annotate src/edu/unc/genomics/ngs/IntervalLengthDistribution.java @ 2:e16016635b2a

Uploaded
author timpalpant
date Mon, 13 Feb 2012 22:12:06 -0500
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
2
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
1 package edu.unc.genomics.ngs;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
2
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
3 import java.io.BufferedWriter;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
4 import java.io.IOException;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
5 import java.nio.charset.Charset;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
6 import java.nio.file.Files;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
7 import java.nio.file.Path;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
8
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
9 import org.apache.commons.math.stat.Frequency;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
10 import org.apache.log4j.Logger;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
11
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
12 import com.beust.jcommander.Parameter;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
13
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
14 import edu.unc.genomics.CommandLineTool;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
15 import edu.unc.genomics.Interval;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
16 import edu.unc.genomics.io.IntervalFile;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
17
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
18 public class IntervalLengthDistribution extends CommandLineTool {
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
19
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
20 private static final Logger log = Logger.getLogger(IntervalLengthDistribution.class);
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
21
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
22 @Parameter(names = {"-i", "--input"}, description = "Interval file", required = true)
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
23 public IntervalFile<? extends Interval> inputFile;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
24 @Parameter(names = {"-o", "--output"}, description = "Output file", required = true)
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
25 public Path outputFile;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
26
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
27
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
28 @Override
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
29 public void run() throws IOException {
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
30 log.debug("Generating histogram of interval lengths");
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
31 Frequency freq = new Frequency();
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
32 int min = Integer.MAX_VALUE;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
33 int max = -1;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
34 for (Interval i : inputFile) {
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
35 int L = i.length();
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
36 freq.addValue(L);
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
37
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
38 if (L < min) {
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
39 min = L;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
40 }
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
41 if (L > max) {
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
42 max = L;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
43 }
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
44 }
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
45
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
46 log.debug("Writing histogram output");
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
47 try (BufferedWriter writer = Files.newBufferedWriter(outputFile, Charset.defaultCharset())) {
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
48 for (int i = min; i <= max; i++) {
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
49 writer.write(i+"\t"+freq.getCount(i));
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
50 writer.newLine();
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
51 }
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
52 }
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
53 }
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
54
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
55 public static void main(String[] args) {
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
56 new IntervalLengthDistribution().instanceMain(args);
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
57 }
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
58
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
59 }