annotate java-genomics-toolkit/src/edu/unc/genomics/visualization/MatrixAligner.java @ 0:1daf3026d231

Upload alpha version
author timpalpant
date Mon, 13 Feb 2012 21:55:55 -0500
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
1 package edu.unc.genomics.visualization;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
2
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
3 import java.io.BufferedWriter;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
4 import java.io.IOException;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
5 import java.nio.charset.Charset;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
6 import java.nio.file.Files;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
7 import java.nio.file.Path;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
8 import java.util.Arrays;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
9 import java.util.Iterator;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
10 import java.util.List;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
11
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
12 import org.apache.commons.lang3.ArrayUtils;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
13 import org.apache.log4j.Logger;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
14 import org.broad.igv.bbfile.WigItem;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
15
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
16 import com.beust.jcommander.Parameter;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
17
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
18 import edu.unc.genomics.BedEntry;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
19 import edu.unc.genomics.CommandLineTool;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
20 import edu.unc.genomics.ReadablePathValidator;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
21 import edu.unc.genomics.io.BedFile;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
22 import edu.unc.genomics.io.WigFile;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
23 import edu.unc.genomics.io.WigFileException;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
24
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
25 public class MatrixAligner extends CommandLineTool {
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
26
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
27 private static final Logger log = Logger.getLogger(MatrixAligner.class);
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
28
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
29 @Parameter(names = {"-i", "--input"}, description = "Input file (Wig)", required = true)
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
30 public WigFile inputFile;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
31 @Parameter(names = {"-l", "--loci"}, description = "Loci file (Bed)", required = true, validateWith = ReadablePathValidator.class)
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
32 public Path lociFile;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
33 @Parameter(names = {"-m", "--max"}, description = "Truncate width (base pairs)")
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
34 public Integer maxWidth;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
35 @Parameter(names = {"-o", "--output"}, description = "Output file (matrix2png format)", required = true)
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
36 public Path outputFile;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
37
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
38 private List<BedEntry> loci;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
39
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
40 @Override
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
41 public void run() throws IOException {
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
42 log.debug("Loading alignment intervals");
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
43 try (BedFile bed = new BedFile(lociFile)) {
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
44 loci = bed.loadAll();
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
45 }
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
46
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
47 // Compute the matrix dimensions
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
48 int leftMax = Integer.MIN_VALUE;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
49 int rightMax = Integer.MIN_VALUE;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
50 for (BedEntry entry : loci) {
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
51 int left = Math.abs(entry.getValue().intValue()-entry.getStart());
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
52 int right = Math.abs(entry.getValue().intValue()-entry.getStop());
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
53 if (left > leftMax) {
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
54 leftMax = left;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
55 }
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
56 if (right > rightMax) {
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
57 rightMax = right;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
58 }
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
59 }
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
60
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
61 int m = loci.size();
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
62 int n = leftMax + rightMax + 1;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
63 int alignmentPoint = leftMax;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
64 log.info("Intervals aligned into: " + m+"x"+n + " matrix");
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
65 log.info("Alignment point: " + alignmentPoint);
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
66
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
67 int leftBound = 0;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
68 int rightBound = n-1;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
69 if (maxWidth != null && maxWidth < n) {
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
70 log.info("Truncated to: " + m+"x"+maxWidth);
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
71 int leftAlignDistance = alignmentPoint;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
72 int rightAlignDistance = n - alignmentPoint - 1;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
73 int halfMax = maxWidth / 2;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
74
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
75 if (halfMax < leftAlignDistance && halfMax < rightAlignDistance) {
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
76 leftBound = alignmentPoint - halfMax;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
77 rightBound = alignmentPoint + halfMax;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
78 } else {
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
79 if (leftAlignDistance <= rightAlignDistance) {
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
80 rightBound = maxWidth;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
81 } else {
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
82 leftBound = n - maxWidth;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
83 }
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
84 }
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
85 }
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
86
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
87 log.debug("Initializing output file");
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
88 int count = 0, skipped = 0;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
89 try (BufferedWriter writer = Files.newBufferedWriter(outputFile, Charset.defaultCharset())) {
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
90 writer.write("ID");
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
91 for (int i = leftBound-alignmentPoint; i <= rightBound-alignmentPoint; i++) {
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
92 writer.write("\t"+i);
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
93 }
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
94 writer.newLine();
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
95
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
96 log.debug("Iterating over all intervals");
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
97 String[] row = new String[n];
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
98 for (BedEntry entry : loci) {
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
99 Iterator<WigItem> result = null;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
100 try {
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
101 result = inputFile.query(entry);
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
102 } catch (WigFileException e) {
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
103 skipped++;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
104 continue;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
105 }
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
106
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
107 float[] data = WigFile.flattenData(result, entry.getStart(), entry.getStop());
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
108 // Reverse if on the crick strand
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
109 if (entry.isCrick()) {
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
110 ArrayUtils.reverse(data);
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
111 }
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
112
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
113 // Position the data in the matrix
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
114 // Locus alignment point (entry value) should be positioned over the matrix alignment point
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
115 int n1 = alignmentPoint - Math.abs(entry.getValue().intValue()-entry.getStart());
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
116 int n2 = alignmentPoint + Math.abs(entry.getValue().intValue()-entry.getStop());
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
117 assert data.length == n2-n1+1;
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
118
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
119 Arrays.fill(row, "-");
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
120 for (int i = 0; i < data.length; i++) {
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
121 if (!Float.isNaN(data[i])) {
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
122 row[n1+i] = String.valueOf(data[i]);
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
123 }
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
124 }
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
125
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
126 // Write to output
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
127 String id = ((entry.getId() == null) ? entry.getId() : "Row "+(count++));
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
128 writer.write(id);
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
129 for (int i = leftBound; i <= rightBound; i++) {
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
130 writer.write("\t"+row[i]);
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
131 }
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
132 writer.newLine();
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
133 }
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
134 }
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
135
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
136 inputFile.close();
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
137 log.info(count + " intervals processed");
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
138 log.info(skipped + " intervals skipped");
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
139 }
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
140
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
141 public static void main(String[] args) {
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
142 new MatrixAligner().instanceMain(args);
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
143 }
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
144
1daf3026d231 Upload alpha version
timpalpant
parents:
diff changeset
145 }