2 * Jalview - A Sequence Alignment Editor and Viewer (Version 2.4)
\r
3 * Copyright (C) 2008 AM Waterhouse, J Procter, G Barton, M Clamp, S Searle
\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
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
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
24 import jalview.datamodel.*;
\r
25 import jalview.schemes.*;
\r
28 * Parse and create Jalview Features files Detects GFF format features files and
\r
29 * parses. Does not implement standard print() - call specific printFeatures or
\r
30 * printGFF. Uses AlignmentI.findSequence(String id) to find the sequence object
\r
31 * for the features annotation - this normally works on an exact match.
\r
34 * @version $Revision$
\r
36 public class FeaturesFile extends AlignFile
\r
39 * work around for GFF interpretation bug where source string becomes description rather than a group
\r
41 private boolean doGffSource = true;
\r
44 * Creates a new FeaturesFile object.
\r
46 public FeaturesFile()
\r
51 * Creates a new FeaturesFile object.
\r
58 * @throws IOException
\r
61 public FeaturesFile(String inFile, String type) throws IOException
\r
63 super(inFile, type);
\r
66 public FeaturesFile(FileParse source) throws IOException
\r
72 * The Application can render HTML, but the applet will remove HTML tags and
\r
73 * replace links with %LINK% Both need to read links in HTML however
\r
75 * @throws IOException
\r
78 public boolean parse(AlignmentI align, Hashtable colours,
\r
81 return parse(align, colours, null, removeHTML);
\r
85 * The Application can render HTML, but the applet will remove HTML tags and
\r
86 * replace links with %LINK% Both need to read links in HTML however
\r
88 * @throws IOException
\r
91 public boolean parse(AlignmentI align, Hashtable colours,
\r
92 Hashtable featureLink, boolean removeHTML)
\r
97 SequenceI seq = null;
\r
98 String type, desc, token = null;
\r
100 int index, start, end;
\r
102 StringTokenizer st;
\r
103 SequenceFeature sf;
\r
104 String featureGroup = null, groupLink = null;
\r
105 Hashtable typeLink = new Hashtable();
\r
107 boolean GFFFile = true;
\r
109 while ((line = nextLine()) != null)
\r
111 if (line.startsWith("#"))
\r
116 st = new StringTokenizer(line, "\t");
\r
117 if (st.countTokens() > 1 && st.countTokens() < 4)
\r
120 type = st.nextToken();
\r
121 if (type.equalsIgnoreCase("startgroup"))
\r
123 featureGroup = st.nextToken();
\r
124 if (st.hasMoreElements())
\r
126 groupLink = st.nextToken();
\r
127 featureLink.put(featureGroup, groupLink);
\r
130 else if (type.equalsIgnoreCase("endgroup"))
\r
132 // We should check whether this is the current group,
\r
133 // but at present theres no way of showing more than 1 group
\r
135 featureGroup = null;
\r
140 // TODO: graduated feature colour needs to be parsed here
\r
141 UserColourScheme ucs = new UserColourScheme(st.nextToken());
\r
142 colours.put(type, ucs.findColour('A'));
\r
143 if (st.hasMoreElements())
\r
145 String link = st.nextToken();
\r
146 typeLink.put(type, link);
\r
147 if (featureLink == null)
\r
149 featureLink = new Hashtable();
\r
151 featureLink.put(type, link);
\r
158 while (st.hasMoreElements())
\r
163 // Still possible this is an old Jalview file,
\r
164 // which does not have type colours at the beginning
\r
165 seqId = token = st.nextToken();
\r
166 seq = align.findName(seqId, true);
\r
169 desc = st.nextToken();
\r
171 if (doGffSource && desc.indexOf(' ')==-1) {
\r
172 // could also be a source term rather than description line
\r
173 group = new String(desc);
\r
175 type = st.nextToken();
\r
178 String stt = st.nextToken();
\r
179 if (stt.length()==0 || stt.equals("-"))
\r
183 start = Integer.parseInt(stt);
\r
185 } catch (NumberFormatException ex)
\r
191 String stt = st.nextToken();
\r
192 if (stt.length()==0 || stt.equals("-"))
\r
196 end = Integer.parseInt(stt);
\r
198 } catch (NumberFormatException ex)
\r
202 // TODO: decide if non positional feature assertion for input data where end==0 is generally valid
\r
205 // treat as non-positional feature, regardless.
\r
210 score = new Float(st.nextToken()).floatValue();
\r
211 } catch (NumberFormatException ex)
\r
216 sf = new SequenceFeature(type, desc, start, end, score, group);
\r
220 sf.setValue("STRAND", st.nextToken());
\r
221 sf.setValue("FRAME", st.nextToken());
\r
222 } catch (Exception ex)
\r
226 if (st.hasMoreTokens())
\r
228 StringBuffer attributes = new StringBuffer();
\r
229 while (st.hasMoreTokens())
\r
231 attributes.append("\t" + st.nextElement());
\r
233 // TODO validate and split GFF2 attributes field ? parse out ([A-Za-z][A-Za-z0-9_]*) <value> ; and add as sf.setValue(attrib, val);
\r
234 sf.setValue("ATTRIBUTES", attributes.toString());
\r
237 seq.addSequenceFeature(sf);
\r
238 while ((seq = align.findName(seq, seqId, true)) != null)
\r
240 seq.addSequenceFeature(new SequenceFeature(sf));
\r
246 if (GFFFile && seq == null)
\r
252 desc = st.nextToken();
\r
254 if (!st.hasMoreTokens())
\r
257 .println("DEBUG: Run out of tokens when trying to identify the destination for the feature.. giving up.");
\r
258 // in all probability, this isn't a file we understand, so bail
\r
263 token = st.nextToken();
\r
265 if (!token.equals("ID_NOT_SPECIFIED"))
\r
267 seq = align.findName(seqId = token, true);
\r
275 index = Integer.parseInt(st.nextToken());
\r
276 seq = align.getSequenceAt(index);
\r
277 } catch (NumberFormatException ex)
\r
285 System.out.println("Sequence not found: " + line);
\r
289 start = Integer.parseInt(st.nextToken());
\r
290 end = Integer.parseInt(st.nextToken());
\r
292 type = st.nextToken();
\r
294 if (!colours.containsKey(type))
\r
296 // Probably the old style groups file
\r
297 UserColourScheme ucs = new UserColourScheme(type);
\r
298 colours.put(type, ucs.findColour('A'));
\r
301 sf = new SequenceFeature(type, desc, "", start, end, featureGroup);
\r
303 if (groupLink != null && removeHTML)
\r
305 sf.addLink(groupLink);
\r
306 sf.description += "%LINK%";
\r
308 if (typeLink.containsKey(type) && removeHTML)
\r
310 sf.addLink(typeLink.get(type).toString());
\r
311 sf.description += "%LINK%";
\r
314 parseDescriptionHTML(sf, removeHTML);
\r
316 seq.addSequenceFeature(sf);
\r
318 while (seqId != null
\r
319 && (seq = align.findName(seq, seqId, false)) != null)
\r
321 seq.addSequenceFeature(new SequenceFeature(sf));
\r
323 // If we got here, its not a GFFFile
\r
327 } catch (Exception ex)
\r
329 System.out.println(line);
\r
330 System.out.println("Error parsing feature file: " + ex + "\n" + line);
\r
331 ex.printStackTrace(System.err);
\r
338 public void parseDescriptionHTML(SequenceFeature sf, boolean removeHTML)
\r
340 if (sf.getDescription() == null)
\r
346 && sf.getDescription().toUpperCase().indexOf("<HTML>") == -1)
\r
348 removeHTML = false;
\r
351 StringBuffer sb = new StringBuffer();
\r
352 StringTokenizer st = new StringTokenizer(sf.getDescription(), "<");
\r
353 String token, link;
\r
356 while (st.hasMoreElements())
\r
358 token = st.nextToken("&>");
\r
359 if (token.equalsIgnoreCase("html") || token.startsWith("/"))
\r
365 startTag = token.indexOf("<");
\r
369 tag = token.substring(startTag + 1);
\r
370 token = token.substring(0, startTag);
\r
373 if (tag != null && tag.toUpperCase().startsWith("A HREF="))
\r
375 if (token.length() > 0)
\r
379 link = tag.substring(tag.indexOf("\"") + 1, tag.length() - 1);
\r
380 String label = st.nextToken("<>");
\r
381 sf.addLink(label + "|" + link);
\r
382 sb.append(label + "%LINK%");
\r
384 else if (tag != null && tag.equalsIgnoreCase("br"))
\r
388 else if (token.startsWith("lt;"))
\r
390 sb.append("<" + token.substring(3));
\r
392 else if (token.startsWith("gt;"))
\r
394 sb.append(">" + token.substring(3));
\r
396 else if (token.startsWith("amp;"))
\r
398 sb.append("&" + token.substring(4));
\r
408 sf.description = sb.toString();
\r
414 * generate a features file for seqs
\r
417 * source of sequence features
\r
419 * hash of feature types and colours
\r
420 * @return features file contents
\r
422 public String printJalviewFormat(SequenceI[] seqs, Hashtable visible)
\r
424 return printJalviewFormat(seqs, visible, true);
\r
428 * generate a features file for seqs with colours from visible (if any)
\r
431 * source of features
\r
433 * hash of Colours for each feature type
\r
435 * when true only feature types in 'visible' will be output
\r
436 * @return features file contents
\r
438 public String printJalviewFormat(SequenceI[] seqs, Hashtable visible,
\r
441 StringBuffer out = new StringBuffer();
\r
442 SequenceFeature[] next;
\r
444 if (visOnly && (visible == null || visible.size() < 1))
\r
446 return "No Features Visible";
\r
448 if (visible != null && visOnly)
\r
450 // write feature colours only if we're given them and we are generating
\r
452 Enumeration en = visible.keys();
\r
455 while (en.hasMoreElements())
\r
457 type = en.nextElement().toString();
\r
458 // TODO: Extend to output graduated feature colour details
\r
460 color = Integer.parseInt(visible.get(type).toString());
\r
463 + jalview.util.Format
\r
464 .getHexString(new java.awt.Color(color)) + "\n");
\r
467 // Work out which groups are both present and visible
\r
468 Vector groups = new Vector();
\r
469 int groupIndex = 0;
\r
471 for (int i = 0; i < seqs.length; i++)
\r
473 next = seqs[i].getSequenceFeatures();
\r
476 for (int j = 0; j < next.length; j++)
\r
478 if (visOnly && !visible.containsKey(next[j].type))
\r
483 if (next[j].featureGroup != null
\r
484 && !groups.contains(next[j].featureGroup))
\r
486 groups.addElement(next[j].featureGroup);
\r
492 String group = null;
\r
497 if (groups.size() > 0 && groupIndex < groups.size())
\r
499 group = groups.elementAt(groupIndex).toString();
\r
500 out.append("\nSTARTGROUP\t" + group + "\n");
\r
507 for (int i = 0; i < seqs.length; i++)
\r
509 next = seqs[i].getSequenceFeatures();
\r
512 for (int j = 0; j < next.length; j++)
\r
514 if (visOnly && !visible.containsKey(next[j].type))
\r
520 && (next[j].featureGroup == null || !next[j].featureGroup
\r
526 if (group == null && next[j].featureGroup != null)
\r
531 if (next[j].description == null
\r
532 || next[j].description.equals(""))
\r
534 out.append(next[j].type + "\t");
\r
538 if (next[j].links != null
\r
539 && next[j].getDescription().indexOf("<html>") == -1)
\r
541 out.append("<html>");
\r
544 out.append(next[j].description + " ");
\r
545 if (next[j].links != null)
\r
547 for (int l = 0; l < next[j].links.size(); l++)
\r
549 String label = next[j].links.elementAt(l).toString();
\r
550 String href = label.substring(label.indexOf("|") + 1);
\r
551 label = label.substring(0, label.indexOf("|"));
\r
553 if (next[j].description.indexOf(href) == -1)
\r
556 .append("<a href=\"" + href + "\">" + label
\r
561 if (next[j].getDescription().indexOf("</html>") == -1)
\r
563 out.append("</html>");
\r
570 out.append(seqs[i].getName() + "\t-1\t" + next[j].begin + "\t"
\r
571 + next[j].end + "\t" + next[j].type + "\n");
\r
578 out.append("ENDGROUP\t" + group + "\n");
\r
586 } while (groupIndex < groups.size() + 1);
\r
588 return out.toString();
\r
591 public String printGFFFormat(SequenceI[] seqs, Hashtable visible)
\r
593 return printGFFFormat(seqs, visible, true);
\r
596 public String printGFFFormat(SequenceI[] seqs, Hashtable visible,
\r
599 StringBuffer out = new StringBuffer();
\r
600 SequenceFeature[] next;
\r
603 for (int i = 0; i < seqs.length; i++)
\r
605 if (seqs[i].getSequenceFeatures() != null)
\r
607 next = seqs[i].getSequenceFeatures();
\r
608 for (int j = 0; j < next.length; j++)
\r
610 if (visOnly && visible!=null && !visible.containsKey(next[j].type))
\r
615 source = next[j].featureGroup;
\r
616 if (source == null)
\r
618 source = next[j].getDescription();
\r
621 out.append(seqs[i].getName() + "\t" + source + "\t"
\r
622 + next[j].type + "\t" + next[j].begin + "\t"
\r
623 + next[j].end + "\t" + next[j].score + "\t");
\r
625 if (next[j].getValue("STRAND") != null)
\r
627 out.append(next[j].getValue("STRAND") + "\t");
\r
634 if (next[j].getValue("FRAME") != null)
\r
636 out.append(next[j].getValue("FRAME"));
\r
643 if (next[j].getValue("ATTRIBUTES") != null)
\r
645 out.append(next[j].getValue("ATTRIBUTES"));
\r
654 return out.toString();
\r
658 * this is only for the benefit of object polymorphism - method does nothing.
\r
660 public void parse()
\r
666 * this is only for the benefit of object polymorphism - method does nothing.
\r
668 * @return error message
\r
670 public String print()
\r
672 return "USE printGFFFormat() or printJalviewFormat()";
\r