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