srcjar/fr/orsay/lri/varna/factories/StockholmIO.java

   1 package fr.orsay.lri.varna.factories;
   2
   3 import java.io.BufferedReader;
   4 import java.io.FileNotFoundException;
   5 import java.io.FileReader;
   6 import java.io.IOException;
   7 import java.io.InputStreamReader;
   8 import java.io.Reader;
   9 import java.io.UnsupportedEncodingException;
  10 import java.net.MalformedURLException;
  11 import java.net.URL;
  12 import java.net.URLConnection;
  13 import java.util.ArrayList;
  14 import java.util.HashMap;
  15 import java.util.LinkedHashMap;
  16 import java.util.Map;
  17
  18
  19 public class StockholmIO {
  20         public static RNAAlignment readAlignementFromFile(String path) throws IOException
  21         {
  22                 return StockholmIO.readAlignement(new BufferedReader(new FileReader(path)));
  23         }
  24         public static RNAAlignment readAlignementFromURL(String url) throws UnsupportedEncodingException, IOException
  25         {
  26                 URL urlAb = new URL(url);
  27                 URLConnection urlConn = urlAb.openConnection();
  28                 urlConn.setUseCaches(false);
  29                 Reader r = new InputStreamReader(urlConn.getInputStream(),"UTF-8");
  30                 return readAlignement(new BufferedReader(r));
  31         }
  32
  33
  34         /*public static Alignment readAlignement(Reader r) throws IOException
  35         {
  36                 return readAlignement(new BufferedReader(r));
  37         }*/
  38
  39         public static RNAAlignment readAlignement(BufferedReader r) throws IOException
  40         {
  41                 LinkedHashMap<String,StringBuffer> rawSeqs = new LinkedHashMap<String,StringBuffer>();
  42                 RNAAlignment result = new RNAAlignment();
  43                 String line = r.readLine();
  44                 String str = "";
  45
  46                 while(line!=null)
  47                 {
  48                         if (!line.startsWith("#"))
  49                         {
  50                                 String[] data = line.split("\\s+");
  51                                 if (data.length>1)
  52                                 {
  53                                         String seqName = data[0].trim();
  54                                         String seq = data[1].trim();
  55                                         if (!rawSeqs.containsKey(seqName))
  56                                         {
  57                                                 rawSeqs.put(seqName,new StringBuffer());
  58                                         }
  59                                         StringBuffer val =  rawSeqs.get(seqName);
  60                                         val.append(seq);
  61                                 }
  62
  63                         }
  64                         else if (line.startsWith("#"))
  65                         {
  66                                 String[] data = line.split("\\s+");
  67                                 if (line.startsWith("#=GC SS_cons"))
  68                                 {
  69                                         str += data[2].trim();
  70                                 }
  71                                 else if (line.startsWith("#=GS"))
  72                                 {
  73                                         if (data[2].trim().equals("AC"))
  74                                         {
  75                                           result.setAccession(data[1].trim(),data[3].trim());
  76
  77                                         }
  78                                 }
  79                         }
  80
  81                         line = r.readLine();
  82                 }
  83                 result.setSecStr(str);
  84                 for(Map.Entry<String,StringBuffer> entry : rawSeqs.entrySet())
  85                 {
  86                         String s = entry.getValue().toString();
  87                         result.addSequence(entry.getKey(), s);
  88                 }
  89                 return result;
  90         }
  91 }