2 * Jalview - A Sequence Alignment Editor and Viewer (Version 2.7)
3 * Copyright (C) 2011 J Procter, AM Waterhouse, J Engelhardt, LM Lui, G Barton, M Clamp, S Searle
5 * This file is part of Jalview.
7 * Jalview is free software: you can redistribute it and/or
8 * modify it under the terms of the GNU General Public License
9 * as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.
11 * Jalview is distributed in the hope that it will be useful, but
12 * WITHOUT ANY WARRANTY; without even the implied warranty
13 * of MERCHANTABILITY or FITNESS FOR A PARTICULAR
14 * PURPOSE. See the GNU General Public License for more details.
16 * You should have received a copy of the GNU General Public License along with Jalview. If not, see <http://www.gnu.org/licenses/>.
18 package jalview.analysis;
22 import jalview.datamodel.*;
23 import jalview.util.*;
26 * Routines for manipulating the order of a multiple sequence alignment TODO:
27 * this class retains some global states concerning sort-order which should be
28 * made attributes for the caller's alignment visualization. TODO: refactor to
29 * allow a subset of selected sequences to be sorted within the context of a
30 * whole alignment. Sort method template is: SequenceI[] tobesorted, [ input
31 * data mapping to each tobesorted element to use ], Alignment context of
32 * tobesorted that are to be re-ordered, boolean sortinplace, [special data - ie
33 * seuqence to be sorted w.r.t.]) sortinplace implies that the sorted vector
34 * resulting from applying the operation to tobesorted should be mapped back to
35 * the original positions in alignment. Otherwise, normal behaviour is to re
36 * order alignment so that tobesorted is sorted and grouped together starting
37 * from the first tobesorted position in the alignment. e.g. (a,tb2,b,tb1,c,tb3
38 * becomes a,tb1,tb2,tb3,b,c)
40 public class AlignmentSorter
43 * todo: refactor searches to follow a basic pattern: (search property, last
44 * search state, current sort direction)
46 static boolean sortIdAscending = true;
48 static int lastGroupHash = 0;
50 static boolean sortGroupAscending = true;
52 static AlignmentOrder lastOrder = null;
54 static boolean sortOrderAscending = true;
56 static NJTree lastTree = null;
58 static boolean sortTreeAscending = true;
61 * last Annotation Label used by sortByScore
63 private static String lastSortByScore;
65 private static boolean sortByScoreAscending = true;
68 * compact representation of last arguments to SortByFeatureScore
70 private static String lastSortByFeatureScore;
72 private static boolean sortByFeatureScoreAscending = true;
74 private static boolean sortLengthAscending;
77 * Sort by Percentage Identity w.r.t. s
84 * sequences from align that are to be sorted.
86 public static void sortByPID(AlignmentI align, SequenceI s,
89 sortByPID(align, s, tosort, 0, -1);
93 * Sort by Percentage Identity w.r.t. s
100 * sequences from align that are to be sorted.
102 * start column (0 for beginning
105 public static void sortByPID(AlignmentI align, SequenceI s,
106 SequenceI[] tosort, int start, int end)
108 int nSeq = align.getHeight();
110 float[] scores = new float[nSeq];
111 SequenceI[] seqs = new SequenceI[nSeq];
113 for (int i = 0; i < nSeq; i++)
115 scores[i] = Comparison.PID(align.getSequenceAt(i)
116 .getSequenceAsString(), s.getSequenceAsString());
117 seqs[i] = align.getSequenceAt(i);
120 QuickSort.sort(scores, 0, scores.length - 1, seqs);
122 setReverseOrder(align, seqs);
126 * Reverse the order of the sort
133 private static void setReverseOrder(AlignmentI align, SequenceI[] seqs)
135 int nSeq = seqs.length;
145 len = (nSeq + 1) / 2;
148 // NOTE: DO NOT USE align.setSequenceAt() here - it will NOT work
149 for (int i = 0; i < len; i++)
151 // SequenceI tmp = seqs[i];
152 align.getSequences().setElementAt(seqs[nSeq - i - 1], i);
153 align.getSequences().setElementAt(seqs[i], nSeq - i - 1);
158 * Sets the Alignment object with the given sequences
161 * Alignment object to be updated
163 * sequences as a vector
165 private static void setOrder(AlignmentI align, Vector tmp)
167 setOrder(align, vectorSubsetToArray(tmp, align.getSequences()));
171 * Sets the Alignment object with the given sequences
176 * sequences as an array
178 public static void setOrder(AlignmentI align, SequenceI[] seqs)
180 // NOTE: DO NOT USE align.setSequenceAt() here - it will NOT work
181 Vector algn = align.getSequences();
182 Vector tmp = new Vector();
184 for (int i = 0; i < seqs.length; i++)
186 if (algn.contains(seqs[i]))
188 tmp.addElement(seqs[i]);
192 algn.removeAllElements();
193 // User may have hidden seqs, then clicked undo or redo
194 for (int i = 0; i < tmp.size(); i++)
196 algn.addElement(tmp.elementAt(i));
202 * Sorts by ID. Numbers are sorted before letters.
205 * The alignment object to sort
207 public static void sortByID(AlignmentI align)
209 int nSeq = align.getHeight();
211 String[] ids = new String[nSeq];
212 SequenceI[] seqs = new SequenceI[nSeq];
214 for (int i = 0; i < nSeq; i++)
216 ids[i] = align.getSequenceAt(i).getName();
217 seqs[i] = align.getSequenceAt(i);
220 QuickSort.sort(ids, seqs);
224 setReverseOrder(align, seqs);
228 setOrder(align, seqs);
231 sortIdAscending = !sortIdAscending;
235 * Sorts by sequence length
238 * The alignment object to sort
240 public static void sortByLength(AlignmentI align)
242 int nSeq = align.getHeight();
244 float[] length = new float[nSeq];
245 SequenceI[] seqs = new SequenceI[nSeq];
247 for (int i = 0; i < nSeq; i++)
249 seqs[i] = align.getSequenceAt(i);
250 length[i] = (float) (seqs[i].getEnd() - seqs[i].getStart());
253 QuickSort.sort(length, seqs);
255 if (sortLengthAscending)
257 setReverseOrder(align, seqs);
261 setOrder(align, seqs);
264 sortLengthAscending = !sortLengthAscending;
268 * Sorts the alignment by size of group. <br>
269 * Maintains the order of sequences in each group by order in given alignment
273 * sorts the given alignment object by group
275 public static void sortByGroup(AlignmentI align)
277 // MAINTAINS ORIGNAL SEQUENCE ORDER,
278 // ORDERS BY GROUP SIZE
279 Vector groups = new Vector();
281 if (groups.hashCode() != lastGroupHash)
283 sortGroupAscending = true;
284 lastGroupHash = groups.hashCode();
288 sortGroupAscending = !sortGroupAscending;
291 // SORTS GROUPS BY SIZE
292 // ////////////////////
293 for (int i = 0; i < align.getGroups().size(); i++)
295 SequenceGroup sg = (SequenceGroup) align.getGroups().elementAt(i);
297 for (int j = 0; j < groups.size(); j++)
299 SequenceGroup sg2 = (SequenceGroup) groups.elementAt(j);
301 if (sg.getSize() > sg2.getSize())
303 groups.insertElementAt(sg, j);
309 if (!groups.contains(sg))
311 groups.addElement(sg);
315 // NOW ADD SEQUENCES MAINTAINING ALIGNMENT ORDER
316 // /////////////////////////////////////////////
317 Vector seqs = new Vector();
319 for (int i = 0; i < groups.size(); i++)
321 SequenceGroup sg = (SequenceGroup) groups.elementAt(i);
322 SequenceI[] orderedseqs = sg.getSequencesInOrder(align);
324 for (int j = 0; j < orderedseqs.length; j++)
326 seqs.addElement(orderedseqs[j]);
330 if (sortGroupAscending)
332 setOrder(align, seqs);
336 setReverseOrder(align,
337 vectorSubsetToArray(seqs, align.getSequences()));
342 * Converts Vector to array. java 1.18 does not have Vector.toArray()
345 * Vector of SequenceI objects
347 * @return array of Sequence[]
349 private static SequenceI[] vectorToArray(Vector tmp)
351 SequenceI[] seqs = new SequenceI[tmp.size()];
353 for (int i = 0; i < tmp.size(); i++)
355 seqs[i] = (SequenceI) tmp.elementAt(i);
369 * @return DOCUMENT ME!
371 private static SequenceI[] vectorSubsetToArray(Vector tmp, Vector mask)
373 Vector seqs = new Vector();
375 boolean[] tmask = new boolean[mask.size()];
377 for (i = 0; i < mask.size(); i++)
382 for (i = 0; i < tmp.size(); i++)
384 Object sq = tmp.elementAt(i);
385 idx = mask.indexOf(sq);
386 if (idx > -1 && tmask[idx])
393 for (i = 0; i < tmask.length; i++)
397 seqs.addElement(mask.elementAt(i));
401 return vectorToArray(seqs);
405 * Sorts by a given AlignmentOrder object
410 * specified order for alignment
412 public static void sortBy(AlignmentI align, AlignmentOrder order)
414 // Get an ordered vector of sequences which may also be present in align
415 Vector tmp = order.getOrder();
417 if (lastOrder == order)
419 sortOrderAscending = !sortOrderAscending;
423 sortOrderAscending = true;
426 if (sortOrderAscending)
428 setOrder(align, tmp);
432 setReverseOrder(align, vectorSubsetToArray(tmp, align.getSequences()));
444 * @return DOCUMENT ME!
446 private static Vector getOrderByTree(AlignmentI align, NJTree tree)
448 int nSeq = align.getHeight();
450 Vector tmp = new Vector();
452 tmp = _sortByTree(tree.getTopNode(), tmp, align.getSequences());
454 if (tmp.size() != nSeq)
456 // TODO: JBPNote - decide if this is always an error
457 // (eg. not when a tree is associated to another alignment which has more
459 if (tmp.size() != nSeq)
461 addStrays(align, tmp);
464 if (tmp.size() != nSeq)
466 System.err.println("WARNING: tmp.size()=" + tmp.size() + " != nseq="
467 + nSeq + " in getOrderByTree - tree contains sequences not in alignment");
475 * Sorts the alignment by a given tree
482 public static void sortByTree(AlignmentI align, NJTree tree)
484 Vector tmp = getOrderByTree(align, tree);
486 // tmp should properly permute align with tree.
487 if (lastTree != tree)
489 sortTreeAscending = true;
494 sortTreeAscending = !sortTreeAscending;
497 if (sortTreeAscending)
499 setOrder(align, tmp);
503 setReverseOrder(align, vectorSubsetToArray(tmp, align.getSequences()));
515 private static void addStrays(AlignmentI align, Vector seqs)
517 int nSeq = align.getHeight();
519 for (int i = 0; i < nSeq; i++)
521 if (!seqs.contains(align.getSequenceAt(i)))
523 seqs.addElement(align.getSequenceAt(i));
527 if (nSeq != seqs.size())
530 .println("ERROR: Size still not right even after addStrays");
544 * @return DOCUMENT ME!
546 private static Vector _sortByTree(SequenceNode node, Vector tmp,
554 SequenceNode left = (SequenceNode) node.left();
555 SequenceNode right = (SequenceNode) node.right();
557 if ((left == null) && (right == null))
559 if (!node.isPlaceholder() && (node.element() != null))
561 if (node.element() instanceof SequenceI)
563 if (!tmp.contains(node.element())) // && (seqset==null || seqset.size()==0 || seqset.contains(tmp)))
565 tmp.addElement((SequenceI) node.element());
574 _sortByTree(left, tmp, seqset);
575 _sortByTree(right, tmp, seqset);
582 // Alignment.sortBy(OrderObj) - sequence of sequence pointer refs in
587 * recover the order of sequences given by the safe numbering scheme introducd
588 * SeqsetUtils.uniquify.
590 public static void recoverOrder(SequenceI[] alignment)
592 float[] ids = new float[alignment.length];
594 for (int i = 0; i < alignment.length; i++)
596 ids[i] = (new Float(alignment[i].getName().substring(8)))
600 jalview.util.QuickSort.sort(ids, alignment);
604 * Sort sequence in order of increasing score attribute for annotation with a
605 * particular scoreLabel. Or reverse if same label was used previously
608 * exact label for sequence associated AlignmentAnnotation scores to
611 * sequences to be sorted
613 public static void sortByAnnotationScore(String scoreLabel,
614 AlignmentI alignment)
616 SequenceI[] seqs = alignment.getSequencesArray();
617 boolean[] hasScore = new boolean[seqs.length]; // per sequence score
619 int hasScores = 0; // number of scores present on set
620 double[] scores = new double[seqs.length];
621 double min = 0, max = 0;
622 for (int i = 0; i < seqs.length; i++)
624 AlignmentAnnotation[] scoreAnn = seqs[i].getAnnotation(scoreLabel);
625 if (scoreAnn != null)
629 scores[i] = scoreAnn[0].getScore(); // take the first instance of this
633 max = min = scores[i];
654 return; // do nothing - no scores present to sort by.
656 if (hasScores < seqs.length)
658 for (int i = 0; i < seqs.length; i++)
662 scores[i] = (max + i + 1.0);
667 jalview.util.QuickSort.sort(scores, seqs);
668 if (lastSortByScore != scoreLabel)
670 lastSortByScore = scoreLabel;
671 setOrder(alignment, seqs);
675 setReverseOrder(alignment, seqs);
680 * types of feature ordering: Sort by score : average score - or total score -
681 * over all features in region Sort by feature label text: (or if null -
682 * feature type text) - numerical or alphabetical Sort by feature density:
683 * based on counts - ignoring individual text or scores for each feature
685 public static String FEATURE_SCORE = "average_score";
687 public static String FEATURE_LABEL = "text";
689 public static String FEATURE_DENSITY = "density";
692 * sort the alignment using the features on each sequence found between start
693 * and stop with the given featureLabel (and optional group qualifier)
695 * @param featureLabel
700 * (-1 to include non-positional features)
702 * (-1 to only sort on non-positional features)
704 * - aligned sequences containing features
706 * - one of the string constants FEATURE_SCORE, FEATURE_LABEL,
709 public static void sortByFeature(String featureLabel, String groupLabel,
710 int start, int stop, AlignmentI alignment, String method)
712 sortByFeature(featureLabel == null ? null : new String[]
713 { featureLabel }, groupLabel == null ? null : new String[]
714 { groupLabel }, start, stop, alignment, method);
717 private static boolean containsIgnoreCase(final String lab,
728 for (int q = 0; q < labs.length; q++)
730 if (labs[q] != null && lab.equalsIgnoreCase(labs[q]))
738 public static void sortByFeature(String[] featureLabels,
739 String[] groupLabels, int start, int stop, AlignmentI alignment,
742 if (method != FEATURE_SCORE && method != FEATURE_LABEL
743 && method != FEATURE_DENSITY)
746 "Implementation Error - sortByFeature method must be one of FEATURE_SCORE, FEATURE_LABEL or FEATURE_DENSITY.");
748 boolean ignoreScore = method != FEATURE_SCORE;
749 StringBuffer scoreLabel = new StringBuffer();
750 scoreLabel.append(start + stop + method);
751 // This doesn't quite work yet - we'd like to have a canonical ordering that
752 // can be preserved from call to call
753 for (int i = 0; featureLabels != null && i < featureLabels.length; i++)
755 scoreLabel.append(featureLabels[i] == null ? "null"
758 for (int i = 0; groupLabels != null && i < groupLabels.length; i++)
760 scoreLabel.append(groupLabels[i] == null ? "null" : groupLabels[i]);
762 SequenceI[] seqs = alignment.getSequencesArray();
764 boolean[] hasScore = new boolean[seqs.length]; // per sequence score
766 int hasScores = 0; // number of scores present on set
767 double[] scores = new double[seqs.length];
768 int[] seqScores = new int[seqs.length];
769 Object[] feats = new Object[seqs.length];
770 double min = 0, max = 0;
771 for (int i = 0; i < seqs.length; i++)
773 SequenceFeature[] sf = seqs[i].getSequenceFeatures();
774 if (sf == null && seqs[i].getDatasetSequence() != null)
776 sf = seqs[i].getDatasetSequence().getSequenceFeatures();
780 sf = new SequenceFeature[0];
784 SequenceFeature[] tmp = new SequenceFeature[sf.length];
785 for (int s = 0; s < tmp.length; s++)
791 int sstart = (start == -1) ? start : seqs[i].findPosition(start);
792 int sstop = (stop == -1) ? stop : seqs[i].findPosition(stop);
796 for (int f = 0; f < sf.length; f++)
798 // filter for selection criteria
800 // ignore features outwith alignment start-stop positions.
801 (sf[f].end < sstart || sf[f].begin > sstop) ||
802 // or ignore based on selection criteria
803 (featureLabels != null && !AlignmentSorter
804 .containsIgnoreCase(sf[f].type, featureLabels))
805 || (groupLabels != null
806 // problem here: we cannot eliminate null feature group features
807 && (sf[f].getFeatureGroup() != null && !AlignmentSorter
808 .containsIgnoreCase(sf[f].getFeatureGroup(),
811 // forget about this feature
817 // or, also take a look at the scores if necessary.
818 if (!ignoreScore && sf[f].getScore() != Float.NaN)
820 if (seqScores[i] == 0)
826 scores[i] += sf[f].getScore(); // take the first instance of this
831 SequenceFeature[] fs;
832 feats[i] = fs = new SequenceFeature[n];
836 for (int f = 0; f < sf.length; f++)
840 ((SequenceFeature[]) feats[i])[n++] = sf[f];
843 if (method == FEATURE_LABEL)
845 // order the labels by alphabet
846 String[] labs = new String[fs.length];
847 for (int l = 0; l < labs.length; l++)
849 labs[l] = (fs[l].getDescription() != null ? fs[l]
850 .getDescription() : fs[l].getType());
852 jalview.util.QuickSort.sort(labs, ((Object[]) feats[i]));
857 // compute average score
858 scores[i] /= seqScores[i];
859 // update the score bounds.
862 max = min = scores[i];
878 if (method == FEATURE_SCORE)
882 return; // do nothing - no scores present to sort by.
885 if (hasScores < seqs.length)
887 for (int i = 0; i < seqs.length; i++)
891 scores[i] = (max + 1 + i);
895 int nf = (feats[i] == null) ? 0
896 : ((SequenceFeature[]) feats[i]).length;
897 // System.err.println("Sorting on Score: seq "+seqs[i].getName()+
898 // " Feats: "+nf+" Score : "+scores[i]);
903 jalview.util.QuickSort.sort(scores, seqs);
905 else if (method == FEATURE_DENSITY)
908 // break ties between equivalent numbers for adjacent sequences by adding
909 // 1/Nseq*i on the original order
910 double fr = 0.9 / (1.0 * seqs.length);
911 for (int i = 0; i < seqs.length; i++)
914 scores[i] = (0.05 + fr * i)
915 + (nf = ((feats[i] == null) ? 0.0
916 : 1.0 * ((SequenceFeature[]) feats[i]).length));
917 // System.err.println("Sorting on Density: seq "+seqs[i].getName()+
918 // " Feats: "+nf+" Score : "+scores[i]);
920 jalview.util.QuickSort.sort(scores, seqs);
924 if (method == FEATURE_LABEL)
926 throw new Error("Not yet implemented.");
929 if (lastSortByFeatureScore == null
930 || !scoreLabel.toString().equals(lastSortByFeatureScore))
932 sortByFeatureScoreAscending = true;
936 sortByFeatureScoreAscending = !sortByFeatureScoreAscending;
938 if (sortByFeatureScoreAscending)
940 setOrder(alignment, seqs);
944 setReverseOrder(alignment, seqs);
946 lastSortByFeatureScore = scoreLabel.toString();