*/
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.util.Comparison;
import jalview.util.Format;
import jalview.util.MapList;
int gapExtend = 20;
- float[][] lookup = ScoreModels.getInstance().getBlosum62().getMatrix();
-
- int defInt = 23;
-
StringBuffer output = new StringBuffer();
- String type;
+ String type; // AlignSeq.PEP or AlignSeq.DNA
+
+ private ScoreMatrix scoreMatrix;
- private int[] charToInt;
+ private static final int GAP_INDEX = -1;
/**
* Creates a new AlignSeq object.
*
* @param s1
- * DOCUMENT ME!
+ * first sequence for alignment
* @param s2
- * DOCUMENT ME!
+ * second sequence for alignment
* @param type
- * DOCUMENT ME!
+ * molecule type, either AlignSeq.PEP or AlignSeq.DNA
*/
public AlignSeq(SequenceI s1, SequenceI s2, String type)
{
SequenceI alSeq1 = new Sequence(s1.getName(), getAStr1());
alSeq1.setStart(s1.getStart() + getSeq1Start() - 1);
alSeq1.setEnd(s1.getStart() + getSeq1End() - 1);
- alSeq1.setDatasetSequence(s1.getDatasetSequence() == null ? s1 : s1
- .getDatasetSequence());
+ alSeq1.setDatasetSequence(
+ s1.getDatasetSequence() == null ? s1 : s1.getDatasetSequence());
return alSeq1;
}
SequenceI alSeq2 = new Sequence(s2.getName(), getAStr2());
alSeq2.setStart(s2.getStart() + getSeq2Start() - 1);
alSeq2.setEnd(s2.getStart() + getSeq2End() - 1);
- alSeq2.setDatasetSequence(s2.getDatasetSequence() == null ? s2 : s2
- .getDatasetSequence());
+ alSeq2.setDatasetSequence(
+ s2.getDatasetSequence() == null ? s2 : s2.getDatasetSequence());
return alSeq2;
}
if (s1str.length() == 0 || s2str.length() == 0)
{
- output.append("ALL GAPS: "
- + (s1str.length() == 0 ? s1.getName() : " ")
- + (s2str.length() == 0 ? s2.getName() : ""));
+ output.append(
+ "ALL GAPS: " + (s1str.length() == 0 ? s1.getName() : " ")
+ + (s2str.length() == 0 ? s2.getName() : ""));
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 float[s1str.length()][s2str.length()];
- // System.out.println("score " + rt.freeMemory() + " " + rt.totalMemory());
E = new float[s1str.length()][s2str.length()];
- // System.out.println("E " + rt.freeMemory() + " " + rt.totalMemory());
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 = ScoreModels.getInstance().getDefaultModel(true).getMatrix();
- }
- else if (type.equals(AlignSeq.DNA))
- {
- lookup = ScoreModels.getInstance().getDefaultModel(false).getMatrix();
- }
+ seq2 = indexEncode(s2str);
}
- private void setType(String type2)
+ private void setDefaultParams(String moleculeType)
{
- this.type = type2;
- if (type.equals(AlignSeq.PEP))
- {
- charToInt = ResidueProperties.aaIndex;
- defInt = ResidueProperties.maxProteinIndex;
- }
- else if (type.equals(AlignSeq.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;
+ scoreMatrix = ScoreModels.getInstance()
+ .getDefaultModel(PEP.equals(type));
}
/**
public void traceAlignment()
{
// Find the maximum score along the rhs or bottom row
- float 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] = GAP_INDEX;
+ sb1.replace(sb1.length() - 1, sb1.length(), "-");
}
else if (trace == -1)
{
i--;
- aseq2[count] = defInt;
- astr2 = "-" + astr2.substring(1);
+ aseq2[count] = GAP_INDEX;
+ sb2.replace(sb2.length() - 1, sb2.length(), "-");
}
count--;
seq1start = i + 1;
seq2start = j + 1;
- if (aseq1[count] != defInt)
+ if (aseq1[count] != GAP_INDEX)
{
aseq1[count] = seq1[i];
- astr1 = s1str.charAt(i) + astr1;
+ sb1.append(s1str.charAt(i));
}
- if (aseq2[count] != defInt)
+ if (aseq2[count] != GAP_INDEX)
{
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();
}
/**
public int findTrace(int i, int j)
{
int t = 0;
- float max = score[i - 1][j - 1] + (lookup[seq1[i]][seq2[j]] * 10);
+ // 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);
if (F[i][j] > max)
{
int m = seq2.length;
// top left hand element
- score[0][0] = lookup[seq1[0]][seq2[0]] * 10;
+ score[0][0] = scoreMatrix.getPairwiseScore(s1str.charAt(0),
+ s2str.charAt(0)) * 10;
E[0][0] = -gapExtend;
F[0][0] = 0;
E[0][j] = max(score[0][j - 1] - gapOpen, E[0][j - 1] - gapExtend);
F[0][j] = -gapExtend;
- score[0][j] = max(lookup[seq1[0]][seq2[j]] * 10, -gapOpen, -gapExtend);
+ float pairwiseScore = scoreMatrix.getPairwiseScore(s1str.charAt(0),
+ s2str.charAt(j));
+ score[0][j] = max(pairwiseScore * 10, -gapOpen, -gapExtend);
traceback[0][j] = 1;
}
E[i][0] = -gapOpen;
F[i][0] = max(score[i - 1][0] - gapOpen, F[i - 1][0] - gapExtend);
- score[i][0] = max(lookup[seq1[i]][seq2[0]] * 10, E[i][0], F[i][0]);
+ float pairwiseScore = scoreMatrix.getPairwiseScore(s1str.charAt(i),
+ s2str.charAt(0));
+ score[i][0] = max(pairwiseScore * 10, E[i][0], F[i][0]);
traceback[i][0] = -1;
}
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);
- score[i][j] = max(score[i - 1][j - 1]
- + (lookup[seq1[i]][seq2[j]] * 10), E[i][j], F[i][j]);
+ float pairwiseScore = scoreMatrix.getPairwiseScore(s1str.charAt(i),
+ s2str.charAt(j));
+ score[i][j] = max(score[i - 1][j - 1] + (pairwiseScore * 10),
+ E[i][j], F[i][j]);
traceback[i][j] = findTrace(i, j);
}
}
}
/**
- * 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
*/
- 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] = scoreMatrix.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??
+ // TODO method doesn't seem to be referenced anywhere delete??
int max = -1000;
int min = 1000;
*/
public jalview.datamodel.Mapping getMappingFromS1(boolean allowmismatch)
{
- ArrayList<Integer> as1 = new ArrayList<Integer>(), as2 = new ArrayList<Integer>();
+ ArrayList<Integer> as1 = new ArrayList<Integer>(),
+ as2 = new ArrayList<Integer>();
int pdbpos = s2.getStart() + getSeq2Start() - 2;
int alignpos = s1.getStart() + getSeq1Start() - 2;
int lp2 = pdbpos - 3, lp1 = alignpos - 3;
}
// construct range pairs
- int[] mapseq1 = new int[as1.size() + (lastmatch ? 1 : 0)], mapseq2 = new int[as2
- .size() + (lastmatch ? 1 : 0)];
+ int[] mapseq1 = new int[as1.size() + (lastmatch ? 1 : 0)],
+ mapseq2 = new int[as2.size() + (lastmatch ? 1 : 0)];
int i = 0;
for (Integer ip : as1)
{
List<SequenceI> ochains, AlignmentI al, String dnaOrProtein,
boolean removeOldAnnots)
{
- List<SequenceI> orig = new ArrayList<SequenceI>(), repl = new ArrayList<SequenceI>();
+ List<SequenceI> orig = new ArrayList<SequenceI>(),
+ repl = new ArrayList<SequenceI>();
List<AlignSeq> aligs = new ArrayList<AlignSeq>();
if (al != null && al.getHeight() > 0)
{
bestm = msq;
}
}
- System.out.println("Best Score for " + (matches.size() + 1) + " :"
- + bestscore);
+ // System.out.println("Best Score for " + (matches.size() + 1) + " :"
+ // + bestscore);
matches.add(bestm);
aligns.add(bestaseq);
al.deleteSequence(bestm);
// 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