srcjar/fr/orsay/lri/varna/utils/TranslateFormatRNaseP.java

   1 /**
   2  * File written by Raphael Champeimont
   3  * UMR 7238 Genomique des Microorganismes
   4  */
   5 package fr.orsay.lri.varna.utils;
   6
   7 import java.io.BufferedReader;
   8 import java.io.BufferedWriter;
   9 import java.io.File;
  10 import java.io.FileReader;
  11 import java.io.FileWriter;
  12 import java.io.IOException;
  13 import java.util.ArrayList;
  14 import java.util.List;
  15
  16 public class TranslateFormatRNaseP {
  17         public static void main(String[] args) throws Exception {
  18                 File templatesDir = new File(new File(System.getProperty("user.dir")), "templates");
  19                 File infile = new File(new File(templatesDir, "RNaseP_bact_a"), "a_bacterial_rnas.gb");
  20                 File outfile = new File(new File(templatesDir, "RNaseP_bact_a"), "alignment.fasta");
  21
  22                 BufferedReader inbuf = new BufferedReader(new FileReader(infile));
  23                 String line = inbuf.readLine();
  24                 String seqname;
  25                 List<String> seqnames = new ArrayList<String>();
  26                 List<String> sequences = new ArrayList<String>();
  27                 while (line != null) {
  28                         if (line.length() != 0) {
  29                                 if (line.startsWith("LOCUS")) {
  30                                         String parts[] = line.split("\\s+");
  31                                         seqname = parts[1];
  32                                         seqnames.add(seqname);
  33                                         sequences.add("");
  34                                 }
  35                                 if (line.startsWith(" ")) {
  36                                         String parts[] = line.split("\\s+");
  37                                         for (int i=2; i<parts.length; i++) {
  38                                                 sequences.set(sequences.size()-1, sequences.get(sequences.size()-1) + parts[i]);
  39                                         }
  40                                 }
  41                         }
  42                         line = inbuf.readLine();
  43                 }
  44                 inbuf.close();
  45
  46                 BufferedWriter outbuf = new BufferedWriter(new FileWriter(outfile));
  47                 for (int i=2; i<seqnames.size(); i++) {
  48                         outbuf.write(">" + seqnames.get(i) + "\n");
  49                         outbuf.write(sequences.get(i) + "\n");
  50                 }
  51                 outbuf.close();
  52         }
  53 }