2 * Jalview - A Sequence Alignment Editor and Viewer (Version 2.5)
\r
3 * Copyright (C) 2010 J Procter, AM Waterhouse, G Barton, M Clamp, S Searle
\r
5 * This file is part of Jalview.
\r
7 * Jalview is free software: you can redistribute it and/or
\r
8 * modify it under the terms of the GNU General Public License
\r
9 * as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.
\r
11 * Jalview is distributed in the hope that it will be useful, but
\r
12 * WITHOUT ANY WARRANTY; without even the implied warranty
\r
13 * of MERCHANTABILITY or FITNESS FOR A PARTICULAR
\r
14 * PURPOSE. See the GNU General Public License for more details.
\r
16 * You should have received a copy of the GNU General Public License along with Jalview. If not, see <http://www.gnu.org/licenses/>.
\r
23 import jalview.datamodel.*;
\r
24 import jalview.schemes.*;
\r
25 import jalview.util.Format;
\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
\r
40 * description rather than a group
\r
42 private boolean doGffSource = true;
\r
45 * Creates a new FeaturesFile object.
\r
47 public FeaturesFile()
\r
52 * Creates a new FeaturesFile object.
\r
59 * @throws IOException
\r
62 public FeaturesFile(String inFile, String type) throws IOException
\r
64 super(inFile, type);
\r
67 public FeaturesFile(FileParse source) throws IOException
\r
73 * The Application can render HTML, but the applet will remove HTML tags and
\r
74 * replace links with %LINK% Both need to read links in HTML however
\r
76 * @throws IOException
\r
79 public boolean parse(AlignmentI align, Hashtable colours,
\r
82 return parse(align, colours, null, removeHTML);
\r
86 * The Application can render HTML, but the applet will remove HTML tags and
\r
87 * replace links with %LINK% Both need to read links in HTML however
\r
89 * @throws IOException
\r
92 public boolean parse(AlignmentI align, Hashtable colours,
\r
93 Hashtable featureLink, boolean removeHTML)
\r
98 SequenceI seq = null;
\r
99 String type, desc, token = null;
\r
101 int index, start, end;
\r
103 StringTokenizer st;
\r
104 SequenceFeature sf;
\r
105 String featureGroup = null, groupLink = null;
\r
106 Hashtable typeLink = new Hashtable();
\r
108 boolean GFFFile = true;
\r
110 while ((line = nextLine()) != null)
\r
112 if (line.startsWith("#"))
\r
117 st = new StringTokenizer(line, "\t");
\r
118 if (st.countTokens() > 1 && st.countTokens() < 4)
\r
121 type = st.nextToken();
\r
122 if (type.equalsIgnoreCase("startgroup"))
\r
124 featureGroup = st.nextToken();
\r
125 if (st.hasMoreElements())
\r
127 groupLink = st.nextToken();
\r
128 featureLink.put(featureGroup, groupLink);
\r
131 else if (type.equalsIgnoreCase("endgroup"))
\r
133 // We should check whether this is the current group,
\r
134 // but at present theres no way of showing more than 1 group
\r
136 featureGroup = null;
\r
141 Object colour = null;
\r
142 String colscheme = st.nextToken();
\r
143 if (colscheme.indexOf("|") > -1)
\r
145 // Parse '|' separated graduated colourscheme fields:
\r
146 // mincolour|maxcolour|[absolute|]minvalue|maxvalue|thresholdtype|thresholdvalue
\r
147 // first four are required.
\r
148 // first two are hexadecimal or word equivalent colours.
\r
149 // second two are values parsed as floats.
\r
150 // fifth is either 'above','below', or 'none'.
\r
151 // sixth is a float value and only required when fifth is either
\r
152 // 'above' or 'below'.
\r
153 StringTokenizer gcol = new StringTokenizer(colscheme, "|");
\r
154 String mincol = gcol.nextToken(), maxcol = gcol.nextToken();
\r
155 String abso = gcol.nextToken(), minval, maxval;
\r
156 if (abso.toLowerCase().indexOf("abso") != 0)
\r
163 minval = gcol.nextToken();
\r
165 maxval = gcol.nextToken();
\r
167 int threshtype = AnnotationColourGradient.NO_THRESHOLD;
\r
168 float min = Float.MIN_VALUE, max = Float.MAX_VALUE, threshval = Float.NaN;
\r
171 if (minval.length() > 0)
\r
173 min = new Float(minval).floatValue();
\r
175 } catch (Exception e)
\r
178 .println("Couldn't parse the minimum value for graduated colour for type ("
\r
180 + ") - did you misspell 'auto' for the optional automatic colour switch ?");
\r
181 e.printStackTrace();
\r
185 if (maxval.length() > 0)
\r
187 max = new Float(maxval).floatValue();
\r
189 } catch (Exception e)
\r
192 .println("Couldn't parse the maximum value for graduated colour for type ("
\r
193 + colscheme + ")");
\r
194 e.printStackTrace();
\r
198 colour = new jalview.schemes.GraduatedColor(
\r
199 new UserColourScheme(mincol).findColour('A'),
\r
200 new UserColourScheme(maxcol).findColour('A'), min,
\r
202 } catch (Exception e)
\r
205 .println("Couldn't parse the graduated colour scheme ("
\r
206 + colscheme + ")");
\r
207 e.printStackTrace();
\r
209 if (colour != null)
\r
211 ((jalview.schemes.GraduatedColor) colour)
\r
212 .setAutoScaled(abso == null);
\r
213 // add in any additional parameters
\r
214 String ttype = null, tval = null;
\r
215 if (gcol.hasMoreTokens())
\r
217 // threshold type and possibly a threshold value
\r
218 ttype = gcol.nextToken();
\r
219 if (ttype.toLowerCase().startsWith("below"))
\r
221 ((jalview.schemes.GraduatedColor) colour)
\r
222 .setThreshType(AnnotationColourGradient.BELOW_THRESHOLD);
\r
224 else if (ttype.toLowerCase().startsWith("above"))
\r
226 ((jalview.schemes.GraduatedColor) colour)
\r
227 .setThreshType(AnnotationColourGradient.ABOVE_THRESHOLD);
\r
231 ((jalview.schemes.GraduatedColor) colour)
\r
232 .setThreshType(AnnotationColourGradient.NO_THRESHOLD);
\r
233 if (!ttype.toLowerCase().startsWith("no"))
\r
236 .println("Ignoring unrecognised threshold type : "
\r
241 if (((GraduatedColor) colour).getThreshType() != AnnotationColourGradient.NO_THRESHOLD)
\r
243 tval = gcol.nextToken();
\r
246 ((jalview.schemes.GraduatedColor) colour)
\r
247 .setThresh(new Float(tval).floatValue());
\r
248 } catch (Exception e)
\r
251 .println("Couldn't parse threshold value as a float: ("
\r
253 e.printStackTrace();
\r
256 // parse the thresh-is-min token ?
\r
257 if (gcol.hasMoreTokens())
\r
260 .println("Ignoring additional tokens in parameters in graduated colour specification\n");
\r
261 while (gcol.hasMoreTokens())
\r
263 System.err.println("|" + gcol);
\r
265 System.err.println("\n");
\r
271 UserColourScheme ucs = new UserColourScheme(colscheme);
\r
272 colour = ucs.findColour('A');
\r
274 if (colour != null)
\r
276 colours.put(type, colour);
\r
278 if (st.hasMoreElements())
\r
280 String link = st.nextToken();
\r
281 typeLink.put(type, link);
\r
282 if (featureLink == null)
\r
284 featureLink = new Hashtable();
\r
286 featureLink.put(type, link);
\r
292 while (st.hasMoreElements())
\r
297 // Still possible this is an old Jalview file,
\r
298 // which does not have type colours at the beginning
\r
299 seqId = token = st.nextToken();
\r
300 seq = align.findName(seqId, true);
\r
303 desc = st.nextToken();
\r
304 String group = null;
\r
305 if (doGffSource && desc.indexOf(' ') == -1)
\r
307 // could also be a source term rather than description line
\r
308 group = new String(desc);
\r
310 type = st.nextToken();
\r
313 String stt = st.nextToken();
\r
314 if (stt.length() == 0 || stt.equals("-"))
\r
320 start = Integer.parseInt(stt);
\r
322 } catch (NumberFormatException ex)
\r
328 String stt = st.nextToken();
\r
329 if (stt.length() == 0 || stt.equals("-"))
\r
335 end = Integer.parseInt(stt);
\r
337 } catch (NumberFormatException ex)
\r
341 // TODO: decide if non positional feature assertion for input data
\r
342 // where end==0 is generally valid
\r
345 // treat as non-positional feature, regardless.
\r
350 score = new Float(st.nextToken()).floatValue();
\r
351 } catch (NumberFormatException ex)
\r
356 sf = new SequenceFeature(type, desc, start, end, score, group);
\r
360 sf.setValue("STRAND", st.nextToken());
\r
361 sf.setValue("FRAME", st.nextToken());
\r
362 } catch (Exception ex)
\r
366 if (st.hasMoreTokens())
\r
368 StringBuffer attributes = new StringBuffer();
\r
369 while (st.hasMoreTokens())
\r
371 attributes.append("\t" + st.nextElement());
\r
373 // TODO validate and split GFF2 attributes field ? parse out
\r
374 // ([A-Za-z][A-Za-z0-9_]*) <value> ; and add as
\r
375 // sf.setValue(attrib, val);
\r
376 sf.setValue("ATTRIBUTES", attributes.toString());
\r
379 seq.addSequenceFeature(sf);
\r
380 while ((seq = align.findName(seq, seqId, true)) != null)
\r
382 seq.addSequenceFeature(new SequenceFeature(sf));
\r
388 if (GFFFile && seq == null)
\r
394 desc = st.nextToken();
\r
396 if (!st.hasMoreTokens())
\r
399 .println("DEBUG: Run out of tokens when trying to identify the destination for the feature.. giving up.");
\r
400 // in all probability, this isn't a file we understand, so bail
\r
405 token = st.nextToken();
\r
407 if (!token.equals("ID_NOT_SPECIFIED"))
\r
409 seq = align.findName(seqId = token, true);
\r
417 index = Integer.parseInt(st.nextToken());
\r
418 seq = align.getSequenceAt(index);
\r
419 } catch (NumberFormatException ex)
\r
427 System.out.println("Sequence not found: " + line);
\r
431 start = Integer.parseInt(st.nextToken());
\r
432 end = Integer.parseInt(st.nextToken());
\r
434 type = st.nextToken();
\r
436 if (!colours.containsKey(type))
\r
438 // Probably the old style groups file
\r
439 UserColourScheme ucs = new UserColourScheme(type);
\r
440 colours.put(type, ucs.findColour('A'));
\r
442 sf = new SequenceFeature(type, desc, "", start, end, featureGroup);
\r
443 if (st.hasMoreTokens())
\r
447 score = new Float(st.nextToken()).floatValue();
\r
448 // update colourgradient bounds if allowed to
\r
449 } catch (NumberFormatException ex)
\r
453 sf.setScore(score);
\r
455 if (groupLink != null && removeHTML)
\r
457 sf.addLink(groupLink);
\r
458 sf.description += "%LINK%";
\r
460 if (typeLink.containsKey(type) && removeHTML)
\r
462 sf.addLink(typeLink.get(type).toString());
\r
463 sf.description += "%LINK%";
\r
466 parseDescriptionHTML(sf, removeHTML);
\r
468 seq.addSequenceFeature(sf);
\r
470 while (seqId != null
\r
471 && (seq = align.findName(seq, seqId, false)) != null)
\r
473 seq.addSequenceFeature(new SequenceFeature(sf));
\r
475 // If we got here, its not a GFFFile
\r
479 } catch (Exception ex)
\r
481 System.out.println(line);
\r
482 System.out.println("Error parsing feature file: " + ex + "\n" + line);
\r
483 ex.printStackTrace(System.err);
\r
490 public void parseDescriptionHTML(SequenceFeature sf, boolean removeHTML)
\r
492 if (sf.getDescription() == null)
\r
498 && sf.getDescription().toUpperCase().indexOf("<HTML>") == -1)
\r
500 removeHTML = false;
\r
503 StringBuffer sb = new StringBuffer();
\r
504 StringTokenizer st = new StringTokenizer(sf.getDescription(), "<");
\r
505 String token, link;
\r
508 while (st.hasMoreElements())
\r
510 token = st.nextToken("&>");
\r
511 if (token.equalsIgnoreCase("html") || token.startsWith("/"))
\r
517 startTag = token.indexOf("<");
\r
521 tag = token.substring(startTag + 1);
\r
522 token = token.substring(0, startTag);
\r
525 if (tag != null && tag.toUpperCase().startsWith("A HREF="))
\r
527 if (token.length() > 0)
\r
531 link = tag.substring(tag.indexOf("\"") + 1, tag.length() - 1);
\r
532 String label = st.nextToken("<>");
\r
533 sf.addLink(label + "|" + link);
\r
534 sb.append(label + "%LINK%");
\r
536 else if (tag != null && tag.equalsIgnoreCase("br"))
\r
540 else if (token.startsWith("lt;"))
\r
542 sb.append("<" + token.substring(3));
\r
544 else if (token.startsWith("gt;"))
\r
546 sb.append(">" + token.substring(3));
\r
548 else if (token.startsWith("amp;"))
\r
550 sb.append("&" + token.substring(4));
\r
560 sf.description = sb.toString();
\r
566 * generate a features file for seqs includes non-pos features by default.
\r
569 * source of sequence features
\r
571 * hash of feature types and colours
\r
572 * @return features file contents
\r
574 public String printJalviewFormat(SequenceI[] seqs, Hashtable visible)
\r
576 return printJalviewFormat(seqs, visible, true, true);
\r
580 * generate a features file for seqs with colours from visible (if any)
\r
583 * source of features
\r
585 * hash of Colours for each feature type
\r
587 * when true only feature types in 'visible' will be output
\r
589 * indicates if non-positional features should be output (regardless
\r
590 * of group or type)
\r
591 * @return features file contents
\r
593 public String printJalviewFormat(SequenceI[] seqs, Hashtable visible,
\r
594 boolean visOnly, boolean nonpos)
\r
596 StringBuffer out = new StringBuffer();
\r
597 SequenceFeature[] next;
\r
598 boolean featuresGen = false;
\r
599 if (visOnly && !nonpos && (visible == null || visible.size() < 1))
\r
601 // no point continuing.
\r
602 return "No Features Visible";
\r
605 if (visible != null && visOnly)
\r
607 // write feature colours only if we're given them and we are generating
\r
609 // TODO: decide if feature links should also be written here ?
\r
610 Enumeration en = visible.keys();
\r
611 String type, color;
\r
612 while (en.hasMoreElements())
\r
614 type = en.nextElement().toString();
\r
616 if (visible.get(type) instanceof GraduatedColor)
\r
618 GraduatedColor gc = (GraduatedColor) visible.get(type);
\r
619 // TODO: NOW: colour by label, autoscale flags.
\r
620 color = Format.getHexString(gc.getMinColor()) + "|"
\r
621 + Format.getHexString(gc.getMaxColor())
\r
622 + (gc.isAutoScale() ? "|" : "|abso|") + gc.getMin() + "|"
\r
623 + gc.getMax() + "|";
\r
624 if (gc.getThreshType() != AnnotationColourGradient.NO_THRESHOLD)
\r
626 if (gc.getThreshType() == AnnotationColourGradient.BELOW_THRESHOLD)
\r
632 if (gc.getThreshType() != AnnotationColourGradient.ABOVE_THRESHOLD)
\r
634 System.err.println("WARNING: Unsupported threshold type ("
\r
635 + gc.getThreshType() + ") : Assuming 'above'");
\r
640 color += "|" + gc.getThresh();
\r
647 else if (visible.get(type) instanceof java.awt.Color)
\r
649 color = Format.getHexString((java.awt.Color) visible.get(type));
\r
653 // legacy support for integer objects containing colour triplet values
\r
654 color = Format.getHexString(new java.awt.Color(Integer
\r
655 .parseInt(visible.get(type).toString())));
\r
657 out.append(type + "\t" + color + "\n");
\r
660 // Work out which groups are both present and visible
\r
661 Vector groups = new Vector();
\r
662 int groupIndex = 0;
\r
663 boolean isnonpos = false;
\r
665 for (int i = 0; i < seqs.length; i++)
\r
667 next = seqs[i].getSequenceFeatures();
\r
670 for (int j = 0; j < next.length; j++)
\r
672 isnonpos = next[j].begin == 0 && next[j].end == 0;
\r
673 if ((!nonpos && isnonpos)
\r
674 || (!isnonpos && visOnly && !visible
\r
675 .containsKey(next[j].type)))
\r
680 if (next[j].featureGroup != null
\r
681 && !groups.contains(next[j].featureGroup))
\r
683 groups.addElement(next[j].featureGroup);
\r
689 String group = null;
\r
693 if (groups.size() > 0 && groupIndex < groups.size())
\r
695 group = groups.elementAt(groupIndex).toString();
\r
696 out.append("\nSTARTGROUP\t" + group + "\n");
\r
703 for (int i = 0; i < seqs.length; i++)
\r
705 next = seqs[i].getSequenceFeatures();
\r
708 for (int j = 0; j < next.length; j++)
\r
710 isnonpos = next[j].begin == 0 && next[j].end == 0;
\r
711 if ((!nonpos && isnonpos)
\r
712 || (!isnonpos && visOnly && !visible
\r
713 .containsKey(next[j].type)))
\r
715 // skip if feature is nonpos and we ignore them or if we only
\r
716 // output visible and it isn't non-pos and it's not visible
\r
721 && (next[j].featureGroup == null || !next[j].featureGroup
\r
727 if (group == null && next[j].featureGroup != null)
\r
731 // we have features to output
\r
732 featuresGen = true;
\r
733 if (next[j].description == null
\r
734 || next[j].description.equals(""))
\r
736 out.append(next[j].type + "\t");
\r
740 if (next[j].links != null
\r
741 && next[j].getDescription().indexOf("<html>") == -1)
\r
743 out.append("<html>");
\r
746 out.append(next[j].description + " ");
\r
747 if (next[j].links != null)
\r
749 for (int l = 0; l < next[j].links.size(); l++)
\r
751 String label = next[j].links.elementAt(l).toString();
\r
752 String href = label.substring(label.indexOf("|") + 1);
\r
753 label = label.substring(0, label.indexOf("|"));
\r
755 if (next[j].description.indexOf(href) == -1)
\r
758 .append("<a href=\"" + href + "\">" + label
\r
763 if (next[j].getDescription().indexOf("</html>") == -1)
\r
765 out.append("</html>");
\r
771 out.append(seqs[i].getName()
\r
778 + ((next[j].score != Float.NaN) ? "\t" + next[j].score
\r
786 out.append("ENDGROUP\t" + group + "\n");
\r
794 } while (groupIndex < groups.size() + 1);
\r
798 return "No Features Visible";
\r
801 return out.toString();
\r
805 * generate a gff file for sequence features includes non-pos features by
\r
812 public String printGFFFormat(SequenceI[] seqs, Hashtable visible)
\r
814 return printGFFFormat(seqs, visible, true, true);
\r
817 public String printGFFFormat(SequenceI[] seqs, Hashtable visible,
\r
818 boolean visOnly, boolean nonpos)
\r
820 StringBuffer out = new StringBuffer();
\r
821 SequenceFeature[] next;
\r
824 for (int i = 0; i < seqs.length; i++)
\r
826 if (seqs[i].getSequenceFeatures() != null)
\r
828 next = seqs[i].getSequenceFeatures();
\r
829 for (int j = 0; j < next.length; j++)
\r
831 isnonpos = next[j].begin == 0 && next[j].end == 0;
\r
832 if ((!nonpos && isnonpos)
\r
833 || (!isnonpos && visOnly && !visible
\r
834 .containsKey(next[j].type)))
\r
839 source = next[j].featureGroup;
\r
840 if (source == null)
\r
842 source = next[j].getDescription();
\r
845 out.append(seqs[i].getName() + "\t" + source + "\t"
\r
846 + next[j].type + "\t" + next[j].begin + "\t"
\r
847 + next[j].end + "\t" + next[j].score + "\t");
\r
849 if (next[j].getValue("STRAND") != null)
\r
851 out.append(next[j].getValue("STRAND") + "\t");
\r
858 if (next[j].getValue("FRAME") != null)
\r
860 out.append(next[j].getValue("FRAME"));
\r
867 if (next[j].getValue("ATTRIBUTES") != null)
\r
869 out.append(next[j].getValue("ATTRIBUTES"));
\r
878 return out.toString();
\r
882 * this is only for the benefit of object polymorphism - method does nothing.
\r
884 public void parse()
\r
890 * this is only for the benefit of object polymorphism - method does nothing.
\r
892 * @return error message
\r
894 public String print()
\r
896 return "USE printGFFFormat() or printJalviewFormat()";
\r