ("*", tl.chrom(), "dummy", tl.start(), tl.end()));
|
|
|
66 return ((a != null) ? a.gene() : "unknown");
|
|
|
67 }
|
|
|
68 }
|
