/*
- * Jalview - A Sequence Alignment Editor and Viewer (Version 2.8.2)
- * Copyright (C) 2014 The Jalview Authors
+ * Jalview - A Sequence Alignment Editor and Viewer ($$Version-Rel$$)
+ * Copyright (C) $$Year-Rel$$ The Jalview Authors
*
* This file is part of Jalview.
*
*/
package jalview.analysis;
+import jalview.analysis.scoremodels.PIDModel;
+import jalview.analysis.scoremodels.ScoreMatrix;
+import jalview.analysis.scoremodels.ScoreModels;
+import jalview.analysis.scoremodels.SimilarityParams;
import jalview.datamodel.AlignmentAnnotation;
import jalview.datamodel.AlignmentI;
import jalview.datamodel.Mapping;
import jalview.datamodel.Sequence;
import jalview.datamodel.SequenceI;
-import jalview.schemes.ResidueProperties;
-import jalview.schemes.ScoreMatrix;
import jalview.util.Comparison;
import jalview.util.Format;
import jalview.util.MapList;
public static final String DNA = "dna";
- static String[] dna =
- { "A", "C", "G", "T", "-" };
+ private static final String NEWLINE = System.lineSeparator();
- // "C", "T", "A", "G", "-"};
- static String[] pep =
- { "A", "R", "N", "D", "C", "Q", "E", "G", "H", "I", "L", "K", "M", "F",
- "P", "S", "T", "W", "Y", "V", "B", "Z", "X", "-" };
+ float[][] score;
- int[][] score;
+ float[][] E;
- int[][] E;
-
- int[][] F;
+ float[][] F;
int[][] traceback;
int count;
/** DOCUMENT ME!! */
- public int maxscore;
+ public float maxscore;
float pid;
int gapExtend = 20;
- int[][] lookup = ResidueProperties.getBLOSUM62();
-
- String[] intToStr = pep;
+ float[][] lookup;
- int defInt = 23;
+ int gapIndex = 23;
StringBuffer output = new StringBuffer();
- String type;
+ String type; // AlignSeq.PEP or AlignSeq.DNA
- private int[] charToInt;
+ private ScoreMatrix scoreModel;
/**
* Creates a new AlignSeq object.
*
- * @param s1
- * DOCUMENT ME!
- * @param s2
- * DOCUMENT ME!
- * @param type
- * DOCUMENT ME!
+ * @param s1 first sequence for alignment
+ * @param s2 second sequence for alignment
+ * @param type molecule type, either AlignSeq.PEP or AlignSeq.DNA
*/
public AlignSeq(SequenceI s1, SequenceI s2, String type)
{
- SeqInit(s1, s1.getSequenceAsString(), s2, s2.getSequenceAsString(),
+ seqInit(s1, s1.getSequenceAsString(), s2, s2.getSequenceAsString(),
type);
}
public AlignSeq(SequenceI s1, String string1, SequenceI s2,
String string2, String type)
{
- SeqInit(s1, string1.toUpperCase(), s2, string2.toUpperCase(), type);
+ seqInit(s1, string1.toUpperCase(), s2, string2.toUpperCase(), type);
}
/**
*
* @return DOCUMENT ME!
*/
- public int getMaxScore()
+ public float getMaxScore()
{
return maxscore;
}
}
/**
- * DOCUMENT ME!
- *
- * @return DOCUMENT ME!
- */
- public SequenceI getS1()
- {
- return s1;
- }
-
- /**
- * DOCUMENT ME!
- *
- * @return DOCUMENT ME!
- */
- public SequenceI getS2()
- {
- return s2;
- }
-
- /**
*
* @return aligned instance of Seq 1
*/
* @param type
* DNA or PEPTIDE
*/
- public void SeqInit(SequenceI s1, String string1, SequenceI s2,
+ public void seqInit(SequenceI s1, String string1, SequenceI s2,
String string2, String type)
{
this.s1 = s1;
this.s2 = s2;
setDefaultParams(type);
- SeqInit(string1, string2);
- }
-
- /**
- * Construct score matrix for sequences with custom substitution matrix
- *
- * @param s1
- * - sequence 1
- * @param string1
- * - string to use for s1
- * @param s2
- * - sequence 2
- * @param string2
- * - string to use for s2
- * @param scoreMatrix
- * - substitution matrix to use for alignment
- */
- public void SeqInit(SequenceI s1, String string1, SequenceI s2,
- String string2, ScoreMatrix scoreMatrix)
- {
- this.s1 = s1;
- this.s2 = s2;
- setType(scoreMatrix.isDNA() ? AlignSeq.DNA : AlignSeq.PEP);
- lookup = scoreMatrix.getMatrix();
+ seqInit(string1, string2);
}
/**
* @param string1
* @param string2
*/
- private void SeqInit(String string1, String string2)
+ private void seqInit(String string1, String string2)
{
s1str = extractGaps(jalview.util.Comparison.GapChars, string1);
s2str = extractGaps(jalview.util.Comparison.GapChars, string2);
return;
}
- // System.out.println("lookuip " + rt.freeMemory() + " "+ rt.totalMemory());
seq1 = new int[s1str.length()];
- // System.out.println("seq1 " + rt.freeMemory() +" " + rt.totalMemory());
seq2 = new int[s2str.length()];
- // System.out.println("seq2 " + rt.freeMemory() + " " + rt.totalMemory());
- score = new int[s1str.length()][s2str.length()];
+ score = new float[s1str.length()][s2str.length()];
- // System.out.println("score " + rt.freeMemory() + " " + rt.totalMemory());
- E = new int[s1str.length()][s2str.length()];
+ E = new float[s1str.length()][s2str.length()];
- // System.out.println("E " + rt.freeMemory() + " " + rt.totalMemory());
- F = new int[s1str.length()][s2str.length()];
+ F = new float[s1str.length()][s2str.length()];
traceback = new int[s1str.length()][s2str.length()];
- // System.out.println("F " + rt.freeMemory() + " " + rt.totalMemory());
- seq1 = stringToInt(s1str, type);
-
- // System.out.println("seq1 " + rt.freeMemory() + " " + rt.totalMemory());
- seq2 = stringToInt(s2str, type);
-
- // System.out.println("Seq2 " + rt.freeMemory() + " " + rt.totalMemory());
- // long tstart = System.currentTimeMillis();
- // calcScoreMatrix();
- // long tend = System.currentTimeMillis();
- // System.out.println("Time take to calculate score matrix = " +
- // (tend-tstart) + " ms");
- // printScoreMatrix(score);
- // System.out.println();
- // printScoreMatrix(traceback);
- // System.out.println();
- // printScoreMatrix(E);
- // System.out.println();
- // /printScoreMatrix(F);
- // System.out.println();
- // tstart = System.currentTimeMillis();
- // traceAlignment();
- // tend = System.currentTimeMillis();
- // System.out.println("Time take to traceback alignment = " + (tend-tstart)
- // + " ms");
- }
-
- private void setDefaultParams(String type)
- {
- setType(type);
+ seq1 = indexEncode(s1str);
- if (type.equals(AlignSeq.PEP))
- {
- lookup = ResidueProperties.getDefaultPeptideMatrix();
- }
- else if (type.equals(AlignSeq.DNA))
- {
- lookup = ResidueProperties.getDefaultDnaMatrix();
- }
+ seq2 = indexEncode(s2str);
}
- private void setType(String type2)
+ private void setDefaultParams(String moleculeType)
{
- this.type = type2;
- if (type.equals(AlignSeq.PEP))
- {
- intToStr = pep;
- charToInt = ResidueProperties.aaIndex;
- defInt = ResidueProperties.maxProteinIndex;
- }
- else if (type.equals(AlignSeq.DNA))
- {
- intToStr = dna;
- charToInt = ResidueProperties.nucleotideIndex;
- defInt = ResidueProperties.maxNucleotideIndex;
- }
- else
+ if (!PEP.equals(moleculeType) && !DNA.equals(moleculeType))
{
output.append("Wrong type = dna or pep only");
- throw new Error(MessageManager.formatMessage("error.unknown_type_dna_or_pep", new String[]{type2}));
+ throw new Error(MessageManager.formatMessage(
+ "error.unknown_type_dna_or_pep",
+ new String[] { moleculeType }));
}
+
+ type = moleculeType;
+ scoreModel = ScoreModels.getInstance().getDefaultModel(
+ PEP.equals(type));
+ lookup = scoreModel.getMatrix();
+ gapIndex = scoreModel.getMatrixIndex(' ');
}
/**
public void traceAlignment()
{
// Find the maximum score along the rhs or bottom row
- int max = -9999;
+ float max = -Float.MAX_VALUE;
for (int i = 0; i < seq1.length; i++)
{
aseq1 = new int[seq1.length + seq2.length];
aseq2 = new int[seq1.length + seq2.length];
+ StringBuilder sb1 = new StringBuilder(aseq1.length);
+ StringBuilder sb2 = new StringBuilder(aseq2.length);
+
count = (seq1.length + seq2.length) - 1;
- while ((i > 0) && (j > 0))
+ while (i > 0 && j > 0)
{
- if ((aseq1[count] != defInt) && (i >= 0))
- {
- aseq1[count] = seq1[i];
- astr1 = s1str.charAt(i) + astr1;
- }
-
- if ((aseq2[count] != defInt) && (j > 0))
- {
- aseq2[count] = seq2[j];
- astr2 = s2str.charAt(j) + astr2;
- }
+ aseq1[count] = seq1[i];
+ sb1.append(s1str.charAt(i));
+ aseq2[count] = seq2[j];
+ sb2.append(s2str.charAt(j));
trace = findTrace(i, j);
else if (trace == 1)
{
j--;
- aseq1[count] = defInt;
- astr1 = "-" + astr1.substring(1);
+ aseq1[count] = gapIndex;
+ sb1.replace(sb1.length() - 1, sb1.length(), "-");
}
else if (trace == -1)
{
i--;
- aseq2[count] = defInt;
- astr2 = "-" + astr2.substring(1);
+ aseq2[count] = gapIndex;
+ sb2.replace(sb2.length() - 1, sb2.length(), "-");
}
count--;
seq1start = i + 1;
seq2start = j + 1;
- if (aseq1[count] != defInt)
+ if (aseq1[count] != gapIndex)
{
aseq1[count] = seq1[i];
- astr1 = s1str.charAt(i) + astr1;
+ sb1.append(s1str.charAt(i));
}
- if (aseq2[count] != defInt)
+ if (aseq2[count] != gapIndex)
{
aseq2[count] = seq2[j];
- astr2 = s2str.charAt(j) + astr2;
+ sb2.append(s2str.charAt(j));
}
+
+ /*
+ * we built the character strings backwards, so now
+ * reverse them to convert to sequence strings
+ */
+ astr1 = sb1.reverse().toString();
+ astr2 = sb2.reverse().toString();
}
/**
}
}
int len = 72 - maxid - 1;
- int nochunks = ((aseq1.length - count) / len) + 1;
+ int nochunks = ((aseq1.length - count) / len)
+ + ((aseq1.length - count) % len > 0 ? 1 : 0);
pid = 0;
- output.append("Score = " + score[maxi][maxj] + "\n");
- output.append("Length of alignment = " + (aseq1.length - count) + "\n");
+ 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(" : " + s1.getStart() + " - " + s1.getEnd()
- + " (Sequence length = " + s1str.length() + ")\n");
+ output.append(" : ").append(String.valueOf(s1.getStart()))
+ .append(" - ").append(String.valueOf(s1.getEnd()));
+ 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(" : " + s2.getStart() + " - " + s2.getEnd()
- + " (Sequence length = " + s2str.length() + ")\n\n");
+ output.append(" : ").append(String.valueOf(s2.getStart()))
+ .append(" - ").append(String.valueOf(s2.getEnd()));
+ output.append(" (Sequence length = ")
+ .append(String.valueOf(s2str.length())).append(")")
+ .append(NEWLINE).append(NEWLINE);
+
+ ScoreMatrix pam250 = ScoreModels.getInstance().getPam250();
for (int j = 0; j < nochunks; j++)
{
// Print the first aligned sequence
- output.append(new Format("%" + (maxid) + "s").form(s1id) + " ");
+ output.append(new Format("%" + (maxid) + "s").form(s1id)).append(" ");
for (int i = 0; i < len; i++)
{
}
}
- output.append("\n");
- output.append(new Format("%" + (maxid) + "s").form(" ") + " ");
+ output.append(NEWLINE);
+ output.append(new Format("%" + (maxid) + "s").form(" ")).append(" ");
- // Print out the matching chars
+ /*
+ * Print out the match symbols:
+ * | for exact match (ignoring case)
+ * . if PAM250 score is positive
+ * else a space
+ */
for (int i = 0; i < len; i++)
{
if ((i + (j * len)) < astr1.length())
{
- if (astr1.charAt(i + (j * len)) == astr2.charAt(i + (j * len))
- && !jalview.util.Comparison.isGap(astr1.charAt(i
- + (j * len))))
+ char c1 = astr1.charAt(i + (j * len));
+ char c2 = astr2.charAt(i + (j * len));
+ boolean sameChar = Comparison.isSameResidue(c1, c2, false);
+ if (sameChar && !Comparison.isGap(c1))
{
pid++;
output.append("|");
}
else if (type.equals("pep"))
{
- if (ResidueProperties.getPAM250(astr1.charAt(i + (j * len)),
- astr2.charAt(i + (j * len))) > 0)
+ if (pam250.getPairwiseScore(c1, c2) > 0)
{
output.append(".");
}
}
// Now print the second aligned sequence
- output = output.append("\n");
- output = output.append(new Format("%" + (maxid) + "s").form(s2id)
- + " ");
+ output = output.append(NEWLINE);
+ output = output.append(new Format("%" + (maxid) + "s").form(s2id))
+ .append(" ");
for (int i = 0; i < len; i++)
{
}
}
- output = output.append("\n\n");
+ output.append(NEWLINE).append(NEWLINE);
}
pid = pid / (aseq1.length - count) * 100;
- output = output.append(new Format("Percentage ID = %2.2f\n\n")
- .form(pid));
-
+ output = output.append(new Format("Percentage ID = %2.2f\n").form(pid));
try
{
os.print(output.toString());
/**
* DOCUMENT ME!
*
- * @param mat
- * DOCUMENT ME!
- */
- public void printScoreMatrix(int[][] mat)
- {
- int n = seq1.length;
- int m = seq2.length;
-
- for (int i = 0; i < n; i++)
- {
- // Print the top sequence
- if (i == 0)
- {
- Format.print(System.out, "%8s", s2str.substring(0, 1));
-
- for (int jj = 1; jj < m; jj++)
- {
- Format.print(System.out, "%5s", s2str.substring(jj, jj + 1));
- }
-
- System.out.println();
- }
-
- for (int j = 0; j < m; j++)
- {
- if (j == 0)
- {
- Format.print(System.out, "%3s", s1str.substring(i, i + 1));
- }
-
- Format.print(System.out, "%3d ", mat[i][j] / 10);
- }
-
- System.out.println();
- }
- }
-
- /**
- * DOCUMENT ME!
- *
* @param i
* DOCUMENT ME!
* @param j
public int findTrace(int i, int j)
{
int t = 0;
- int max = score[i - 1][j - 1] + (lookup[seq1[i]][seq2[j]] * 10);
+ float max = score[i - 1][j - 1] + (lookup[seq1[i]][seq2[j]] * 10);
if (F[i][j] > max)
{
}
/**
- * DOCUMENT ME!
+ * Returns the given sequence with all of the given gap characters removed.
*
- * @param gapChar
- * DOCUMENT ME!
+ * @param gapChars
+ * a string of characters to be treated as gaps
* @param seq
- * DOCUMENT ME!
+ * the input sequence
*
- * @return DOCUMENT ME!
+ * @return
*/
- public static String extractGaps(String gapChar, String seq)
+ public static String extractGaps(String gapChars, String seq)
{
- StringTokenizer str = new StringTokenizer(seq, gapChar);
- StringBuffer newString = new StringBuffer();
+ if (gapChars == null || seq == null)
+ {
+ return null;
+ }
+ StringTokenizer str = new StringTokenizer(seq, gapChars);
+ StringBuilder newString = new StringBuilder(seq.length());
while (str.hasMoreTokens())
{
/**
* DOCUMENT ME!
*
- * @param i1
+ * @param f1
* DOCUMENT ME!
- * @param i2
+ * @param f2
* DOCUMENT ME!
- * @param i3
+ * @param f3
* DOCUMENT ME!
*
* @return DOCUMENT ME!
*/
- public int max(int i1, int i2, int i3)
+ private static float max(float f1, float f2, float f3)
{
- int max = i1;
+ float max = f1;
- if (i2 > i1)
+ if (f2 > f1)
{
- max = i2;
+ max = f2;
}
- if (i3 > max)
+ if (f3 > max)
{
- max = i3;
+ max = f3;
}
return max;
/**
* DOCUMENT ME!
*
- * @param i1
+ * @param f1
* DOCUMENT ME!
- * @param i2
+ * @param f2
* DOCUMENT ME!
*
* @return DOCUMENT ME!
*/
- public int max(int i1, int i2)
+ private static float max(float f1, float f2)
{
- int max = i1;
+ float max = f1;
- if (i2 > i1)
+ if (f2 > f1)
{
- max = i2;
+ max = f2;
}
return max;
}
/**
- * DOCUMENT ME!
+ * Converts the character string to an array of integers which are the
+ * corresponding indices to the characters in the score matrix
*
* @param s
- * DOCUMENT ME!
- * @param type
- * DOCUMENT ME!
*
- * @return DOCUMENT ME!
+ * @return
*/
- public int[] stringToInt(String s, String type)
+ int[] indexEncode(String s)
{
- int[] seq1 = new int[s.length()];
+ int[] encoded = new int[s.length()];
for (int i = 0; i < s.length(); i++)
{
- // String ss = s.substring(i, i + 1).toUpperCase();
char c = s.charAt(i);
- if ('a' <= c && c <= 'z')
- {
- // TO UPPERCASE !!!
- c -= ('a' - 'A');
- }
-
- try
- {
- seq1[i] = charToInt[c]; // set accordingly from setType
- if (seq1[i] < 0 || seq1[i] > defInt) // set from setType: 23 for
- // peptides, or 4 for NA.
- {
- seq1[i] = defInt;
- }
-
- } catch (Exception e)
- {
- seq1[i] = defInt;
- }
+ encoded[i] = scoreModel.getMatrixIndex(c);
}
- return seq1;
+ return encoded;
}
/**
public static void displayMatrix(Graphics g, int[][] mat, int n, int m,
int psize)
{
+ // TODO method dosen't seem to be referenced anywhere delete??
int max = -1000;
int min = 1000;
if (allowmismatch || c1 == c2)
{
- lastmatch = true;
- // extend mapping interval.
+ // extend mapping interval
if (lp1 + 1 != alignpos || lp2 + 1 != pdbpos)
{
as1.add(Integer.valueOf(alignpos));
as2.add(Integer.valueOf(pdbpos));
}
+ lastmatch = true;
lp1 = alignpos;
lp2 = pdbpos;
}
else
{
+ // extend mapping interval
+ if (lastmatch)
+ {
+ as1.add(Integer.valueOf(lp1));
+ as2.add(Integer.valueOf(lp2));
+ }
lastmatch = false;
}
}
// construct range pairs
+
int[] mapseq1 = new int[as1.size() + (lastmatch ? 1 : 0)], mapseq2 = new int[as2
.size() + (lastmatch ? 1 : 0)];
int i = 0;
* @param ochains
* @param al
* @param dnaOrProtein
- * @param removeOldAnnots when true, old annotation is cleared before new annotation transferred
+ * @param removeOldAnnots
+ * when true, old annotation is cleared before new annotation
+ * transferred
+ * @return List<List<SequenceI> originals, List<SequenceI> replacement,
+ * List<AlignSeq> alignment between each>
*/
- public static void replaceMatchingSeqsWith(List<SequenceI> seqs, List<AlignmentAnnotation> annotations, List<SequenceI> ochains,
- AlignmentI al, String dnaOrProtein, boolean removeOldAnnots)
+ public static List<List<? extends Object>> replaceMatchingSeqsWith(
+ List<SequenceI> seqs, List<AlignmentAnnotation> annotations,
+ List<SequenceI> ochains, AlignmentI al, String dnaOrProtein,
+ boolean removeOldAnnots)
{
+ List<SequenceI> orig = new ArrayList<SequenceI>(), repl = new ArrayList<SequenceI>();
+ List<AlignSeq> aligs = new ArrayList<AlignSeq>();
if (al != null && al.getHeight() > 0)
{
ArrayList<SequenceI> matches = new ArrayList<SequenceI>();
ArrayList<AlignSeq> aligns = new ArrayList<AlignSeq>();
-
+
for (SequenceI sq : ochains)
{
SequenceI bestm = null;
AlignSeq bestaseq = null;
- int bestscore = 0;
+ float bestscore = 0;
for (SequenceI msq : al.getSequences())
{
- AlignSeq aseq = doGlobalNWAlignment(msq, sq,
- dnaOrProtein);
+ AlignSeq aseq = doGlobalNWAlignment(msq, sq, dnaOrProtein);
if (bestm == null || aseq.getMaxScore() > bestscore)
{
bestscore = aseq.getMaxScore();
if ((q = ochains.indexOf(sp)) > -1)
{
seqs.set(p, sq = matches.get(q));
+ orig.add(sp);
+ repl.add(sq);
sq.setName(sp.getName());
sq.setDescription(sp.getDescription());
Mapping sp2sq;
- sq.transferAnnotation(sp, sp2sq = aligns.get(q).getMappingFromS1(false));
+ sq.transferAnnotation(sp,
+ sp2sq = aligns.get(q).getMappingFromS1(false));
+ aligs.add(aligns.get(q));
int inspos = -1;
for (int ap = 0; ap < annotations.size();)
{
{
inspos = ap;
}
- if (removeOldAnnots) {
+ if (removeOldAnnots)
+ {
annotations.remove(ap);
- } else {
+ }
+ else
+ {
AlignmentAnnotation alan = annotations.remove(ap);
alan.liftOver(sq, sp2sq);
alan.setSequenceRef(sq);
ap++;
}
}
- if (sq.getAnnotation() != null)
+ if (sq.getAnnotation() != null && sq.getAnnotation().length > 0)
{
- annotations.addAll(inspos, Arrays.asList(sq.getAnnotation()));
+ annotations.addAll(inspos == -1 ? annotations.size() : inspos,
+ Arrays.asList(sq.getAnnotation()));
}
}
}
}
+ return Arrays.asList(orig, repl, aligs);
}
/**
// long start = System.currentTimeMillis();
+ SimilarityParams pidParams = new SimilarityParams(true, true, true,
+ true);
float pid;
String seqi, seqj;
for (int i = 0; i < height; i++)
seqj = ug;
}
}
- pid = Comparison.PID(seqi, seqj);
+ pid = (float) PIDModel.computePID(seqi, seqj, pidParams);
// use real sequence length rather than string length
if (lngth[j] < lngth[i])
git://source.jalview.org / jalview.git / blobdiff