src/jalview/io/TCoffeeScoreFile.java

   1 package jalview.io;
   2
   3 import java.io.BufferedReader;
   4 import java.io.File;
   5 import java.io.FileNotFoundException;
   6 import java.io.FileReader;
   7 import java.io.IOException;
   8 import java.io.Reader;
   9 import java.util.ArrayList;
  10 import java.util.HashMap;
  11 import java.util.LinkedHashMap;
  12 import java.util.List;
  13 import java.util.Map;
  14
  15 /**
  16  * A file parse for T-Coffee score ascii format. This file contains the alignment consensus
  17  * for each resude in any sequence.
  18  * <p>
  19  * This file is procuded by <code>t_coffee</code> providing the option
  20  * <code>-output=score_ascii </code> to the program command line
  21  *
  22  * An example file is the following
  23  *
  24  * <pre>
  25  * T-COFFEE, Version_9.02.r1228 (2012-02-16 18:15:12 - Revision 1228 - Build 336)
  26  * Cedric Notredame
  27  * CPU TIME:0 sec.
  28  * SCORE=90
  29  * *
  30  *  BAD AVG GOOD
  31  * *
  32  * 1PHT   :  89
  33  * 1BB9   :  90
  34  * 1UHC   :  94
  35  * 1YCS   :  94
  36  * 1OOT   :  93
  37  * 1ABO   :  94
  38  * 1FYN   :  94
  39  * 1QCF   :  94
  40  * cons   :  90
  41  *
  42  * 1PHT   999999999999999999999999998762112222543211112134
  43  * 1BB9   99999999999999999999999999987-------4322----2234
  44  * 1UHC   99999999999999999999999999987-------5321----2246
  45  * 1YCS   99999999999999999999999999986-------4321----1-35
  46  * 1OOT   999999999999999999999999999861-------3------1135
  47  * 1ABO   99999999999999999999999999986-------422-------34
  48  * 1FYN   99999999999999999999999999985-------32--------35
  49  * 1QCF   99999999999999999999999999974-------2---------24
  50  * cons   999999999999999999999999999851000110321100001134
  51  *
  52  *
  53  * 1PHT   ----------5666642367889999999999889
  54  * 1BB9   1111111111676653-355679999999999889
  55  * 1UHC   ----------788774--66789999999999889
  56  * 1YCS   ----------78777--356789999999999889
  57  * 1OOT   ----------78877--356789999999997-67
  58  * 1ABO   ----------687774--56779999999999889
  59  * 1FYN   ----------6888842356789999999999889
  60  * 1QCF   ----------6878742356789999999999889
  61  * cons   00100000006877641356789999999999889
  62  * </pre>
  63  *
  64  *
  65  * @author Paolo Di Tommaso
  66  *
  67  */
  68 public class TCoffeeScoreFile {
  69
  70         /** The {@link Header} structure holder */
  71         Header header;
  72
  73         /**
  74          * Holds the consensues values for each sequences. It uses a LinkedHashMap to maintaint the
  75          * insertion order.
  76          */
  77         LinkedHashMap<String,StringBuilder> scores = new LinkedHashMap<String,StringBuilder>();
  78
  79
  80         /**
  81          * Get the string of score values for the specified seqeunce ID.
  82          * @param id The sequence ID
  83          * @return The scores as a string of values e.g. {@code 99999987-------432}.
  84          *      It return an empty string when the specified ID is missing.
  85          */
  86         public String getScoresFor( String id ) {
  87                 return scores.containsKey(id) ? scores.get(id).toString() : "";
  88         }
  89
  90         /**
  91          * @return The list of score string as a {@link List} object, in the same ordeer of the insertion i.e. in the MSA
  92          */
  93         public List<String> getScoresList() {
  94                 List<String> result = new ArrayList<String>( scores.size() );
  95                 for( Map.Entry<String,StringBuilder> it : scores.entrySet() ) {
  96                         result.add(it.getValue().toString());
  97                 }
  98
  99                 return result;
 100         }
 101
 102         /**
 103          * @return The parsed score values a matrix of bytes
 104          */
 105         public byte[][] getScoresArray() {
 106                 byte[][] result = new byte[ scores.size() ][];
 107
 108                 int rowCount = 0;
 109                 for( Map.Entry<String,StringBuilder> it : scores.entrySet() ) {
 110                         String line = it.getValue().toString();
 111                         byte[] seqValues = new byte[ line.length() ];
 112                         for( int j=0, c=line.length(); j<c; j++ ) {
 113
 114                                 byte val = (byte)(line.charAt(j) - '0');
 115
 116                                 seqValues[j] = ( val >= 0 && val <= 9 ) ? val : -1;
 117                         }
 118
 119                         result[rowCount++] = seqValues;
 120                 }
 121
 122                 return result;
 123         }
 124
 125         /**
 126          * Parse the specified file.
 127          *
 128          * @param file The file to be parsed
 129          */
 130         public void parse(File file) {
 131                 try {
 132                         parse(new FileReader(file));
 133                 }
 134                 catch (FileNotFoundException e) {
 135                         throw new RuntimeException(e);
 136                 }
 137         }
 138
 139         /**
 140          * Parse the provided reader for the T-Coffee scores file format
 141          *
 142          * @param reader
 143          */
 144         public void parse(Reader reader) {
 145
 146                 try {
 147                         BufferedReader in = (BufferedReader) (reader instanceof BufferedReader ? reader : new BufferedReader(reader));
 148                         doParsing(in);
 149                 }
 150                 catch( Exception e) {
 151                         throw new RuntimeException(e);
 152                 }
 153         }
 154
 155         private void doParsing(BufferedReader in) throws IOException {
 156
 157                 /*
 158                  * read the header
 159                  */
 160                 header = readHeader(in);
 161
 162                 /*
 163                  * initilize the structure
 164                  */
 165                 for( Map.Entry<String,Integer> entry : header.scores.entrySet() ) {
 166                         scores.put( entry.getKey(), new StringBuilder());
 167                 }
 168
 169                 /*
 170                  * go with the reading
 171                  */
 172                 Block block;
 173                 while( (block = readBlock(in, header.scores.size())) != null  ) {
 174
 175                         /*
 176                          * append sequences read in the block
 177                          */
 178                         for( Map.Entry<String,String> entry : block.items.entrySet() ) {
 179                                 StringBuilder scoreStringBuilder = scores.get(entry.getKey());
 180                                 if( scoreStringBuilder == null ) {
 181                                         throw new RuntimeException(String.format("Invalid T-Coffee score file. Sequence ID '%s' is not declared in header section", entry.getKey()));
 182                                 }
 183
 184                                 scoreStringBuilder.append( entry.getValue() );
 185                         }
 186
 187                 }
 188
 189         }
 190
 191
 192         static int parseInt( String str ) {
 193                 try {
 194                         return Integer.parseInt(str);
 195                 }
 196                 catch( NumberFormatException e ) {
 197                         // TODO report a warning ?
 198                         return 0;
 199                 }
 200         }
 201
 202         /**
 203          * Reaad the header section in the T-Coffee score file format
 204          *
 205          * @param reader The scores reader
 206          * @return The parser {@link Header} instance
 207          * @throws RuntimeException when the header is not in the expected format
 208          */
 209         static Header readHeader(BufferedReader reader) {
 210
 211                 Header result = null;
 212                 try {
 213                         result = new Header();
 214                         result.head = reader.readLine();
 215
 216                         String line;
 217
 218                         while( (line = reader.readLine()) != null ) {
 219                                 if( line.startsWith("SCORE=")) {
 220                                         result.score = parseInt( line.substring(6).trim() );
 221                                         break;
 222                                 }
 223                         }
 224
 225                         if( (line=reader.readLine())==null || !"*".equals(line.trim())) return null;
 226                         if( (line=reader.readLine())==null || !"BAD AVG GOOD".equals(line.trim())) return null;
 227                         if( (line=reader.readLine())==null || !"*".equals(line.trim())) return null;
 228
 229                         /*
 230                          * now are expected a list if sequences ID up to the first blank line
 231                          */
 232                         while( (line=reader.readLine()) != null ) {
 233                                 if( "".equals(line) ) {
 234                                         break;
 235                                 }
 236
 237                                 int p = line.indexOf(":");
 238                                 if( p == -1 ) {
 239                                         // TODO report a warning
 240                                         continue;
 241                                 }
 242
 243                                 String id = line.substring(0,p).trim();
 244                                 int val = parseInt(line.substring(p+1).trim());
 245                                 if( "".equals(id) ) {
 246                                         // TODO report warning
 247                                         continue;
 248                                 }
 249
 250                                 result.scores.put(id,val);
 251                         }
 252
 253                 }
 254                 catch( IOException e ) {
 255                         throw new RuntimeException("Cannot parse T-Coffee score ascii file", e);
 256                 }
 257
 258                 return result;
 259         }
 260
 261         /**
 262          * Read a scores block ihe provided stream.
 263          *
 264          * @param reader The stream to parse
 265          * @param size The expected number of the sequence to be read
 266          * @return The {@link Block} instance read or {link null} null if the end of file has reached.
 267          * @throws IOException Something went wrong on the 'wire'
 268          */
 269         static Block readBlock( BufferedReader reader, int size ) throws IOException {
 270                 Block result = new Block(size);
 271                 String line;
 272
 273                 /*
 274                  * read blank lines (eventually)
 275                  */
 276                 while( (line=reader.readLine()) != null && "".equals(line.trim())) {
 277                         // consume blank lines
 278                 }
 279
 280                 if( line == null ) return null;
 281
 282                 /*
 283                  * read the scores block
 284                  */
 285                 do {
 286                         if( "".equals(line.trim()) ) {
 287                                 // terminated
 288                                 break;
 289                         }
 290
 291                         // split the line on the first blank
 292                         // the first part have to contain the sequence id
 293                         // theramining part are the scores values
 294                         int p = line.indexOf(" ");
 295                         if( p == -1 ) {
 296                                 //TODO This is an unexpected condition, log a warning or throw an exception ?
 297                                 continue;
 298                         }
 299
 300                         String id = line.substring(0,p).trim();
 301                         String val = line.substring(p+1).trim();
 302
 303                         result.items.put(id, val);
 304
 305                 } while( (line = reader.readLine()) != null );
 306
 307
 308                 return result;
 309         }
 310
 311         /*
 312          * The score file header
 313          */
 314         static class Header {
 315                 String head;
 316                 int score;
 317
 318                 LinkedHashMap<String,Integer> scores = new LinkedHashMap<String,Integer>();
 319
 320                 public int getScoreAvg() { return score; }
 321
 322                 public int getScoreFor( String ID ) {
 323
 324                         return scores.containsKey(ID) ? scores.get(ID) : -1;
 325
 326                 }
 327         }
 328
 329         /*
 330          * Hold a single block values block in the score file
 331          */
 332         static class Block {
 333                 int size;
 334                 Map<String,String> items;
 335
 336                 public Block( int size ) {
 337                         this.size = size;
 338                         this.items = new HashMap<String,String>(size);
 339                 }
 340
 341                 String getScoresFor( String id ) {
 342                         return items.get(id);
 343                 }
 344
 345                 String getConsensus() {
 346                         return items.get("cons");
 347                 }
 348         }
 349
 350
 351
 352 }