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