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 import javax.swing.*;\r
  32 \r
  33 \r
  34 public class SequenceFeatureFetcher implements Runnable {\r
  35     AlignmentI align;\r
  36     AlignmentPanel ap;\r
  37     ArrayList unknownSequences;\r
  38     CutAndPasteTransfer output = new CutAndPasteTransfer();\r
  39     StringBuffer sbuffer = new StringBuffer();\r
  40 \r
  41     public SequenceFeatureFetcher(AlignmentI align, AlignmentPanel ap) {\r
  42         unknownSequences = new ArrayList();\r
  43         this.align = align;\r
  44         this.ap = ap;\r
  45 \r
  46         Thread thread = new Thread(this);\r
  47         thread.start();\r
  48     }\r
  49 \r
  50     public void run() {\r
  51         String cache = jalview.bin.Cache.getProperty("UNIPROT_CACHE");\r
  52 \r
  53         RandomAccessFile out = null;\r
  54 \r
  55         try {\r
  56             if (cache == null) {\r
  57                 jalview.bin.Cache.setProperty("UNIPROT_CACHE",\r
  58                     System.getProperty("user.home") + "/uniprot.xml");\r
  59                 cache = jalview.bin.Cache.getProperty("UNIPROT_CACHE");\r
  60             }\r
  61 \r
  62             File test = new File(cache);\r
  63 \r
  64             if (!test.exists()) {\r
  65                 out = new RandomAccessFile(cache, "rw");\r
  66                 out.writeBytes("<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n");\r
  67                 out.writeBytes("<UNIPROT_CACHE>\n");\r
  68             } else {\r
  69                 out = new RandomAccessFile(cache, "rw");\r
  70 \r
  71                 // open exisiting cache and remove </UNIPROT_CACHE> from the end\r
  72                 long lastLine = 0;\r
  73                 String data;\r
  74 \r
  75                 while ((data = out.readLine()) != null) {\r
  76                     if (data.indexOf("</entry>") > -1) {\r
  77                         lastLine = out.getFilePointer();\r
  78                     }\r
  79                 }\r
  80 \r
  81                 out.seek(lastLine);\r
  82             }\r
  83 \r
  84             int seqIndex = 0;\r
  85             Vector sequences = align.getSequences();\r
  86 \r
  87             while (seqIndex < sequences.size()) {\r
  88                 ArrayList ids = new ArrayList();\r
  89 \r
  90                 for (int i = 0; (seqIndex < sequences.size()) && (i < 50);\r
  91                         seqIndex++, i++) {\r
  92                     SequenceI sequence = (SequenceI) sequences.get(seqIndex);\r
  93                     ids.add(sequence.getName());\r
  94                 }\r
  95 \r
  96                 tryLocalCacheFirst(ids, align);\r
  97 \r
  98                 if (ids.size() > 0) {\r
  99                     StringBuffer remainingIds = new StringBuffer("uniprot:");\r
 100 \r
 101                     for (int i = 0; i < ids.size(); i++)\r
 102                         remainingIds.append(ids.get(i) + ";");\r
 103 \r
 104                     EBIFetchClient ebi = new EBIFetchClient();\r
 105                     String[] result = ebi.fetchData(remainingIds.toString(),\r
 106                             "xml", null);\r
 107 \r
 108                     if (result != null) {\r
 109                         ReadUniprotFile(result, out, align);\r
 110                     }\r
 111                 }\r
 112             }\r
 113 \r
 114             if (out != null) {\r
 115                 out.writeBytes("</UNIPROT_CACHE>\n");\r
 116                 out.close();\r
 117             }\r
 118         } catch (Exception ex) {\r
 119             ex.printStackTrace();\r
 120         }\r
 121 \r
 122         ap.repaint();\r
 123         findMissingIds(align);\r
 124 \r
 125         if (sbuffer.length() > 0) {\r
 126             output.setText(\r
 127                 "Your sequences have been matched to Uniprot. Some of the ids have been\n" +\r
 128                 "altered, most likely the start/end residue will have been updated.\n" +\r
 129                 "Save your alignment to maintain the updated id.\n\n" +\r
 130                 sbuffer.toString());\r
 131             Desktop.addInternalFrame(output, "Sequence names updated ", 600, 300);\r
 132         }\r
 133 \r
 134         if (unknownSequences.size() > 0) {\r
 135             //ignore for now!!!!!!!!!!\r
 136             //  WSWUBlastClient blastClient = new WSWUBlastClient(align, unknownSequences);\r
 137         }\r
 138     }\r
 139 \r
 140     void ReadUniprotFile(String[] result, RandomAccessFile out, AlignmentI align) {\r
 141         SequenceI sequence = null;\r
 142         Vector features = null;\r
 143         String type;\r
 144         String description;\r
 145         String status;\r
 146         String start;\r
 147         String end;\r
 148         String pdb = null;\r
 149 \r
 150         for (int r = 0; r < result.length; r++) {\r
 151             if ((sequence == null) && (result[r].indexOf("<name>") > -1)) {\r
 152                 long filePointer = 0;\r
 153 \r
 154                 if (out != null) {\r
 155                     try {\r
 156                         filePointer = out.getFilePointer();\r
 157                         out.writeBytes("<entry>\n");\r
 158                     } catch (Exception ex) {\r
 159                     }\r
 160                 }\r
 161 \r
 162                 String seqName = parseElement(result[r], "<name>", out);\r
 163                 sequence = align.findName(seqName);\r
 164 \r
 165                 if (sequence == null) {\r
 166                     sequence = align.findName(seqName.substring(0,\r
 167                                 seqName.indexOf('_')));\r
 168 \r
 169                     if (sequence != null) {\r
 170                         sbuffer.append("changing " + sequence.getName() +\r
 171                             " to " + seqName + "\n");\r
 172                         sequence.setName(seqName);\r
 173                     }\r
 174                 }\r
 175 \r
 176                 if (sequence == null) {\r
 177                     sbuffer.append("UNIPROT updated suggestion is " +\r
 178                         result[r] + "\n");\r
 179                     sequence = align.findName(result[r]);\r
 180 \r
 181                     // this entry has been suggested by ebi.\r
 182                     // doesn't match id in alignment file\r
 183                     try {\r
 184                         out.setLength(filePointer);\r
 185                     } catch (Exception ex) {\r
 186                     }\r
 187 \r
 188                     // now skip to next entry\r
 189                     while (result[r].indexOf("</entry>") == -1)\r
 190                         r++;\r
 191                 }\r
 192 \r
 193                 features = new Vector();\r
 194                 type = "";\r
 195                 start = "0";\r
 196                 end = "0";\r
 197                 description = "";\r
 198                 status = "";\r
 199                 pdb = "";\r
 200             }\r
 201 \r
 202             if (sequence == null) {\r
 203                 continue;\r
 204             }\r
 205 \r
 206             if (result[r].indexOf("<property type=\"pdb accession\"") > -1) {\r
 207                 pdb = parseValue(result[r], "value=", out);\r
 208                 sequence.setPDBId(pdb);\r
 209             }\r
 210 \r
 211             if (result[r].indexOf("feature type") > -1) {\r
 212                 type = parseValue(result[r], "type=", out);\r
 213                 description = parseValue(result[r], "description=", null);\r
 214                 status = parseValue(result[r], "status=", null);\r
 215 \r
 216                 while (result[r].indexOf("position") == -1) {\r
 217                     r++; //<location>\r
 218                 }\r
 219 \r
 220                 // r++;\r
 221                 if (result[r].indexOf("begin") > -1) {\r
 222                     start = parseValue(result[r], "position=", out);\r
 223                     end = parseValue(result[++r], "position=", out);\r
 224                 } else {\r
 225                     start = parseValue(result[r], "position=", out);\r
 226                     end = parseValue(result[r], "position=", null);\r
 227                 }\r
 228 \r
 229                 int sstart = Integer.parseInt(start);\r
 230                 int eend = Integer.parseInt(end);\r
 231 \r
 232                 if (out != null) {\r
 233                     try {\r
 234                         out.writeBytes("</feature>\n");\r
 235                     } catch (Exception ex) {\r
 236                     }\r
 237                 }\r
 238 \r
 239                 SequenceFeature sf = new SequenceFeature(type, sstart, eend,\r
 240                         description, status);\r
 241                 features.add(sf);\r
 242             }\r
 243 \r
 244             if (result[r].indexOf("<sequence") > -1) {\r
 245                 StringBuffer seqString = new StringBuffer();\r
 246 \r
 247                 if (out != null) {\r
 248                     try {\r
 249                         out.writeBytes(result[r] + "\n");\r
 250                     } catch (Exception ex) {\r
 251                     }\r
 252                 }\r
 253 \r
 254                 while (result[++r].indexOf("</sequence>") == -1) {\r
 255                     seqString.append(result[r]);\r
 256 \r
 257                     if (out != null) {\r
 258                         try {\r
 259                             out.writeBytes(result[r] + "\n");\r
 260                         } catch (Exception ex) {\r
 261                         }\r
 262                     }\r
 263                 }\r
 264 \r
 265                 if (out != null) {\r
 266                     try {\r
 267                         out.writeBytes(result[r] + "\n");\r
 268                     } catch (Exception ex) {\r
 269                     }\r
 270                 }\r
 271 \r
 272                 StringBuffer nonGapped = new StringBuffer();\r
 273 \r
 274                 for (int i = 0; i < sequence.getSequence().length(); i++) {\r
 275                     if (!jalview.util.Comparison.isGap(sequence.getCharAt(i))) {\r
 276                         nonGapped.append(sequence.getCharAt(i));\r
 277                     }\r
 278                 }\r
 279 \r
 280                 int absStart = seqString.toString().indexOf(nonGapped.toString());\r
 281 \r
 282                 if (absStart == -1) {\r
 283                     unknownSequences.add(sequence.getName());\r
 284                     features = null;\r
 285                     sbuffer.append(sequence.getName() +\r
 286                         " SEQUENCE NOT %100 MATCH \n");\r
 287 \r
 288                     continue;\r
 289                 }\r
 290 \r
 291                 int absEnd = absStart + nonGapped.toString().length();\r
 292                 absStart += 1;\r
 293 \r
 294                 if ((absStart != sequence.getStart()) ||\r
 295                         (absEnd != sequence.getEnd())) {\r
 296                     sbuffer.append("Updated: " + sequence.getName() + " " +\r
 297                         sequence.getStart() + "/" + sequence.getEnd() +\r
 298                         "  to  " + absStart + "/" + absEnd + "\n");\r
 299                 }\r
 300 \r
 301                 sequence.setStart(absStart);\r
 302                 sequence.setEnd(absEnd);\r
 303             }\r
 304 \r
 305             if (result[r].indexOf("</entry>") > -1) {\r
 306                 if (features != null) {\r
 307                     sequence.setSequenceFeatures(features);\r
 308                 }\r
 309 \r
 310                 features = null;\r
 311                 sequence = null;\r
 312 \r
 313                 if (out != null) {\r
 314                     try {\r
 315                         out.writeBytes("</entry>\n");\r
 316                     } catch (Exception ex) {\r
 317                     }\r
 318                 }\r
 319             }\r
 320         }\r
 321     }\r
 322 \r
 323     void findMissingIds(AlignmentI align) {\r
 324         String data;\r
 325         ArrayList cachedIds = new ArrayList();\r
 326 \r
 327         try {\r
 328             BufferedReader in = new BufferedReader(new FileReader(\r
 329                         jalview.bin.Cache.getProperty("UNIPROT_CACHE")));\r
 330 \r
 331             while ((data = in.readLine()) != null) {\r
 332                 if (data.indexOf("name") > -1) {\r
 333                     String name = parseElement(data, "<name>", null);\r
 334                     cachedIds.add(name);\r
 335                 }\r
 336             }\r
 337         } catch (Exception ex) {\r
 338             ex.printStackTrace();\r
 339         }\r
 340 \r
 341         for (int i = 0; i < align.getHeight(); i++)\r
 342             if (!cachedIds.contains(align.getSequenceAt(i).getName())) {\r
 343                 unknownSequences.add(align.getSequenceAt(i).getName());\r
 344             }\r
 345     }\r
 346 \r
 347     void tryLocalCacheFirst(ArrayList ids, AlignmentI align) {\r
 348         ArrayList cacheData = new ArrayList();\r
 349 \r
 350         try {\r
 351             BufferedReader in = new BufferedReader(new FileReader(\r
 352                         jalview.bin.Cache.getProperty("UNIPROT_CACHE")));\r
 353 \r
 354             // read through cache file, if the cache has sequences we're looking for\r
 355             // add the lines to a new String array, Readthis new array and\r
 356             // make sure we remove the ids from the list to retrieve from EBI\r
 357             String data;\r
 358 \r
 359             while ((data = in.readLine()) != null) {\r
 360                 if (data.indexOf("name") > -1) {\r
 361                     String name = parseElement(data, "<name>", null);\r
 362 \r
 363                     if (ids.contains(name)) {\r
 364                         cacheData.add("<entry>");\r
 365                         cacheData.add(data);\r
 366 \r
 367                         while (data.indexOf("</entry>") == -1) {\r
 368                             data = in.readLine();\r
 369                             cacheData.add(data);\r
 370                         }\r
 371 \r
 372                         cacheData.add(data);\r
 373 \r
 374                         ids.remove(name);\r
 375                     }\r
 376                 }\r
 377             }\r
 378         } catch (Exception ex) {\r
 379             ex.printStackTrace();\r
 380         }\r
 381 \r
 382         String[] localData = new String[cacheData.size()];\r
 383         cacheData.toArray(localData);\r
 384 \r
 385         if ((localData != null) && (localData.length > 0)) {\r
 386             ReadUniprotFile(localData, null, align);\r
 387         }\r
 388     }\r
 389 \r
 390     String parseValue(String line, String tag, RandomAccessFile out) {\r
 391         if (out != null) {\r
 392             try {\r
 393                 out.writeBytes(line + "\n");\r
 394             } catch (Exception ex) {\r
 395             }\r
 396         }\r
 397 \r
 398         int index = line.indexOf(tag) + tag.length() + 1;\r
 399 \r
 400         if (index == tag.length()) {\r
 401             return "";\r
 402         }\r
 403 \r
 404         return line.substring(index, line.indexOf("\"", index + 1));\r
 405     }\r
 406 \r
 407     String parseElement(String line, String tag, RandomAccessFile out) {\r
 408         if (out != null) {\r
 409             try {\r
 410                 out.writeBytes(line + "\n");\r
 411             } catch (Exception ex) {\r
 412             }\r
 413         }\r
 414 \r
 415         int index = line.indexOf(tag) + tag.length();\r
 416 \r
 417         return line.substring(index, line.indexOf("</"));\r
 418     }\r
 419 }\r