2 * Jalview - A Sequence Alignment Editor and Viewer (Version 2.6)
\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 * when true, assume GFF style features rather than Jalview style.
\r
110 boolean GFFFile = true;
\r
111 while ((line = nextLine()) != null)
\r
113 if (line.startsWith("#"))
\r
118 st = new StringTokenizer(line, "\t");
\r
119 if (st.countTokens() == 1)
\r
121 if (line.trim().equalsIgnoreCase("GFF"))
\r
123 // Start parsing file as if it might be GFF again.
\r
128 if (st.countTokens() > 1 && st.countTokens() < 4)
\r
131 type = st.nextToken();
\r
132 if (type.equalsIgnoreCase("startgroup"))
\r
134 featureGroup = st.nextToken();
\r
135 if (st.hasMoreElements())
\r
137 groupLink = st.nextToken();
\r
138 featureLink.put(featureGroup, groupLink);
\r
141 else if (type.equalsIgnoreCase("endgroup"))
\r
143 // We should check whether this is the current group,
\r
144 // but at present theres no way of showing more than 1 group
\r
146 featureGroup = null;
\r
151 Object colour = null;
\r
152 String colscheme = st.nextToken();
\r
153 if (colscheme.indexOf("|") > -1
\r
154 || colscheme.trim().equalsIgnoreCase("label"))
\r
156 // Parse '|' separated graduated colourscheme fields:
\r
157 // [label|][mincolour|maxcolour|[absolute|]minvalue|maxvalue|thresholdtype|thresholdvalue]
\r
158 // can either provide 'label' only, first is optional, next two
\r
159 // colors are required (but may be
\r
160 // left blank), next is optional, nxt two min/max are required.
\r
161 // first is either 'label'
\r
162 // first/second and third are both hexadecimal or word equivalent
\r
164 // next two are values parsed as floats.
\r
165 // fifth is either 'above','below', or 'none'.
\r
166 // sixth is a float value and only required when fifth is either
\r
167 // 'above' or 'below'.
\r
168 StringTokenizer gcol = new StringTokenizer(colscheme, "|",
\r
171 int threshtype = AnnotationColourGradient.NO_THRESHOLD;
\r
172 float min = Float.MIN_VALUE, max = Float.MAX_VALUE, threshval = Float.NaN;
\r
173 boolean labelCol = false;
\r
175 String mincol = gcol.nextToken();
\r
179 .println("Expected either 'label' or a colour specification in the line: "
\r
183 String maxcol = null;
\r
184 if (mincol.toLowerCase().indexOf("label") == 0)
\r
187 mincol = (gcol.hasMoreTokens() ? gcol.nextToken() : null); // skip
\r
189 mincol = (gcol.hasMoreTokens() ? gcol.nextToken() : null);
\r
191 String abso = null, minval, maxval;
\r
192 if (mincol != null)
\r
194 // at least four more tokens
\r
195 if (mincol.equals("|"))
\r
201 gcol.nextToken(); // skip next '|'
\r
203 // continue parsing rest of line
\r
204 maxcol = gcol.nextToken();
\r
205 if (maxcol.equals("|"))
\r
211 gcol.nextToken(); // skip next '|'
\r
213 abso = gcol.nextToken();
\r
214 gcol.nextToken(); // skip next '|'
\r
215 if (abso.toLowerCase().indexOf("abso") != 0)
\r
222 minval = gcol.nextToken();
\r
223 gcol.nextToken(); // skip next '|'
\r
225 maxval = gcol.nextToken();
\r
226 if (gcol.hasMoreTokens())
\r
228 gcol.nextToken(); // skip next '|'
\r
232 if (minval.length() > 0)
\r
234 min = new Float(minval).floatValue();
\r
236 } catch (Exception e)
\r
239 .println("Couldn't parse the minimum value for graduated colour for type ("
\r
241 + ") - did you misspell 'auto' for the optional automatic colour switch ?");
\r
242 e.printStackTrace();
\r
246 if (maxval.length() > 0)
\r
248 max = new Float(maxval).floatValue();
\r
250 } catch (Exception e)
\r
253 .println("Couldn't parse the maximum value for graduated colour for type ("
\r
254 + colscheme + ")");
\r
255 e.printStackTrace();
\r
260 // add in some dummy min/max colours for the label-only
\r
267 colour = new jalview.schemes.GraduatedColor(
\r
268 new UserColourScheme(mincol).findColour('A'),
\r
269 new UserColourScheme(maxcol).findColour('A'), min,
\r
271 } catch (Exception e)
\r
274 .println("Couldn't parse the graduated colour scheme ("
\r
275 + colscheme + ")");
\r
276 e.printStackTrace();
\r
278 if (colour != null)
\r
280 ((jalview.schemes.GraduatedColor) colour)
\r
281 .setColourByLabel(labelCol);
\r
282 ((jalview.schemes.GraduatedColor) colour)
\r
283 .setAutoScaled(abso == null);
\r
284 // add in any additional parameters
\r
285 String ttype = null, tval = null;
\r
286 if (gcol.hasMoreTokens())
\r
288 // threshold type and possibly a threshold value
\r
289 ttype = gcol.nextToken();
\r
290 if (ttype.toLowerCase().startsWith("below"))
\r
292 ((jalview.schemes.GraduatedColor) colour)
\r
293 .setThreshType(AnnotationColourGradient.BELOW_THRESHOLD);
\r
295 else if (ttype.toLowerCase().startsWith("above"))
\r
297 ((jalview.schemes.GraduatedColor) colour)
\r
298 .setThreshType(AnnotationColourGradient.ABOVE_THRESHOLD);
\r
302 ((jalview.schemes.GraduatedColor) colour)
\r
303 .setThreshType(AnnotationColourGradient.NO_THRESHOLD);
\r
304 if (!ttype.toLowerCase().startsWith("no"))
\r
307 .println("Ignoring unrecognised threshold type : "
\r
312 if (((GraduatedColor) colour).getThreshType() != AnnotationColourGradient.NO_THRESHOLD)
\r
317 tval = gcol.nextToken();
\r
318 ((jalview.schemes.GraduatedColor) colour)
\r
319 .setThresh(new Float(tval).floatValue());
\r
320 } catch (Exception e)
\r
323 .println("Couldn't parse threshold value as a float: ("
\r
325 e.printStackTrace();
\r
328 // parse the thresh-is-min token ?
\r
329 if (gcol.hasMoreTokens())
\r
332 .println("Ignoring additional tokens in parameters in graduated colour specification\n");
\r
333 while (gcol.hasMoreTokens())
\r
335 System.err.println("|" + gcol.nextToken());
\r
337 System.err.println("\n");
\r
343 UserColourScheme ucs = new UserColourScheme(colscheme);
\r
344 colour = ucs.findColour('A');
\r
346 if (colour != null)
\r
348 colours.put(type, colour);
\r
350 if (st.hasMoreElements())
\r
352 String link = st.nextToken();
\r
353 typeLink.put(type, link);
\r
354 if (featureLink == null)
\r
356 featureLink = new Hashtable();
\r
358 featureLink.put(type, link);
\r
364 while (st.hasMoreElements())
\r
369 // Still possible this is an old Jalview file,
\r
370 // which does not have type colours at the beginning
\r
371 seqId = token = st.nextToken();
\r
372 seq = align.findName(seqId, true);
\r
375 desc = st.nextToken();
\r
376 String group = null;
\r
377 if (doGffSource && desc.indexOf(' ') == -1)
\r
379 // could also be a source term rather than description line
\r
380 group = new String(desc);
\r
382 type = st.nextToken();
\r
385 String stt = st.nextToken();
\r
386 if (stt.length() == 0 || stt.equals("-"))
\r
392 start = Integer.parseInt(stt);
\r
394 } catch (NumberFormatException ex)
\r
400 String stt = st.nextToken();
\r
401 if (stt.length() == 0 || stt.equals("-"))
\r
407 end = Integer.parseInt(stt);
\r
409 } catch (NumberFormatException ex)
\r
413 // TODO: decide if non positional feature assertion for input data
\r
414 // where end==0 is generally valid
\r
417 // treat as non-positional feature, regardless.
\r
422 score = new Float(st.nextToken()).floatValue();
\r
423 } catch (NumberFormatException ex)
\r
428 sf = new SequenceFeature(type, desc, start, end, score, group);
\r
432 sf.setValue("STRAND", st.nextToken());
\r
433 sf.setValue("FRAME", st.nextToken());
\r
434 } catch (Exception ex)
\r
438 if (st.hasMoreTokens())
\r
440 StringBuffer attributes = new StringBuffer();
\r
441 while (st.hasMoreTokens())
\r
443 attributes.append("\t" + st.nextElement());
\r
445 // TODO validate and split GFF2 attributes field ? parse out
\r
446 // ([A-Za-z][A-Za-z0-9_]*) <value> ; and add as
\r
447 // sf.setValue(attrib, val);
\r
448 sf.setValue("ATTRIBUTES", attributes.toString());
\r
451 seq.addSequenceFeature(sf);
\r
452 while ((seq = align.findName(seq, seqId, true)) != null)
\r
454 seq.addSequenceFeature(new SequenceFeature(sf));
\r
460 if (GFFFile && seq == null)
\r
466 desc = st.nextToken();
\r
468 if (!st.hasMoreTokens())
\r
471 .println("DEBUG: Run out of tokens when trying to identify the destination for the feature.. giving up.");
\r
472 // in all probability, this isn't a file we understand, so bail
\r
477 token = st.nextToken();
\r
479 if (!token.equals("ID_NOT_SPECIFIED"))
\r
481 seq = align.findName(seqId = token, true);
\r
489 index = Integer.parseInt(st.nextToken());
\r
490 seq = align.getSequenceAt(index);
\r
491 } catch (NumberFormatException ex)
\r
499 System.out.println("Sequence not found: " + line);
\r
503 start = Integer.parseInt(st.nextToken());
\r
504 end = Integer.parseInt(st.nextToken());
\r
506 type = st.nextToken();
\r
508 if (!colours.containsKey(type))
\r
510 // Probably the old style groups file
\r
511 UserColourScheme ucs = new UserColourScheme(type);
\r
512 colours.put(type, ucs.findColour('A'));
\r
514 sf = new SequenceFeature(type, desc, "", start, end, featureGroup);
\r
515 if (st.hasMoreTokens())
\r
519 score = new Float(st.nextToken()).floatValue();
\r
520 // update colourgradient bounds if allowed to
\r
521 } catch (NumberFormatException ex)
\r
525 sf.setScore(score);
\r
527 if (groupLink != null && removeHTML)
\r
529 sf.addLink(groupLink);
\r
530 sf.description += "%LINK%";
\r
532 if (typeLink.containsKey(type) && removeHTML)
\r
534 sf.addLink(typeLink.get(type).toString());
\r
535 sf.description += "%LINK%";
\r
538 parseDescriptionHTML(sf, removeHTML);
\r
540 seq.addSequenceFeature(sf);
\r
542 while (seqId != null
\r
543 && (seq = align.findName(seq, seqId, false)) != null)
\r
545 seq.addSequenceFeature(new SequenceFeature(sf));
\r
547 // If we got here, its not a GFFFile
\r
551 } catch (Exception ex)
\r
553 System.out.println(line);
\r
554 System.out.println("Error parsing feature file: " + ex + "\n" + line);
\r
555 ex.printStackTrace(System.err);
\r
562 public void parseDescriptionHTML(SequenceFeature sf, boolean removeHTML)
\r
564 if (sf.getDescription() == null)
\r
570 && sf.getDescription().toUpperCase().indexOf("<HTML>") == -1)
\r
572 removeHTML = false;
\r
575 StringBuffer sb = new StringBuffer();
\r
576 StringTokenizer st = new StringTokenizer(sf.getDescription(), "<");
\r
577 String token, link;
\r
580 while (st.hasMoreElements())
\r
582 token = st.nextToken("&>");
\r
583 if (token.equalsIgnoreCase("html") || token.startsWith("/"))
\r
589 startTag = token.indexOf("<");
\r
593 tag = token.substring(startTag + 1);
\r
594 token = token.substring(0, startTag);
\r
597 if (tag != null && tag.toUpperCase().startsWith("A HREF="))
\r
599 if (token.length() > 0)
\r
603 link = tag.substring(tag.indexOf("\"") + 1, tag.length() - 1);
\r
604 String label = st.nextToken("<>");
\r
605 sf.addLink(label + "|" + link);
\r
606 sb.append(label + "%LINK%");
\r
608 else if (tag != null && tag.equalsIgnoreCase("br"))
\r
612 else if (token.startsWith("lt;"))
\r
614 sb.append("<" + token.substring(3));
\r
616 else if (token.startsWith("gt;"))
\r
618 sb.append(">" + token.substring(3));
\r
620 else if (token.startsWith("amp;"))
\r
622 sb.append("&" + token.substring(4));
\r
632 sf.description = sb.toString();
\r
638 * generate a features file for seqs includes non-pos features by default.
\r
641 * source of sequence features
\r
643 * hash of feature types and colours
\r
644 * @return features file contents
\r
646 public String printJalviewFormat(SequenceI[] seqs, Hashtable visible)
\r
648 return printJalviewFormat(seqs, visible, true, true);
\r
652 * generate a features file for seqs with colours from visible (if any)
\r
655 * source of features
\r
657 * hash of Colours for each feature type
\r
659 * when true only feature types in 'visible' will be output
\r
661 * indicates if non-positional features should be output (regardless
\r
662 * of group or type)
\r
663 * @return features file contents
\r
665 public String printJalviewFormat(SequenceI[] seqs, Hashtable visible,
\r
666 boolean visOnly, boolean nonpos)
\r
668 StringBuffer out = new StringBuffer();
\r
669 SequenceFeature[] next;
\r
670 boolean featuresGen = false;
\r
671 if (visOnly && !nonpos && (visible == null || visible.size() < 1))
\r
673 // no point continuing.
\r
674 return "No Features Visible";
\r
677 if (visible != null && visOnly)
\r
679 // write feature colours only if we're given them and we are generating
\r
681 // TODO: decide if feature links should also be written here ?
\r
682 Enumeration en = visible.keys();
\r
683 String type, color;
\r
684 while (en.hasMoreElements())
\r
686 type = en.nextElement().toString();
\r
688 if (visible.get(type) instanceof GraduatedColor)
\r
690 GraduatedColor gc = (GraduatedColor) visible.get(type);
\r
691 color = (gc.isColourByLabel() ? "label|" : "")
\r
692 + Format.getHexString(gc.getMinColor()) + "|"
\r
693 + Format.getHexString(gc.getMaxColor())
\r
694 + (gc.isAutoScale() ? "|" : "|abso|") + gc.getMin() + "|"
\r
695 + gc.getMax() + "|";
\r
696 if (gc.getThreshType() != AnnotationColourGradient.NO_THRESHOLD)
\r
698 if (gc.getThreshType() == AnnotationColourGradient.BELOW_THRESHOLD)
\r
704 if (gc.getThreshType() != AnnotationColourGradient.ABOVE_THRESHOLD)
\r
706 System.err.println("WARNING: Unsupported threshold type ("
\r
707 + gc.getThreshType() + ") : Assuming 'above'");
\r
712 color += "|" + gc.getThresh();
\r
719 else if (visible.get(type) instanceof java.awt.Color)
\r
721 color = Format.getHexString((java.awt.Color) visible.get(type));
\r
725 // legacy support for integer objects containing colour triplet values
\r
726 color = Format.getHexString(new java.awt.Color(Integer
\r
727 .parseInt(visible.get(type).toString())));
\r
729 out.append(type + "\t" + color + "\n");
\r
732 // Work out which groups are both present and visible
\r
733 Vector groups = new Vector();
\r
734 int groupIndex = 0;
\r
735 boolean isnonpos = false;
\r
737 for (int i = 0; i < seqs.length; i++)
\r
739 next = seqs[i].getSequenceFeatures();
\r
742 for (int j = 0; j < next.length; j++)
\r
744 isnonpos = next[j].begin == 0 && next[j].end == 0;
\r
745 if ((!nonpos && isnonpos)
\r
746 || (!isnonpos && visOnly && !visible
\r
747 .containsKey(next[j].type)))
\r
752 if (next[j].featureGroup != null
\r
753 && !groups.contains(next[j].featureGroup))
\r
755 groups.addElement(next[j].featureGroup);
\r
761 String group = null;
\r
765 if (groups.size() > 0 && groupIndex < groups.size())
\r
767 group = groups.elementAt(groupIndex).toString();
\r
768 out.append("\nSTARTGROUP\t" + group + "\n");
\r
775 for (int i = 0; i < seqs.length; i++)
\r
777 next = seqs[i].getSequenceFeatures();
\r
780 for (int j = 0; j < next.length; j++)
\r
782 isnonpos = next[j].begin == 0 && next[j].end == 0;
\r
783 if ((!nonpos && isnonpos)
\r
784 || (!isnonpos && visOnly && !visible
\r
785 .containsKey(next[j].type)))
\r
787 // skip if feature is nonpos and we ignore them or if we only
\r
788 // output visible and it isn't non-pos and it's not visible
\r
793 && (next[j].featureGroup == null || !next[j].featureGroup
\r
799 if (group == null && next[j].featureGroup != null)
\r
803 // we have features to output
\r
804 featuresGen = true;
\r
805 if (next[j].description == null
\r
806 || next[j].description.equals(""))
\r
808 out.append(next[j].type + "\t");
\r
812 if (next[j].links != null
\r
813 && next[j].getDescription().indexOf("<html>") == -1)
\r
815 out.append("<html>");
\r
818 out.append(next[j].description + " ");
\r
819 if (next[j].links != null)
\r
821 for (int l = 0; l < next[j].links.size(); l++)
\r
823 String label = next[j].links.elementAt(l).toString();
\r
824 String href = label.substring(label.indexOf("|") + 1);
\r
825 label = label.substring(0, label.indexOf("|"));
\r
827 if (next[j].description.indexOf(href) == -1)
\r
829 out.append("<a href=\"" + href + "\">" + label + "</a>");
\r
833 if (next[j].getDescription().indexOf("</html>") == -1)
\r
835 out.append("</html>");
\r
841 out.append(seqs[i].getName()
\r
848 + ((next[j].score != Float.NaN) ? "\t" + next[j].score
\r
856 out.append("ENDGROUP\t" + group + "\n");
\r
864 } while (groupIndex < groups.size() + 1);
\r
868 return "No Features Visible";
\r
871 return out.toString();
\r
875 * generate a gff file for sequence features includes non-pos features by
\r
882 public String printGFFFormat(SequenceI[] seqs, Hashtable visible)
\r
884 return printGFFFormat(seqs, visible, true, true);
\r
887 public String printGFFFormat(SequenceI[] seqs, Hashtable visible,
\r
888 boolean visOnly, boolean nonpos)
\r
890 StringBuffer out = new StringBuffer();
\r
891 SequenceFeature[] next;
\r
894 for (int i = 0; i < seqs.length; i++)
\r
896 if (seqs[i].getSequenceFeatures() != null)
\r
898 next = seqs[i].getSequenceFeatures();
\r
899 for (int j = 0; j < next.length; j++)
\r
901 isnonpos = next[j].begin == 0 && next[j].end == 0;
\r
902 if ((!nonpos && isnonpos)
\r
903 || (!isnonpos && visOnly && !visible
\r
904 .containsKey(next[j].type)))
\r
909 source = next[j].featureGroup;
\r
910 if (source == null)
\r
912 source = next[j].getDescription();
\r
915 out.append(seqs[i].getName() + "\t" + source + "\t"
\r
916 + next[j].type + "\t" + next[j].begin + "\t"
\r
917 + next[j].end + "\t" + next[j].score + "\t");
\r
919 if (next[j].getValue("STRAND") != null)
\r
921 out.append(next[j].getValue("STRAND") + "\t");
\r
928 if (next[j].getValue("FRAME") != null)
\r
930 out.append(next[j].getValue("FRAME"));
\r
937 if (next[j].getValue("ATTRIBUTES") != null)
\r
939 out.append(next[j].getValue("ATTRIBUTES"));
\r
948 return out.toString();
\r
952 * this is only for the benefit of object polymorphism - method does nothing.
\r
954 public void parse()
\r
960 * this is only for the benefit of object polymorphism - method does nothing.
\r
962 * @return error message
\r
964 public String print()
\r
966 return "USE printGFFFormat() or printJalviewFormat()";
\r