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