From: gmungoc <g.m.carstairs@dundee.ac.uk>
Date: Tue, 7 Feb 2017 13:20:12 +0000 (+0000)
Subject: JAL-2397 flag provided for 2.10.1 PCA gap calculation treatment
X-Git-Tag: Release_2_10_3b1~343^2~2
X-Git-Url: http://source.jalview.org/gitweb/?a=commitdiff_plain;h=454f89f88d84083b9b654367f728b4a9dcb43e65;p=jalview.git

JAL-2397 flag provided for 2.10.1 PCA gap calculation treatment
---

diff --git a/examples/groovy/pcaMode.groovy b/examples/groovy/pcaMode.groovy
deleted file mode 100644
index 817e6e1..0000000
--- a/examples/groovy/pcaMode.groovy
+++ /dev/null
@@ -1,55 +0,0 @@
-/*
- * PCA calculation treatment of gaps (and C->R score) changed in Jalview 2.10.2.
- * To restore behaviour prior as at 2.10.1, run the script below
- * NB this will change Tree calculations using similarity matrices
- */
-
-/*
- * START 2.10.1 mode script
- */
-import jalview.schemes.ResidueProperties 
-
-for (int row = 0 ; row <= ResidueProperties.maxProteinIndex; row++) 
-{ 
-  ResidueProperties.BLOSUM62[row][23] = ResidueProperties.BLOSUM62[row][22] 
-  ResidueProperties.BLOSUM62[23][row] = ResidueProperties.BLOSUM62[row][22] 
-  ResidueProperties.PAM250[row][23] = ResidueProperties.PAM250[row][22] 
-  ResidueProperties.PAM250[23][row] = ResidueProperties.PAM250[row][22] 
-} 
-ResidueProperties.BLOSUM62[4][1]=3 // C-R score
-for (int row = 0 ; row <= ResidueProperties.maxNucleotideIndex; row++) 
-{ 
-  ResidueProperties.DNA[row][10] = ResidueProperties.DNA[row][9] 
-  ResidueProperties.DNA[10][row] = ResidueProperties.DNA[row][9] 
-}
-/*
- * END 2.10.1 mode script
- */
-
-/*
- * To reinstate 2.10.2 calculation mode, including Tree calculations,
- * run the script below
- */
-/*
- * START 2.10.2 mode script
- */
- import jalview.schemes.ResidueProperties 
-
-for (int row = 0 ; row < ResidueProperties.maxProteinIndex; row++) 
-{ 
-  ResidueProperties.BLOSUM62[row][23] = -4 
-  ResidueProperties.BLOSUM62[23][row] = -4 
-  ResidueProperties.PAM250[row][23] = -8 
-  ResidueProperties.PAM250[23][row] = -8 
-} 
-ResidueProperties.BLOSUM62[23][23] = 1 
-ResidueProperties.PAM250[23][23] = 1 
-ResidueProperties.BLOSUM62[4][1]=-3 // C-R score
-for (int row = 0 ; row <= ResidueProperties.maxNucleotideIndex; row++) 
-{ 
-  ResidueProperties.DNA[row][10] = 1 
-  ResidueProperties.DNA[10][row] = 1 
-}
-/*
- * END 2.10.2 mode script
- */
diff --git a/src/jalview/viewmodel/PCAModel.java b/src/jalview/viewmodel/PCAModel.java
index 55e7300..0623dab 100644
--- a/src/jalview/viewmodel/PCAModel.java
+++ b/src/jalview/viewmodel/PCAModel.java
@@ -30,6 +30,13 @@ import java.util.Vector;
 
 public class PCAModel
 {
+  /*
+   * Jalview 2.10.1 treated gaps as X (peptide) or N (nucleotide)
+   * for pairwise scoring; 2.10.2 uses gap score (last column) in
+   * score matrix (JAL-2397)
+   * Set this flag to true (via Groovy) for 2.10.1 behaviour
+   */
+  private static boolean scoreGapAsAny = false;
 
   public PCAModel(AlignmentView seqstrings2, SequenceI[] seqs2,
           boolean nucleotide2)
@@ -69,7 +76,8 @@ public class PCAModel
 
   public void run()
   {
-    String[] sequenceStrings = seqstrings.getSequenceStrings(' ');
+    char gapChar = scoreGapAsAny ? (nucleotide ? 'N' : 'X') : ' ';
+    String[] sequenceStrings = seqstrings.getSequenceStrings(gapChar);
     pca = new PCA(sequenceStrings, nucleotide,
             score_matrix);
     pca.setJvCalcMode(jvCalcMode);