src/jalview/io/SequenceFeatureFetcher.java

   1 /*\r
   2 * Jalview - A Sequence Alignment Editor and Viewer\r
   3 * Copyright (C) 2005 AM Waterhouse, J Procter, G Barton, M Clamp, S Searle\r
   4 *\r
   5 * This program is free software; you can redistribute it and/or\r
   6 * modify it under the terms of the GNU General Public License\r
   7 * as published by the Free Software Foundation; either version 2\r
   8 * of the License, or (at your option) any later version.\r
   9 *\r
  10 * This program is distributed in the hope that it will be useful,\r
  11 * but WITHOUT ANY WARRANTY; without even the implied warranty of\r
  12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the\r
  13 * GNU General Public License for more details.\r
  14 *\r
  15 * You should have received a copy of the GNU General Public License\r
  16 * along with this program; if not, write to the Free Software\r
  17 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA\r
  18 */\r
  19 package jalview.io;\r
  20 \r
  21 import jalview.datamodel.*;\r
  22 \r
  23 import jalview.gui.*;\r
  24 \r
  25 import jalview.io.*;\r
  26 \r
  27 import java.io.*;\r
  28 \r
  29 import java.util.*;\r
  30 \r
  31 \r
  32 /**\r
  33  * DOCUMENT ME!\r
  34  *\r
  35  * @author $author$\r
  36  * @version $Revision$\r
  37  */\r
  38 public class SequenceFeatureFetcher implements Runnable\r
  39 {\r
  40     AlignmentI align;\r
  41     AlignmentPanel ap;\r
  42     ArrayList unknownSequences;\r
  43     CutAndPasteTransfer output = new CutAndPasteTransfer();\r
  44     StringBuffer sbuffer = new StringBuffer();\r
  45 \r
  46     /**\r
  47      * Creates a new SequenceFeatureFetcher object.\r
  48      *\r
  49      * @param align DOCUMENT ME!\r
  50      * @param ap DOCUMENT ME!\r
  51      */\r
  52     public SequenceFeatureFetcher(AlignmentI align, AlignmentPanel ap)\r
  53     {\r
  54         unknownSequences = new ArrayList();\r
  55         this.align = align;\r
  56         this.ap = ap;\r
  57 \r
  58         Thread thread = new Thread(this);\r
  59         thread.start();\r
  60     }\r
  61 \r
  62     /**\r
  63      * DOCUMENT ME!\r
  64      */\r
  65     public void run()\r
  66     {\r
  67         RandomAccessFile out = null;\r
  68 \r
  69         try\r
  70         {\r
  71             String cache = System.getProperty("user.home") +\r
  72                 "/.jalview.uniprot.xml";\r
  73 \r
  74             File test = new File(cache);\r
  75 \r
  76             if (!test.exists())\r
  77             {\r
  78                 out = new RandomAccessFile(cache, "rw");\r
  79                 out.writeBytes("<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n");\r
  80                 out.writeBytes("<UNIPROT_CACHE>\n");\r
  81             }\r
  82             else\r
  83             {\r
  84                 out = new RandomAccessFile(cache, "rw");\r
  85 \r
  86                 // open exisiting cache and remove </UNIPROT_CACHE> from the end\r
  87                 long lastLine = 0;\r
  88                 String data;\r
  89 \r
  90                 while ((data = out.readLine()) != null)\r
  91                 {\r
  92                     if (data.indexOf("</entry>") > -1)\r
  93                     {\r
  94                         lastLine = out.getFilePointer();\r
  95                     }\r
  96                 }\r
  97 \r
  98                 out.seek(lastLine);\r
  99             }\r
 100 \r
 101             int seqIndex = 0;\r
 102             Vector sequences = align.getSequences();\r
 103 \r
 104             while (seqIndex < sequences.size())\r
 105             {\r
 106                 ArrayList ids = new ArrayList();\r
 107 \r
 108                 for (int i = 0; (seqIndex < sequences.size()) && (i < 50);\r
 109                         seqIndex++, i++)\r
 110                 {\r
 111                     SequenceI sequence = (SequenceI) sequences.get(seqIndex);\r
 112                     ids.add(sequence.getName());\r
 113                 }\r
 114 \r
 115                 tryLocalCacheFirst(ids, align);\r
 116 \r
 117                 if (ids.size() > 0)\r
 118                 {\r
 119                     StringBuffer remainingIds = new StringBuffer("uniprot:");\r
 120 \r
 121                     for (int i = 0; i < ids.size(); i++)\r
 122                         remainingIds.append(ids.get(i) + ";");\r
 123 \r
 124                     EBIFetchClient ebi = new EBIFetchClient();\r
 125                     String[] result = ebi.fetchData(remainingIds.toString(),\r
 126                             "xml", null);\r
 127 \r
 128                     if (result != null)\r
 129                     {\r
 130                         ReadUniprotFile(result, out, align);\r
 131                     }\r
 132                 }\r
 133             }\r
 134 \r
 135             if (out != null)\r
 136             {\r
 137                 out.writeBytes("</UNIPROT_CACHE>\n");\r
 138                 out.close();\r
 139             }\r
 140         }\r
 141         catch (Exception ex)\r
 142         {\r
 143             ex.printStackTrace();\r
 144         }\r
 145 \r
 146         findMissingIds(align);\r
 147 \r
 148         if (sbuffer.length() > 0)\r
 149         {\r
 150             output.setText(\r
 151                 "Your sequences have been matched to Uniprot. Some of the ids have been\n" +\r
 152                 "altered, most likely the start/end residue will have been updated.\n" +\r
 153                 "Save your alignment to maintain the updated id.\n\n" +\r
 154                 sbuffer.toString());\r
 155             Desktop.addInternalFrame(output, "Sequence names updated ", 600, 300);\r
 156         }\r
 157 \r
 158         if (unknownSequences.size() > 0)\r
 159         {\r
 160             //ignore for now!!!!!!!!!!\r
 161             //  WSWUBlastClient blastClient = new WSWUBlastClient(align, unknownSequences);\r
 162         }\r
 163 \r
 164         jalview.gui.PaintRefresher.Refresh(null, align);\r
 165     }\r
 166 \r
 167     /**\r
 168      * DOCUMENT ME!\r
 169      *\r
 170      * @param result DOCUMENT ME!\r
 171      * @param out DOCUMENT ME!\r
 172      * @param align DOCUMENT ME!\r
 173      */\r
 174     void ReadUniprotFile(String[] result, RandomAccessFile out, AlignmentI align)\r
 175     {\r
 176         SequenceI sequence = null;\r
 177         Vector features = null;\r
 178         String type;\r
 179         String description;\r
 180         String status;\r
 181         String start;\r
 182         String end;\r
 183         String pdb = null;\r
 184 \r
 185         for (int r = 0; r < result.length; r++)\r
 186         {\r
 187             if ((sequence == null) && (result[r].indexOf("<name>") > -1))\r
 188             {\r
 189                 long filePointer = 0;\r
 190 \r
 191                 if (out != null)\r
 192                 {\r
 193                     try\r
 194                     {\r
 195                         filePointer = out.getFilePointer();\r
 196                         out.writeBytes("<entry>\n");\r
 197                     }\r
 198                     catch (Exception ex)\r
 199                     {\r
 200                     }\r
 201                 }\r
 202 \r
 203                 String seqName = parseElement(result[r], "<name>", out);\r
 204                 sequence = align.findName(seqName);\r
 205 \r
 206                 if (sequence == null)\r
 207                 {\r
 208                     sequence = align.findName(seqName.substring(0,\r
 209                                 seqName.indexOf('_')));\r
 210 \r
 211                     if (sequence != null)\r
 212                     {\r
 213                         sbuffer.append("changing " + sequence.getName() +\r
 214                             " to " + seqName + "\n");\r
 215                         sequence.setName(seqName);\r
 216                     }\r
 217                 }\r
 218 \r
 219                 if (sequence == null)\r
 220                 {\r
 221                     sbuffer.append("UNIPROT updated suggestion is " +\r
 222                         result[r] + "\n");\r
 223                     sequence = align.findName(result[r]);\r
 224 \r
 225                     // this entry has been suggested by ebi.\r
 226                     // doesn't match id in alignment file\r
 227                     try\r
 228                     {\r
 229                         out.setLength(filePointer);\r
 230                     }\r
 231                     catch (Exception ex)\r
 232                     {\r
 233                     }\r
 234 \r
 235                     // now skip to next entry\r
 236                     while (result[r].indexOf("</entry>") == -1)\r
 237                         r++;\r
 238                 }\r
 239 \r
 240                 features = new Vector();\r
 241                 type = "";\r
 242                 start = "0";\r
 243                 end = "0";\r
 244                 description = "";\r
 245                 status = "";\r
 246                 pdb = "";\r
 247             }\r
 248 \r
 249             if (sequence == null)\r
 250             {\r
 251                 continue;\r
 252             }\r
 253 \r
 254             if (result[r].indexOf("<property type=\"pdb accession\"") > -1)\r
 255             {\r
 256                 pdb = parseValue(result[r], "value=", out);\r
 257                 sequence.setPDBId(pdb);\r
 258             }\r
 259 \r
 260             if (result[r].indexOf("feature type") > -1)\r
 261             {\r
 262                 type = parseValue(result[r], "type=", out);\r
 263                 description = parseValue(result[r], "description=", null);\r
 264                 status = parseValue(result[r], "status=", null);\r
 265 \r
 266                 while (result[r].indexOf("position") == -1)\r
 267                 {\r
 268                     r++; //<location>\r
 269                 }\r
 270 \r
 271                 // r++;\r
 272                 if (result[r].indexOf("begin") > -1)\r
 273                 {\r
 274                     start = parseValue(result[r], "position=", out);\r
 275                     end = parseValue(result[++r], "position=", out);\r
 276                 }\r
 277                 else\r
 278                 {\r
 279                     start = parseValue(result[r], "position=", out);\r
 280                     end = parseValue(result[r], "position=", null);\r
 281                 }\r
 282 \r
 283                 int sstart = Integer.parseInt(start);\r
 284                 int eend = Integer.parseInt(end);\r
 285 \r
 286                 if (out != null)\r
 287                 {\r
 288                     try\r
 289                     {\r
 290                         out.writeBytes("</feature>\n");\r
 291                     }\r
 292                     catch (Exception ex)\r
 293                     {\r
 294                     }\r
 295                 }\r
 296 \r
 297                 SequenceFeature sf = new SequenceFeature(type, sstart, eend,\r
 298                         description, status);\r
 299                 features.add(sf);\r
 300             }\r
 301 \r
 302             if (result[r].indexOf("<sequence") > -1)\r
 303             {\r
 304                 StringBuffer seqString = new StringBuffer();\r
 305 \r
 306                 if (out != null)\r
 307                 {\r
 308                     try\r
 309                     {\r
 310                         out.writeBytes(result[r] + "\n");\r
 311                     }\r
 312                     catch (Exception ex)\r
 313                     {\r
 314                     }\r
 315                 }\r
 316 \r
 317                 while (result[++r].indexOf("</sequence>") == -1)\r
 318                 {\r
 319                     seqString.append(result[r]);\r
 320 \r
 321                     if (out != null)\r
 322                     {\r
 323                         try\r
 324                         {\r
 325                             out.writeBytes(result[r] + "\n");\r
 326                         }\r
 327                         catch (Exception ex)\r
 328                         {\r
 329                         }\r
 330                     }\r
 331                 }\r
 332 \r
 333                 if (out != null)\r
 334                 {\r
 335                     try\r
 336                     {\r
 337                         out.writeBytes(result[r] + "\n");\r
 338                     }\r
 339                     catch (Exception ex)\r
 340                     {\r
 341                     }\r
 342                 }\r
 343 \r
 344                 StringBuffer nonGapped = new StringBuffer();\r
 345 \r
 346                 for (int i = 0; i < sequence.getSequence().length(); i++)\r
 347                 {\r
 348                     if (!jalview.util.Comparison.isGap(sequence.getCharAt(i)))\r
 349                     {\r
 350                         nonGapped.append(sequence.getCharAt(i));\r
 351                     }\r
 352                 }\r
 353 \r
 354                 int absStart = seqString.toString().indexOf(nonGapped.toString());\r
 355 \r
 356                 if (absStart == -1)\r
 357                 {\r
 358                     unknownSequences.add(sequence.getName());\r
 359                     features = null;\r
 360                     sbuffer.append(sequence.getName() +\r
 361                         " SEQUENCE NOT %100 MATCH \n");\r
 362 \r
 363                     continue;\r
 364                 }\r
 365 \r
 366                 int absEnd = absStart + nonGapped.toString().length();\r
 367                 absStart += 1;\r
 368 \r
 369                 if ((absStart != sequence.getStart()) ||\r
 370                         (absEnd != sequence.getEnd()))\r
 371                 {\r
 372                     sbuffer.append("Updated: " + sequence.getName() + " " +\r
 373                         sequence.getStart() + "/" + sequence.getEnd() +\r
 374                         "  to  " + absStart + "/" + absEnd + "\n");\r
 375                 }\r
 376 \r
 377                 sequence.setStart(absStart);\r
 378                 sequence.setEnd(absEnd);\r
 379             }\r
 380 \r
 381             if (result[r].indexOf("</entry>") > -1)\r
 382             {\r
 383                 if (features != null)\r
 384                 {\r
 385                     sequence.setSequenceFeatures(features);\r
 386                 }\r
 387 \r
 388                 features = null;\r
 389                 sequence = null;\r
 390 \r
 391                 if (out != null)\r
 392                 {\r
 393                     try\r
 394                     {\r
 395                         out.writeBytes("</entry>\n");\r
 396                     }\r
 397                     catch (Exception ex)\r
 398                     {\r
 399                     }\r
 400                 }\r
 401             }\r
 402         }\r
 403     }\r
 404 \r
 405     /**\r
 406      * DOCUMENT ME!\r
 407      *\r
 408      * @param align DOCUMENT ME!\r
 409      */\r
 410     void findMissingIds(AlignmentI align)\r
 411     {\r
 412         String data;\r
 413         ArrayList cachedIds = new ArrayList();\r
 414 \r
 415         try\r
 416         {\r
 417             if(jalview.bin.Cache.getProperty("UNIPROT_CACHE")==null)\r
 418               return;\r
 419 \r
 420             BufferedReader in = new BufferedReader(new FileReader(\r
 421                         jalview.bin.Cache.getProperty("UNIPROT_CACHE")));\r
 422 \r
 423             while ((data = in.readLine()) != null)\r
 424             {\r
 425                 if (data.indexOf("name") > -1)\r
 426                 {\r
 427                     String name = parseElement(data, "<name>", null);\r
 428                     cachedIds.add(name);\r
 429                 }\r
 430             }\r
 431         }\r
 432         catch (Exception ex)\r
 433         {\r
 434             ex.printStackTrace();\r
 435         }\r
 436 \r
 437         for (int i = 0; i < align.getHeight(); i++)\r
 438             if (!cachedIds.contains(align.getSequenceAt(i).getName()))\r
 439             {\r
 440                 unknownSequences.add(align.getSequenceAt(i).getName());\r
 441             }\r
 442     }\r
 443 \r
 444     /**\r
 445      * DOCUMENT ME!\r
 446      *\r
 447      * @param ids DOCUMENT ME!\r
 448      * @param align DOCUMENT ME!\r
 449      */\r
 450     void tryLocalCacheFirst(ArrayList ids, AlignmentI align)\r
 451     {\r
 452         ArrayList cacheData = new ArrayList();\r
 453 \r
 454         try\r
 455         {\r
 456             if(jalview.bin.Cache.getProperty("UNIPROT_CACHE")==null)\r
 457               return;\r
 458 \r
 459             BufferedReader in = new BufferedReader(new FileReader(\r
 460                         jalview.bin.Cache.getProperty("UNIPROT_CACHE")));\r
 461 \r
 462             // read through cache file, if the cache has sequences we're looking for\r
 463             // add the lines to a new String array, Readthis new array and\r
 464             // make sure we remove the ids from the list to retrieve from EBI\r
 465             String data;\r
 466 \r
 467             while ((data = in.readLine()) != null)\r
 468             {\r
 469                 if (data.indexOf("name") > -1)\r
 470                 {\r
 471                     String name = parseElement(data, "<name>", null);\r
 472 \r
 473                     if (ids.contains(name))\r
 474                     {\r
 475                         cacheData.add("<entry>");\r
 476                         cacheData.add(data);\r
 477 \r
 478                         while (data.indexOf("</entry>") == -1)\r
 479                         {\r
 480                             data = in.readLine();\r
 481                             cacheData.add(data);\r
 482                         }\r
 483 \r
 484                         cacheData.add(data);\r
 485 \r
 486                         ids.remove(name);\r
 487                     }\r
 488                 }\r
 489             }\r
 490         }\r
 491         catch (Exception ex)\r
 492         {\r
 493             ex.printStackTrace();\r
 494         }\r
 495 \r
 496         String[] localData = new String[cacheData.size()];\r
 497         cacheData.toArray(localData);\r
 498 \r
 499         if ((localData != null) && (localData.length > 0))\r
 500         {\r
 501             ReadUniprotFile(localData, null, align);\r
 502         }\r
 503     }\r
 504 \r
 505     /**\r
 506      * DOCUMENT ME!\r
 507      *\r
 508      * @param line DOCUMENT ME!\r
 509      * @param tag DOCUMENT ME!\r
 510      * @param out DOCUMENT ME!\r
 511      *\r
 512      * @return DOCUMENT ME!\r
 513      */\r
 514     String parseValue(String line, String tag, RandomAccessFile out)\r
 515     {\r
 516         if (out != null)\r
 517         {\r
 518             try\r
 519             {\r
 520                 out.writeBytes(line + "\n");\r
 521             }\r
 522             catch (Exception ex)\r
 523             {\r
 524             }\r
 525         }\r
 526 \r
 527         int index = line.indexOf(tag) + tag.length() + 1;\r
 528 \r
 529         if (index == tag.length())\r
 530         {\r
 531             return "";\r
 532         }\r
 533 \r
 534         return line.substring(index, line.indexOf("\"", index + 1));\r
 535     }\r
 536 \r
 537     /**\r
 538      * DOCUMENT ME!\r
 539      *\r
 540      * @param line DOCUMENT ME!\r
 541      * @param tag DOCUMENT ME!\r
 542      * @param out DOCUMENT ME!\r
 543      *\r
 544      * @return DOCUMENT ME!\r
 545      */\r
 546     String parseElement(String line, String tag, RandomAccessFile out)\r
 547     {\r
 548         if (out != null)\r
 549         {\r
 550             try\r
 551             {\r
 552                 out.writeBytes(line + "\n");\r
 553             }\r
 554             catch (Exception ex)\r
 555             {\r
 556             }\r
 557         }\r
 558 \r
 559         int index = line.indexOf(tag) + tag.length();\r
 560 \r
 561         return line.substring(index, line.indexOf("</"));\r
 562     }\r
 563 }\r