From 427421de47dacacac55ac29deca78e2f0a1c5408 Mon Sep 17 00:00:00 2001 From: gmungoc Date: Wed, 30 Aug 2017 12:16:22 +0100 Subject: [PATCH] JAL-2685 use actual selection for report of pairwise alignment, help page updated --- help/html/calculations/pairwise.html | 16 ++-- src/jalview/analysis/AlignSeq.java | 96 +++++++++++----------- src/jalview/gui/PairwiseAlignPanel.java | 63 ++++++++------ test/jalview/gui/PairwiseAlignmentPanelTest.java | 73 ++++++++++++++++ 4 files changed, 167 insertions(+), 81 deletions(-) create mode 100644 test/jalview/gui/PairwiseAlignmentPanelTest.java diff --git a/help/html/calculations/pairwise.html b/help/html/calculations/pairwise.html index bb80b84..bf677e6 100755 --- a/help/html/calculations/pairwise.html +++ b/help/html/calculations/pairwise.html @@ -38,11 +38,15 @@

Gap open : 12
Gap extend : 2

-

When you select the pairwise alignment option a new window will - come up which will display the alignments in a text format as they - are calculated. Also displayed is information about the alignment - such as alignment score, length and percentage identity between the - sequences.

-

 

+

When you select the pairwise alignment option, a new window will + come up which displays the alignments in a text format, for example:

+

+ 

+    FER1_SPIOL/5-13 TTMMGMAT

+                    |. .. ||

+    FER1_MESCR/5-15 TAALSGAT
+
shows the aligned sequences, where '|' links identical residues, and (for peptide) '.' links residues that have a positive PAM250 score. +

The window also shows information about the alignment such as alignment score, length and percentage identity between the sequences.

+

A button is also provided to allow you to view the sequences as an alignment.

diff --git a/src/jalview/analysis/AlignSeq.java b/src/jalview/analysis/AlignSeq.java index 34a21e6..15ea2fc 100755 --- a/src/jalview/analysis/AlignSeq.java +++ b/src/jalview/analysis/AlignSeq.java @@ -36,6 +36,7 @@ import jalview.util.MessageManager; import java.awt.Color; import java.awt.Graphics; +import java.io.PrintStream; import java.util.ArrayList; import java.util.Arrays; import java.util.List; @@ -49,6 +50,14 @@ import java.util.StringTokenizer; */ public class AlignSeq { + private static final int MAX_NAME_LENGTH = 30; + + private static final int GAP_OPEN_COST = 120; + + private static final int GAP_EXTEND_COST = 20; + + private static final int GAP_INDEX = -1; + public static final String PEP = "pep"; public static final String DNA = "dna"; @@ -61,7 +70,7 @@ public class AlignSeq float[][] F; - int[][] traceback; + int[][] traceback; // todo is this actually used? int[] seq1; @@ -96,30 +105,20 @@ public class AlignSeq /** DOCUMENT ME!! */ public int seq2start; - /** DOCUMENT ME!! */ public int seq2end; int count; - /** DOCUMENT ME!! */ public float maxscore; - float pid; - int prev = 0; - int gapOpen = 120; - - int gapExtend = 20; - StringBuffer output = new StringBuffer(); String type; // AlignSeq.PEP or AlignSeq.DNA private ScoreMatrix scoreMatrix; - private static final int GAP_INDEX = -1; - /** * Creates a new AlignSeq object. * @@ -378,11 +377,10 @@ public class AlignSeq } } - // System.out.println(maxi + " " + maxj + " " + score[maxi][maxj]); int i = maxi; int j = maxj; int trace; - maxscore = score[i][j] / 10; + maxscore = score[i][j] / 10f; seq1end = maxi + 1; seq2end = maxj + 1; @@ -451,49 +449,48 @@ public class AlignSeq /** * DOCUMENT ME! */ - public void printAlignment(java.io.PrintStream os) + public void printAlignment(PrintStream os) { // TODO: Use original sequence characters rather than re-translated // characters in output // Find the biggest id length for formatting purposes - String s1id = s1.getName(), s2id = s2.getName(); - int maxid = s1.getName().length(); - if (s2.getName().length() > maxid) - { - maxid = s2.getName().length(); - } - if (maxid > 30) + String s1id = getAlignedSeq1().getDisplayId(true); + String s2id = getAlignedSeq2().getDisplayId(true); + int nameLength = Math.max(s1id.length(), s2id.length()); + if (nameLength > MAX_NAME_LENGTH) { - maxid = 30; + int truncateBy = nameLength - MAX_NAME_LENGTH; + nameLength = MAX_NAME_LENGTH; // JAL-527 - truncate the sequence ids - if (s1.getName().length() > maxid) + if (s1id.length() > nameLength) { - s1id = s1.getName().substring(0, 30); + int slashPos = s1id.lastIndexOf('/'); + s1id = s1id.substring(0, slashPos - truncateBy) + + s1id.substring(slashPos); } - if (s2.getName().length() > maxid) + if (s2id.length() > nameLength) { - s2id = s2.getName().substring(0, 30); + int slashPos = s2id.lastIndexOf('/'); + s2id = s2id.substring(0, slashPos - truncateBy) + + s2id.substring(slashPos); } } - int len = 72 - maxid - 1; + int len = 72 - nameLength - 1; int nochunks = ((aseq1.length - count) / len) + ((aseq1.length - count) % len > 0 ? 1 : 0); - pid = 0; + float pid = 0f; output.append("Score = ").append(score[maxi][maxj]).append(NEWLINE); output.append("Length of alignment = ") .append(String.valueOf(aseq1.length - count)).append(NEWLINE); output.append("Sequence "); - output.append(new Format("%" + maxid + "s").form(s1.getName())); - output.append(" : ").append(String.valueOf(s1.getStart())) - .append(" - ").append(String.valueOf(s1.getEnd())); + Format nameFormat = new Format("%" + nameLength + "s"); + output.append(nameFormat.form(s1id)); output.append(" (Sequence length = ") .append(String.valueOf(s1str.length())).append(")") .append(NEWLINE); output.append("Sequence "); - output.append(new Format("%" + maxid + "s").form(s2.getName())); - output.append(" : ").append(String.valueOf(s2.getStart())) - .append(" - ").append(String.valueOf(s2.getEnd())); + output.append(nameFormat.form(s2id)); output.append(" (Sequence length = ") .append(String.valueOf(s2str.length())).append(")") .append(NEWLINE).append(NEWLINE); @@ -503,7 +500,7 @@ public class AlignSeq for (int j = 0; j < nochunks; j++) { // Print the first aligned sequence - output.append(new Format("%" + (maxid) + "s").form(s1id)).append(" "); + output.append(nameFormat.form(s1id)).append(" "); for (int i = 0; i < len; i++) { @@ -514,7 +511,7 @@ public class AlignSeq } output.append(NEWLINE); - output.append(new Format("%" + (maxid) + "s").form(" ")).append(" "); + output.append(nameFormat.form(" ")).append(" "); /* * Print out the match symbols: @@ -534,7 +531,7 @@ public class AlignSeq pid++; output.append("|"); } - else if (type.equals("pep")) + else if (PEP.equals(type)) { if (pam250.getPairwiseScore(c1, c2) > 0) { @@ -554,8 +551,7 @@ public class AlignSeq // Now print the second aligned sequence output = output.append(NEWLINE); - output = output.append(new Format("%" + (maxid) + "s").form(s2id)) - .append(" "); + output = output.append(nameFormat.form(s2id)).append(" "); for (int i = 0; i < len; i++) { @@ -565,11 +561,12 @@ public class AlignSeq } } - output.append(NEWLINE).append(NEWLINE); + output.append(NEWLINE);// .append(NEWLINE); } pid = pid / (aseq1.length - count) * 100; - output = output.append(new Format("Percentage ID = %2.2f\n").form(pid)); + output.append(new Format("Percentage ID = %3.2f\n").form(pid)); + output.append(NEWLINE); try { os.print(output.toString()); @@ -591,7 +588,6 @@ public class AlignSeq public int findTrace(int i, int j) { int t = 0; - // float pairwiseScore = lookup[seq1[i]][seq2[j]]; float pairwiseScore = scoreMatrix.getPairwiseScore(s1str.charAt(i), s2str.charAt(j)); float max = score[i - 1][j - 1] + (pairwiseScore * 10); @@ -640,19 +636,19 @@ public class AlignSeq // top left hand element score[0][0] = scoreMatrix.getPairwiseScore(s1str.charAt(0), s2str.charAt(0)) * 10; - E[0][0] = -gapExtend; + E[0][0] = -GAP_EXTEND_COST; F[0][0] = 0; // Calculate the top row first for (int j = 1; j < m; j++) { // What should these values be? 0 maybe - E[0][j] = max(score[0][j - 1] - gapOpen, E[0][j - 1] - gapExtend); - F[0][j] = -gapExtend; + E[0][j] = max(score[0][j - 1] - GAP_OPEN_COST, E[0][j - 1] - GAP_EXTEND_COST); + F[0][j] = -GAP_EXTEND_COST; float pairwiseScore = scoreMatrix.getPairwiseScore(s1str.charAt(0), s2str.charAt(j)); - score[0][j] = max(pairwiseScore * 10, -gapOpen, -gapExtend); + score[0][j] = max(pairwiseScore * 10, -GAP_OPEN_COST, -GAP_EXTEND_COST); traceback[0][j] = 1; } @@ -660,8 +656,8 @@ public class AlignSeq // Now do the left hand column for (int i = 1; i < n; i++) { - E[i][0] = -gapOpen; - F[i][0] = max(score[i - 1][0] - gapOpen, F[i - 1][0] - gapExtend); + E[i][0] = -GAP_OPEN_COST; + F[i][0] = max(score[i - 1][0] - GAP_OPEN_COST, F[i - 1][0] - GAP_EXTEND_COST); float pairwiseScore = scoreMatrix.getPairwiseScore(s1str.charAt(i), s2str.charAt(0)); @@ -674,8 +670,8 @@ public class AlignSeq { for (int j = 1; j < m; j++) { - E[i][j] = max(score[i][j - 1] - gapOpen, E[i][j - 1] - gapExtend); - F[i][j] = max(score[i - 1][j] - gapOpen, F[i - 1][j] - gapExtend); + E[i][j] = max(score[i][j - 1] - GAP_OPEN_COST, E[i][j - 1] - GAP_EXTEND_COST); + F[i][j] = max(score[i - 1][j] - GAP_OPEN_COST, F[i - 1][j] - GAP_EXTEND_COST); float pairwiseScore = scoreMatrix.getPairwiseScore(s1str.charAt(i), s2str.charAt(j)); diff --git a/src/jalview/gui/PairwiseAlignPanel.java b/src/jalview/gui/PairwiseAlignPanel.java index f75407c..4aea4b1 100755 --- a/src/jalview/gui/PairwiseAlignPanel.java +++ b/src/jalview/gui/PairwiseAlignPanel.java @@ -22,7 +22,8 @@ package jalview.gui; import jalview.analysis.AlignSeq; import jalview.datamodel.Alignment; -import jalview.datamodel.Sequence; +import jalview.datamodel.AlignmentView; +import jalview.datamodel.SequenceGroup; import jalview.datamodel.SequenceI; import jalview.jbgui.GPairwiseAlignPanel; import jalview.util.MessageManager; @@ -40,49 +41,56 @@ import java.util.Vector; public class PairwiseAlignPanel extends GPairwiseAlignPanel { + private static final String DASHES = "---------------------\n"; + AlignmentViewport av; - Vector sequences; + Vector sequences; /** * Creates a new PairwiseAlignPanel object. * - * @param av + * @param viewport * DOCUMENT ME! */ - public PairwiseAlignPanel(AlignmentViewport av) + public PairwiseAlignPanel(AlignmentViewport viewport) { super(); - this.av = av; + this.av = viewport; - sequences = new Vector(); + sequences = new Vector(); - SequenceI[] seqs; - String[] seqStrings = av.getViewAsString(true); + SequenceGroup selectionGroup = viewport.getSelectionGroup(); + boolean isSelection = selectionGroup != null + && selectionGroup.getSize() > 0; + AlignmentView view = viewport.getAlignmentView(isSelection); + // String[] seqStrings = viewport.getViewAsString(true); + String[] seqStrings = view.getSequenceStrings(viewport + .getGapCharacter()); - if (av.getSelectionGroup() == null) + SequenceI[] seqs; + if (isSelection) { - seqs = av.getAlignment().getSequencesArray(); + seqs = (SequenceI[]) view.getAlignmentAndHiddenColumns(viewport + .getGapCharacter())[0]; } else { - seqs = av.getSelectionGroup().getSequencesInOrder(av.getAlignment()); + seqs = av.getAlignment().getSequencesArray(); } - String type = (av.getAlignment().isNucleotide()) ? AlignSeq.DNA + String type = (viewport.getAlignment().isNucleotide()) ? AlignSeq.DNA : AlignSeq.PEP; float[][] scores = new float[seqs.length][seqs.length]; double totscore = 0; int count = seqs.length; - - Sequence seq; + boolean first = true; for (int i = 1; i < count; i++) { for (int j = 0; j < i; j++) { - AlignSeq as = new AlignSeq(seqs[i], seqStrings[i], seqs[j], seqStrings[j], type); @@ -94,9 +102,15 @@ public class PairwiseAlignPanel extends GPairwiseAlignPanel as.calcScoreMatrix(); as.traceAlignment(); + if (!first) + { + System.out.println(DASHES); + textarea.append(DASHES); + } + first = false; as.printAlignment(System.out); - scores[i][j] = (float) as.getMaxScore() - / (float) as.getASeq1().length; + scores[i][j] = as.getMaxScore() + / as.getASeq1().length; totscore = totscore + scores[i][j]; textarea.append(as.getOutput()); @@ -112,19 +126,17 @@ public class PairwiseAlignPanel extends GPairwiseAlignPanel for (int i = 0; i < count; i++) { - jalview.util.Format.print(System.out, "%s \n", - ("" + i) + " " + seqs[i].getName()); + System.out.println(String.format("%d %s", i, + seqs[i].getDisplayId(true))); } - System.out.println("\n"); - for (int i = 0; i < count; i++) { for (int j = 0; j < i; j++) { - jalview.util.Format.print(System.out, "%7.3f", - scores[i][j] / totscore); + System.out.print(String.format("%7.3f", scores[i][j] / totscore)); } + System.out.println(); } System.out.println("\n"); @@ -137,13 +149,14 @@ public class PairwiseAlignPanel extends GPairwiseAlignPanel * @param e * DOCUMENT ME! */ + @Override protected void viewInEditorButton_actionPerformed(ActionEvent e) { - Sequence[] seq = new Sequence[sequences.size()]; + SequenceI[] seq = new SequenceI[sequences.size()]; for (int i = 0; i < sequences.size(); i++) { - seq[i] = (Sequence) sequences.elementAt(i); + seq[i] = sequences.elementAt(i); } AlignFrame af = new AlignFrame(new Alignment(seq), diff --git a/test/jalview/gui/PairwiseAlignmentPanelTest.java b/test/jalview/gui/PairwiseAlignmentPanelTest.java new file mode 100644 index 0000000..4d09f43 --- /dev/null +++ b/test/jalview/gui/PairwiseAlignmentPanelTest.java @@ -0,0 +1,73 @@ +package jalview.gui; + +import static org.testng.Assert.assertEquals; + +import jalview.datamodel.AlignmentI; +import jalview.datamodel.SequenceGroup; +import jalview.io.DataSourceType; +import jalview.io.FileLoader; + +import javax.swing.JTextArea; + +import junit.extensions.PA; + +import org.testng.annotations.Test; + +public class PairwiseAlignmentPanelTest +{ + @Test(groups = "Functional") + public void testConstructor_withSelectionGroup() + { + AlignFrame af = new FileLoader().LoadFileWaitTillLoaded( + "examples/uniref50.fa", DataSourceType.FILE); + AlignViewport viewport = af.getViewport(); + AlignmentI al = viewport.getAlignment(); + + /* + * select columns 29-36 of sequences 4 and 5 for alignment + * Q93XJ9_SOLTU/23-29 L-KAISNV + * FER1_PEA/26-32 V-TTTKAF + */ + SequenceGroup sg = new SequenceGroup(); + sg.addSequence(al.getSequenceAt(3), false); + sg.addSequence(al.getSequenceAt(4), false); + sg.setStartRes(28); + sg.setEndRes(35); + viewport.setSelectionGroup(sg); + + PairwiseAlignPanel testee = new PairwiseAlignPanel(viewport); + + String text = ((JTextArea) PA.getValue(testee, "textarea")).getText(); + String expected = "Score = 80.0\n" + "Length of alignment = 4\n" + + "Sequence FER1_PEA/29-32 (Sequence length = 7)\n" + + "Sequence Q93XJ9_SOLTU/23-26 (Sequence length = 7)\n\n" + + " FER1_PEA/29-32 TKAF\n" + " ||.\n" + + "Q93XJ9_SOLTU/23-26 LKAI\n" + "Percentage ID = 50.00\n\n"; + assertEquals(text, expected); + } + + /** + * This test aligns the same sequences as testConstructor_withSelectionGroup + * but as a complete alignment (no selection). Note that in fact the user is + * currently required to make a selection in order to calculate pairwise + * alignments, so this case does not arise. + */ + @Test(groups = "Functional") + public void testConstructor_noSelectionGroup() + { + String seqs = ">Q93XJ9_SOLTU/23-29\nL-KAISNV\n>FER1_PEA/26-32\nV-TTTKAF\n"; + AlignFrame af = new FileLoader().LoadFileWaitTillLoaded(seqs, + DataSourceType.PASTE); + AlignViewport viewport = af.getViewport(); + + PairwiseAlignPanel testee = new PairwiseAlignPanel(viewport); + + String text = ((JTextArea) PA.getValue(testee, "textarea")).getText(); + String expected = "Score = 80.0\n" + "Length of alignment = 4\n" + + "Sequence FER1_PEA/29-32 (Sequence length = 7)\n" + + "Sequence Q93XJ9_SOLTU/23-26 (Sequence length = 7)\n\n" + + " FER1_PEA/29-32 TKAF\n" + " ||.\n" + + "Q93XJ9_SOLTU/23-26 LKAI\n" + "Percentage ID = 50.00\n\n"; + assertEquals(text, expected); + } +} -- 1.7.10.2