-/* Jalview - a java multiple alignment editor\r
- * Copyright (C) 1998 Michele Clamp\r
- *\r
- * This program is free software; you can redistribute it and/or\r
- * modify it under the terms of the GNU General Public License\r
- * as published by the Free Software Foundation; either version 2\r
- * of the License, or (at your option) any later version.\r
- *\r
- * This program is distributed in the hope that it will be useful,\r
- * but WITHOUT ANY WARRANTY; without even the implied warranty of\r
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\r
- * GNU General Public License for more details.\r
- *\r
- * You should have received a copy of the GNU General Public License\r
- * along with this program; if not, write to the Free Software\r
- * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.\r
- */\r
+/*\r
+* Jalview - A Sequence Alignment Editor and Viewer\r
+* Copyright (C) 2005 AM Waterhouse, J Procter, G Barton, M Clamp, S Searle\r
+*\r
+* This program is free software; you can redistribute it and/or\r
+* modify it under the terms of the GNU General Public License\r
+* as published by the Free Software Foundation; either version 2\r
+* of the License, or (at your option) any later version.\r
+*\r
+* This program is distributed in the hope that it will be useful,\r
+* but WITHOUT ANY WARRANTY; without even the implied warranty of\r
+* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\r
+* GNU General Public License for more details.\r
+*\r
+* You should have received a copy of the GNU General Public License\r
+* along with this program; if not, write to the Free Software\r
+* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA\r
+*/\r
package jalview.analysis;\r
\r
-import jalview.schemes.*;\r
import jalview.datamodel.SequenceI;\r
+\r
+import jalview.schemes.*;\r
+\r
import jalview.util.*;\r
-import jalview.io.*;\r
\r
-import java.util.*;\r
-import java.io.*;\r
import java.awt.*;\r
\r
-public class AlignSeq {\r
+import java.util.*;\r
\r
- int[][] score;\r
- int[][] E;\r
- int[][] F;\r
- int[][] traceback;\r
-\r
- int[] seq1;\r
- int[] seq2;\r
\r
- SequenceI s1;\r
- SequenceI s2;\r
-\r
- String s1str;\r
- String s2str;\r
-\r
- int maxi;\r
- int maxj;\r
-\r
- int[] aseq1;\r
- int[] aseq2;\r
-\r
- String astr1 = "";\r
- String astr2 = "";\r
-\r
- public int seq1start;\r
- public int seq1end;\r
- public int seq2start;\r
- public int seq2end;\r
-\r
- int count;\r
-\r
- public int maxscore;\r
- float pid;\r
- int prev = 0;\r
-\r
- public static java.util.Hashtable dnaHash = new java.util.Hashtable();\r
-\r
- static {\r
- dnaHash.put("C", new Integer(0));\r
- dnaHash.put("T", new Integer(1));\r
- dnaHash.put("A", new Integer(2));\r
- dnaHash.put("G", new Integer(3));\r
- dnaHash.put("-", new Integer(4));\r
- }\r
-\r
- static String dna[] = {"C","T","A","G","-"};\r
- 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","-"};\r
-\r
- int gapOpen = 120;\r
- int gapExtend = 20;\r
-\r
- int lookup[][] = ResidueProperties.getBLOSUM62();\r
- String intToStr[] = pep;\r
- int defInt = 23;\r
-\r
- String output = "";\r
-\r
- String type;\r
- Runtime rt;\r
- public AlignSeq() {}\r
-\r
- public AlignSeq(SequenceI s1, SequenceI s2,String type) {\r
- rt = Runtime.getRuntime();\r
- SeqInit(s1,s2,type);\r
- }\r
-\r
- public int getMaxScore() {\r
- return maxscore;\r
- }\r
-\r
- public int getSeq2Start() {\r
- return seq2start;\r
- }\r
-\r
- public int getSeq2End() {\r
- return seq2end;\r
- }\r
-\r
- public int getSeq1Start() {\r
- return seq1start;\r
- }\r
-\r
- public int getSeq1End() {\r
- return seq1end;\r
- }\r
-\r
- public String getOutput() {\r
- return output;\r
- }\r
-\r
- public String getAStr1() {\r
- return astr1;\r
- }\r
- public String getAStr2() {\r
- return astr2;\r
- }\r
- public int [] getASeq1() {\r
- return aseq1;\r
- }\r
- public int [] getASeq2() {\r
- return aseq2;\r
- }\r
- public SequenceI getS1() {\r
- return s1;\r
- }\r
- public SequenceI getS2() {\r
- return s2;\r
- }\r
-\r
- public void SeqInit(SequenceI s1, SequenceI s2,String type) {\r
- s1str = extractGaps(".",s1.getSequence());\r
- s2str = extractGaps(".",s2.getSequence());\r
- s1str = extractGaps("-",s1str);\r
- s2str = extractGaps("-",s2str);\r
- s1str = extractGaps(" ",s1str);\r
- s2str = extractGaps(" ",s2str);\r
-\r
- this.s1 = s1;\r
- this.s2 = s2;\r
-\r
- this.type = type;\r
-\r
- if (type.equals("pep")) {\r
- lookup = ResidueProperties.getBLOSUM62();\r
- intToStr = pep;\r
- defInt = 23;\r
- } else if (type.equals("dna")) {\r
- lookup = ResidueProperties.getDNA();\r
- intToStr = dna;\r
- defInt = 4;\r
- } else {\r
- output = output + ("Wrong type = dna or pep only");\r
- System.exit(0);\r
+/**\r
+ *\r
+ *\r
+ * @author $author$\r
+ * @version $Revision$\r
+ */\r
+public class AlignSeq\r
+{\r
+ /** DOCUMENT ME!! */\r
+ public static java.util.Hashtable dnaHash = new java.util.Hashtable();\r
+\r
+ static\r
+ {\r
+ dnaHash.put("C", new Integer(0));\r
+ dnaHash.put("T", new Integer(1));\r
+ dnaHash.put("A", new Integer(2));\r
+ dnaHash.put("G", new Integer(3));\r
+ dnaHash.put("-", new Integer(4));\r
}\r
\r
+ static String[] dna = { "C", "T", "A", "G", "-" };\r
+ static String[] pep =\r
+ {\r
+ "A", "R", "N", "D", "C", "Q", "E", "G", "H", "I", "L", "K", "M", "F",\r
+ "P", "S", "T", "W", "Y", "V", "B", "Z", "X", "-"\r
+ };\r
+ int[][] score;\r
+ int[][] E;\r
+ int[][] F;\r
+ int[][] traceback;\r
+ int[] seq1;\r
+ int[] seq2;\r
+ SequenceI s1;\r
+ SequenceI s2;\r
+ String s1str;\r
+ String s2str;\r
+ int maxi;\r
+ int maxj;\r
+ int[] aseq1;\r
+ int[] aseq2;\r
+ public String astr1 = "";\r
+ public String astr2 = "";\r
+\r
+ /** DOCUMENT ME!! */\r
+ public int seq1start;\r
+\r
+ /** DOCUMENT ME!! */\r
+ public int seq1end;\r
+\r
+ /** DOCUMENT ME!! */\r
+ public int seq2start;\r
+\r
+ /** DOCUMENT ME!! */\r
+ public int seq2end;\r
+ int count;\r
+\r
+ /** DOCUMENT ME!! */\r
+ public int maxscore;\r
+ float pid;\r
+ int prev = 0;\r
+ int gapOpen = 120;\r
+ int gapExtend = 20;\r
+ int[][] lookup = ResidueProperties.getBLOSUM62();\r
+ String[] intToStr = pep;\r
+ int defInt = 23;\r
+ StringBuffer output = new StringBuffer();\r
+ String type;\r
+ Runtime rt;\r
+\r
+\r
+ /**\r
+ * Creates a new AlignSeq object.\r
+ *\r
+ * @param s1 DOCUMENT ME!\r
+ * @param s2 DOCUMENT ME!\r
+ * @param type DOCUMENT ME!\r
+ */\r
+ public AlignSeq(SequenceI s1, SequenceI s2, String type)\r
+ {\r
+ rt = Runtime.getRuntime();\r
+ SeqInit(s1, s2, type);\r
+ }\r
\r
- //System.out.println("lookuip " + rt.freeMemory() + " "+ rt.totalMemory());\r
- seq1 = new int[s1str.length()];\r
- //System.out.println("seq1 " + rt.freeMemory() +" " + rt.totalMemory());\r
- seq2 = new int[s2str.length()];\r
- //System.out.println("seq2 " + rt.freeMemory() + " " + rt.totalMemory());\r
- score = new int[s1str.length()][s2str.length()];\r
- //System.out.println("score " + rt.freeMemory() + " " + rt.totalMemory());\r
- E = new int[s1str.length()][s2str.length()];\r
- //System.out.println("E " + rt.freeMemory() + " " + rt.totalMemory());\r
- F = new int[s1str.length()][s2str.length()];\r
- traceback = new int[s1str.length()][s2str.length()];\r
- //System.out.println("F " + rt.freeMemory() + " " + rt.totalMemory());\r
- seq1 = stringToInt(s1str,type);\r
- //System.out.println("seq1 " + rt.freeMemory() + " " + rt.totalMemory());\r
- seq2 = stringToInt(s2str,type);\r
- //System.out.println("Seq2 " + rt.freeMemory() + " " + rt.totalMemory());\r
-\r
- // long tstart = System.currentTimeMillis();\r
- // calcScoreMatrix();\r
- //long tend = System.currentTimeMillis();\r
-\r
- //System.out.println("Time take to calculate score matrix = " + (tend-tstart) + " ms");\r
-\r
-\r
- // printScoreMatrix(score);\r
- //System.out.println();\r
-\r
- //printScoreMatrix(traceback);\r
- //System.out.println();\r
-\r
- // printScoreMatrix(E);\r
- //System.out.println();\r
-\r
- ///printScoreMatrix(F);\r
- //System.out.println();\r
- // tstart = System.currentTimeMillis();\r
- //traceAlignment();\r
- //tend = System.currentTimeMillis();\r
- //System.out.println("Time take to traceback alignment = " + (tend-tstart) + " ms");\r
- }\r
-\r
- public void traceAlignment() {\r
-\r
- // Find the maximum score along the rhs or bottom row\r
- int max = -9999;\r
- for (int i = 0; i < seq1.length; i++) {\r
- if (score[i][seq2.length - 1] > max ) {\r
- max = score[i][seq2.length - 1];\r
- maxi = i;\r
- maxj = seq2.length-1;\r
- }\r
+ /**\r
+ * DOCUMENT ME!\r
+ *\r
+ * @return DOCUMENT ME!\r
+ */\r
+ public int getMaxScore()\r
+ {\r
+ return maxscore;\r
}\r
- for (int j = 0; j < seq2.length; j++) {\r
- if (score[seq1.length - 1][j] > max ) {\r
- max = score[seq1.length - 1][j];\r
- maxi = seq1.length-1;\r
- maxj = j;\r
- }\r
+\r
+ /**\r
+ * DOCUMENT ME!\r
+ *\r
+ * @return DOCUMENT ME!\r
+ */\r
+ public int getSeq2Start()\r
+ {\r
+ return seq2start;\r
}\r
\r
- // System.out.println(maxi + " " + maxj + " " + score[maxi][maxj]);\r
-\r
- int i = maxi;\r
- int j = maxj;\r
- int trace;\r
- maxscore = score[i][j] / 10;\r
-\r
- seq1end = maxi+1;\r
- seq2end = maxj+1;\r
-\r
- aseq1 = new int[seq1.length + seq2.length];\r
- aseq2 = new int[seq1.length + seq2.length];\r
-\r
- count = seq1.length + seq2.length - 1;\r
-\r
- while (i>0 && j >0) {\r
-\r
- if (aseq1[count] != defInt && i >=0) {\r
- aseq1[count] = seq1[i];\r
- astr1 = intToStr[seq1[i]] + astr1;\r
- }\r
-\r
- if (aseq2[count] != defInt && j > 0) {\r
- aseq2[count] = seq2[j];\r
- astr2 = intToStr[seq2[j]] + astr2;\r
- }\r
- trace = findTrace(i,j);\r
- if (trace == 0) {\r
- i--;\r
- j--;\r
-\r
- } else if (trace == 1) {\r
- j--;\r
- aseq1[count] = defInt;\r
- astr1 = "-" + astr1.substring(1);\r
- } else if (trace == -1) {\r
- i--;\r
- aseq2[count] = defInt;\r
- astr2 = "-" + astr2.substring(1);\r
- }\r
- count--;\r
+ /**\r
+ * DOCUMENT ME!\r
+ *\r
+ * @return DOCUMENT ME!\r
+ */\r
+ public int getSeq2End()\r
+ {\r
+ return seq2end;\r
}\r
\r
- seq1start = i+1;\r
- seq2start = j+1;\r
+ /**\r
+ * DOCUMENT ME!\r
+ *\r
+ * @return DOCUMENT ME!\r
+ */\r
+ public int getSeq1Start()\r
+ {\r
+ return seq1start;\r
+ }\r
\r
- if (aseq1[count] != defInt) {\r
- aseq1[count] = seq1[i];\r
- astr1 = intToStr[seq1[i]] + astr1;\r
+ /**\r
+ * DOCUMENT ME!\r
+ *\r
+ * @return DOCUMENT ME!\r
+ */\r
+ public int getSeq1End()\r
+ {\r
+ return seq1end;\r
}\r
\r
- if (aseq2[count] != defInt) {\r
- aseq2[count] = seq2[j];\r
- astr2 = intToStr[seq2[j]] + astr2;\r
+ /**\r
+ * DOCUMENT ME!\r
+ *\r
+ * @return DOCUMENT ME!\r
+ */\r
+ public String getOutput()\r
+ {\r
+ return output.toString();\r
}\r
- }\r
\r
- public void printAlignment() {\r
- // Find the biggest id length for formatting purposes\r
- int maxid = s1.getName().length();\r
+ /**\r
+ * DOCUMENT ME!\r
+ *\r
+ * @return DOCUMENT ME!\r
+ */\r
+ public String getAStr1()\r
+ {\r
+ return astr1;\r
+ }\r
\r
- if (s2.getName().length() > maxid) {\r
- maxid = s2.getName().length();\r
+ /**\r
+ * DOCUMENT ME!\r
+ *\r
+ * @return DOCUMENT ME!\r
+ */\r
+ public String getAStr2()\r
+ {\r
+ return astr2;\r
}\r
\r
- int len = 72 - maxid - 1;\r
- int nochunks = ((aseq1.length - count) / len) + 1;\r
- pid = 0;\r
- int overlap = 0;\r
-\r
- output = output + ("Score = " + score[maxi][maxj] + "\n");\r
- output = output + ("Length of alignment = " + (aseq1.length-count) + "\n");\r
- output = output + ("Sequence ");\r
- output = output + (new Format("%" + maxid + "s").form(s1.getName()));\r
- output = output + (" : " + seq1start + " - " + seq1end + " (Sequence length = " + s1str.length() + ")\n");\r
- output = output + ("Sequence ");\r
- output = output + (new Format("%" + maxid + "s").form(s2.getName()));\r
- output = output + (" : " + seq2start + " - " + seq2end + " (Sequence length = " + s2str.length() + ")\n\n");\r
-\r
- for (int j = 0; j < nochunks; j++) {\r
- // Print the first aligned sequence\r
- output = output + (new Format("%" + (maxid) + "s").form(s1.getName()) + " ");\r
- for (int i = 0; i < len ; i++) {\r
-\r
- if ((count + i + j*len) < aseq1.length) {\r
- output = output + (new Format("%s").form(intToStr[aseq1[count + i + j*len]]));\r
- }\r
- }\r
-\r
- output = output + ("\n");\r
- output = output + (new Format("%" + (maxid) + "s").form(" ") + " ");\r
- // Print out the matching chars\r
- for (int i = 0; i < len ; i++) {\r
-\r
- if ((count + i + j*len) < aseq1.length) {\r
- if ( intToStr[aseq1[count+i+j*len]].equals(intToStr[aseq2[count+i+j*len]]) && !intToStr[aseq1[count+i+j*len]].equals("-")) {\r
- pid++;\r
- output = output + ("|");\r
- } else if (type.equals("pep")) {\r
- if (ResidueProperties.getPAM250(intToStr[aseq1[count+i+j*len]],intToStr[aseq2[count+i+j*len]]) > 0) {\r
- output = output + (".");\r
- } else {\r
- output = output + (" ");\r
- }\r
- } else {\r
- output = output + (" ");\r
- }\r
+ /**\r
+ * DOCUMENT ME!\r
+ *\r
+ * @return DOCUMENT ME!\r
+ */\r
+ public int[] getASeq1()\r
+ {\r
+ return aseq1;\r
+ }\r
\r
- }\r
- }\r
- // Now print the second aligned sequence\r
- output = output + ("\n");\r
- output = output + (new Format("%" + (maxid) + "s").form(s2.getName()) + " " );\r
- for (int i = 0; i < len ; i++) {\r
- if ((count + i + j*len) < aseq1.length) {\r
- output = output + (new Format("%s").form(intToStr[aseq2[count + i + j*len]]));\r
- }\r
- }\r
- output = output + ("\n\n");\r
+ /**\r
+ * DOCUMENT ME!\r
+ *\r
+ * @return DOCUMENT ME!\r
+ */\r
+ public int[] getASeq2()\r
+ {\r
+ return aseq2;\r
}\r
- pid = pid/(float)(aseq1.length-count)*100;\r
- output = output + (new Format("Percentage ID = %2.2f\n\n").form(pid));\r
\r
- }\r
+ /**\r
+ * DOCUMENT ME!\r
+ *\r
+ * @return DOCUMENT ME!\r
+ */\r
+ public SequenceI getS1()\r
+ {\r
+ return s1;\r
+ }\r
\r
- public void printScoreMatrix(int[][] mat) {\r
- int n = seq1.length;\r
- int m = seq2.length;\r
+ /**\r
+ * DOCUMENT ME!\r
+ *\r
+ * @return DOCUMENT ME!\r
+ */\r
+ public SequenceI getS2()\r
+ {\r
+ return s2;\r
+ }\r
\r
- for (int i = 0; i < n;i++) {\r
- // Print the top sequence\r
- if (i == 0) {\r
- Format.print(System.out,"%8s",s2str.substring(0,1));\r
- for (int jj = 1;jj < m; jj++) {\r
- Format.print(System.out,"%5s",s2str.substring(jj,jj+1));\r
+ /**\r
+ * DOCUMENT ME!\r
+ *\r
+ * @param s1 DOCUMENT ME!\r
+ * @param s2 DOCUMENT ME!\r
+ * @param type DOCUMENT ME!\r
+ */\r
+ public void SeqInit(SequenceI s1, SequenceI s2, String type)\r
+ {\r
+ s1str = extractGaps(jalview.util.Comparison.GapChars, s1.getSequence());\r
+ s2str = extractGaps(jalview.util.Comparison.GapChars, s2.getSequence());\r
+\r
+ this.s1 = s1;\r
+ this.s2 = s2;\r
+\r
+ this.type = type;\r
+\r
+ if (type.equals("pep"))\r
+ {\r
+ lookup = ResidueProperties.getBLOSUM62();\r
+ intToStr = pep;\r
+ defInt = 23;\r
}\r
- System.out.println();\r
- }\r
-\r
- for (int j = 0;j < m; j++) {\r
- if (j == 0) {\r
- Format.print(System.out,"%3s",s1str.substring(i,i+1));\r
+ else if (type.equals("dna"))\r
+ {\r
+ lookup = ResidueProperties.getDNA();\r
+ intToStr = dna;\r
+ defInt = 4;\r
}\r
- Format.print(System.out,"%3d ",mat[i][j]/10);\r
- }\r
- System.out.println();\r
- }\r
- }\r
-\r
- public int findTrace(int i,int j) {\r
- int t = 0;\r
- int max = score[i-1][j-1] + lookup[seq1[i]][seq2[j]] * 10;\r
-\r
- if (F[i][j] > max) {\r
- max = F[i][j];\r
- t = -1;\r
- } else if (F[i][j] == max) {\r
- if (prev == -1) {\r
- max = F[i][j];\r
- t = -1;\r
- }\r
- }\r
- if (E[i][j] >= max) {\r
- max = E[i][j];\r
- t = 1;\r
- } else if (E[i][j] == max) {\r
- if (prev == 1) {\r
- max = E[i][j];\r
- t = 1;\r
- }\r
+ else\r
+ {\r
+ output.append("Wrong type = dna or pep only");\r
+ System.exit(0);\r
+ }\r
+\r
+ //System.out.println("lookuip " + rt.freeMemory() + " "+ rt.totalMemory());\r
+ seq1 = new int[s1str.length()];\r
+\r
+ //System.out.println("seq1 " + rt.freeMemory() +" " + rt.totalMemory());\r
+ seq2 = new int[s2str.length()];\r
+\r
+ //System.out.println("seq2 " + rt.freeMemory() + " " + rt.totalMemory());\r
+ score = new int[s1str.length()][s2str.length()];\r
+\r
+ //System.out.println("score " + rt.freeMemory() + " " + rt.totalMemory());\r
+ E = new int[s1str.length()][s2str.length()];\r
+\r
+ //System.out.println("E " + rt.freeMemory() + " " + rt.totalMemory());\r
+ F = new int[s1str.length()][s2str.length()];\r
+ traceback = new int[s1str.length()][s2str.length()];\r
+\r
+ //System.out.println("F " + rt.freeMemory() + " " + rt.totalMemory());\r
+ seq1 = stringToInt(s1str, type);\r
+\r
+ //System.out.println("seq1 " + rt.freeMemory() + " " + rt.totalMemory());\r
+ seq2 = stringToInt(s2str, type);\r
+\r
+ //System.out.println("Seq2 " + rt.freeMemory() + " " + rt.totalMemory());\r
+ // long tstart = System.currentTimeMillis();\r
+ // calcScoreMatrix();\r
+ //long tend = System.currentTimeMillis();\r
+ //System.out.println("Time take to calculate score matrix = " + (tend-tstart) + " ms");\r
+ // printScoreMatrix(score);\r
+ //System.out.println();\r
+ //printScoreMatrix(traceback);\r
+ //System.out.println();\r
+ // printScoreMatrix(E);\r
+ //System.out.println();\r
+ ///printScoreMatrix(F);\r
+ //System.out.println();\r
+ // tstart = System.currentTimeMillis();\r
+ //traceAlignment();\r
+ //tend = System.currentTimeMillis();\r
+ //System.out.println("Time take to traceback alignment = " + (tend-tstart) + " ms");\r
}\r
- prev = t;\r
- return t;\r
- }\r
\r
- public void calcScoreMatrix() {\r
+ /**\r
+ * DOCUMENT ME!\r
+ */\r
+ public void traceAlignment()\r
+ {\r
+ // Find the maximum score along the rhs or bottom row\r
+ int max = -9999;\r
+\r
+ for (int i = 0; i < seq1.length; i++)\r
+ {\r
+ if (score[i][seq2.length - 1] > max)\r
+ {\r
+ max = score[i][seq2.length - 1];\r
+ maxi = i;\r
+ maxj = seq2.length - 1;\r
+ }\r
+ }\r
\r
+ for (int j = 0; j < seq2.length; j++)\r
+ {\r
+ if (score[seq1.length - 1][j] > max)\r
+ {\r
+ max = score[seq1.length - 1][j];\r
+ maxi = seq1.length - 1;\r
+ maxj = j;\r
+ }\r
+ }\r
\r
- int n = seq1.length;\r
- int m = seq2.length;\r
+ // System.out.println(maxi + " " + maxj + " " + score[maxi][maxj]);\r
+ int i = maxi;\r
+ int j = maxj;\r
+ int trace;\r
+ maxscore = score[i][j] / 10;\r
\r
+ seq1end = maxi + 1;\r
+ seq2end = maxj + 1;\r
\r
- // top left hand element\r
- score[0][0] = lookup[seq1[0]][seq2[0]] * 10;\r
- E[0][0] = -gapExtend;\r
- F[0][0] = 0;\r
+ aseq1 = new int[seq1.length + seq2.length];\r
+ aseq2 = new int[seq1.length + seq2.length];\r
\r
- // Calculate the top row first\r
- for (int j=1; j < m; j++) {\r
- // What should these values be? 0 maybe\r
- E[0][j] = max(score[0][j-1] - gapOpen,E[0][j-1] - gapExtend);\r
- F[0][j] = -gapExtend;\r
+ count = (seq1.length + seq2.length) - 1;\r
\r
- score[0][j] = max( lookup[seq1[0]][seq2[j]] * 10 ,-gapOpen,-gapExtend);\r
+ while ((i > 0) && (j > 0))\r
+ {\r
+ if ((aseq1[count] != defInt) && (i >= 0))\r
+ {\r
+ aseq1[count] = seq1[i];\r
+ astr1 = intToStr[seq1[i]] + astr1;\r
+ }\r
\r
- traceback[0][j] = 1;\r
- }\r
+ if ((aseq2[count] != defInt) && (j > 0))\r
+ {\r
+ aseq2[count] = seq2[j];\r
+ astr2 = intToStr[seq2[j]] + astr2;\r
+ }\r
\r
- // Now do the left hand column\r
- for (int i=1; i < n; i++) {\r
- E[i][0] = -gapOpen;\r
- F[i][0] = max(score[i-1][0]-gapOpen,F[i-1][0]-gapExtend);\r
+ trace = findTrace(i, j);\r
\r
- score[i][0] = max( lookup[seq1[i]][seq2[0]] * 10 ,E[i][0],F[i][0]);\r
- traceback[i][0] = -1;\r
- }\r
+ if (trace == 0)\r
+ {\r
+ i--;\r
+ j--;\r
+ }\r
+ else if (trace == 1)\r
+ {\r
+ j--;\r
+ aseq1[count] = defInt;\r
+ astr1 = "-" + astr1.substring(1);\r
+ }\r
+ else if (trace == -1)\r
+ {\r
+ i--;\r
+ aseq2[count] = defInt;\r
+ astr2 = "-" + astr2.substring(1);\r
+ }\r
\r
- // Now do all the other rows\r
- for (int i = 1; i < n; i++) {\r
- for (int j = 1; j < m; j++) {\r
+ count--;\r
+ }\r
\r
- E[i][j] = max(score[i][j-1] - gapOpen, E[i][j-1] - gapExtend);\r
- F[i][j] = max(score[i-1][j] - gapOpen, F[i-1][j] - gapExtend);\r
+ seq1start = i + 1;\r
+ seq2start = j + 1;\r
\r
- score[i][j] = max(score[i-1][j-1] + lookup[seq1[i]][seq2[j]]*10,\r
- E[i][j],\r
- F[i][j]);\r
- traceback[i][j] = findTrace(i,j);\r
- }\r
- }\r
+ if (aseq1[count] != defInt)\r
+ {\r
+ aseq1[count] = seq1[i];\r
+ astr1 = intToStr[seq1[i]] + astr1;\r
+ }\r
\r
- }\r
- public static String extractChars(String chars, String seq) {\r
- String out = seq;\r
- for (int i=0; i < chars.length(); i++) {\r
- String gap = chars.substring(i,i+1);\r
- out = extractGaps(gap,out);\r
- }\r
- return out;\r
- }\r
- public static String extractGaps(String gapChar, String seq) {\r
- StringTokenizer str = new StringTokenizer(seq,gapChar);\r
- String newString = "";\r
-\r
- while (str.hasMoreTokens()) {\r
- newString = newString + str.nextToken();\r
+ if (aseq2[count] != defInt)\r
+ {\r
+ aseq2[count] = seq2[j];\r
+ astr2 = intToStr[seq2[j]] + astr2;\r
+ }\r
}\r
- return newString;\r
- }\r
\r
+ /**\r
+ * DOCUMENT ME!\r
+ */\r
+ public void printAlignment(java.io.PrintStream os)\r
+ {\r
+ // Find the biggest id length for formatting purposes\r
+ int maxid = s1.getName().length();\r
+\r
+ if (s2.getName().length() > maxid)\r
+ {\r
+ maxid = s2.getName().length();\r
+ }\r
+\r
+ int len = 72 - maxid - 1;\r
+ int nochunks = ((aseq1.length - count) / len) + 1;\r
+ pid = 0;\r
+\r
+ output.append("Score = " + score[maxi][maxj] + "\n");\r
+ output.append("Length of alignment = " + (aseq1.length - count) + "\n");\r
+ output.append("Sequence ");\r
+ output.append(new Format("%" + maxid + "s").form(s1.getName()));\r
+ output.append(" : " + seq1start + " - " + seq1end + " (Sequence length = " +\r
+ s1str.length() + ")\n");\r
+ output .append("Sequence ");\r
+ output.append(new Format("%" + maxid + "s").form(s2.getName()));\r
+ output .append(" : " + seq2start + " - " + seq2end + " (Sequence length = " +\r
+ s2str.length() + ")\n\n");\r
+\r
+ for (int j = 0; j < nochunks; j++)\r
+ {\r
+ // Print the first aligned sequence\r
+ output.append(new Format("%" + (maxid) + "s").form(s1.getName()) + " ");\r
+\r
+ for (int i = 0; i < len; i++)\r
+ {\r
+ if ((count + i + (j * len)) < aseq1.length)\r
+ {\r
+ output.append(new Format("%s").form(intToStr[aseq1[count + i +\r
+ (j * len)]]));\r
+ }\r
+ }\r
+\r
+ output.append("\n");\r
+ output.append(new Format("%" + (maxid) + "s").form(" ") + " ");\r
+\r
+ // Print out the matching chars\r
+ for (int i = 0; i < len; i++)\r
+ {\r
+ if ((count + i + (j * len)) < aseq1.length)\r
+ {\r
+ if (intToStr[aseq1[count + i + (j * len)]].equals(\r
+ intToStr[aseq2[count + i + (j * len)]]) &&\r
+ !intToStr[aseq1[count + i + (j * len)]].equals("-"))\r
+ {\r
+ pid++;\r
+ output.append("|");\r
+ }\r
+ else if (type.equals("pep"))\r
+ {\r
+ if (ResidueProperties.getPAM250(\r
+ intToStr[aseq1[count + i + (j * len)]],\r
+ intToStr[aseq2[count + i + (j * len)]]) > 0)\r
+ {\r
+ output.append(".");\r
+ }\r
+ else\r
+ {\r
+ output.append(" ");\r
+ }\r
+ }\r
+ else\r
+ {\r
+ output.append(" ");\r
+ }\r
+ }\r
+ }\r
+\r
+ // Now print the second aligned sequence\r
+ output = output.append("\n");\r
+ output = output.append(new Format("%" + (maxid) + "s").form(s2.getName()) + " ");\r
+\r
+ for (int i = 0; i < len; i++)\r
+ {\r
+ if ((count + i + (j * len)) < aseq1.length)\r
+ {\r
+ output .append(new Format("%s").form(intToStr[aseq2[count + i +\r
+ (j * len)]]));\r
+ }\r
+ }\r
+\r
+ output = output .append("\n\n");\r
+ }\r
\r
- public int max(int i1, int i2, int i3) {\r
- int max = i1;\r
- if (i2 > i1) {\r
- max = i2;\r
+ pid = pid / (float) (aseq1.length - count) * 100;\r
+ output = output.append(new Format("Percentage ID = %2.2f\n\n").form(pid));\r
+\r
+ try{\r
+ os.println(output.toString());\r
+ }catch(Exception ex){}\r
}\r
- if (i3 > max) {\r
- max = i3;\r
+\r
+ /**\r
+ * DOCUMENT ME!\r
+ *\r
+ * @param mat DOCUMENT ME!\r
+ */\r
+ public void printScoreMatrix(int[][] mat)\r
+ {\r
+ int n = seq1.length;\r
+ int m = seq2.length;\r
+\r
+ for (int i = 0; i < n; i++)\r
+ {\r
+ // Print the top sequence\r
+ if (i == 0)\r
+ {\r
+ Format.print(System.out, "%8s", s2str.substring(0, 1));\r
+\r
+ for (int jj = 1; jj < m; jj++)\r
+ {\r
+ Format.print(System.out, "%5s", s2str.substring(jj, jj + 1));\r
+ }\r
+\r
+ System.out.println();\r
+ }\r
+\r
+ for (int j = 0; j < m; j++)\r
+ {\r
+ if (j == 0)\r
+ {\r
+ Format.print(System.out, "%3s", s1str.substring(i, i + 1));\r
+ }\r
+\r
+ Format.print(System.out, "%3d ", mat[i][j] / 10);\r
+ }\r
+\r
+ System.out.println();\r
+ }\r
}\r
- return max;\r
- }\r
\r
- public int max(int i1, int i2) {\r
- int max = i1;\r
- if (i2 > i1) {\r
- max = i2;\r
+ /**\r
+ * DOCUMENT ME!\r
+ *\r
+ * @param i DOCUMENT ME!\r
+ * @param j DOCUMENT ME!\r
+ *\r
+ * @return DOCUMENT ME!\r
+ */\r
+ public int findTrace(int i, int j)\r
+ {\r
+ int t = 0;\r
+ int max = score[i - 1][j - 1] + (lookup[seq1[i]][seq2[j]] * 10);\r
+\r
+ if (F[i][j] > max)\r
+ {\r
+ max = F[i][j];\r
+ t = -1;\r
+ }\r
+ else if (F[i][j] == max)\r
+ {\r
+ if (prev == -1)\r
+ {\r
+ max = F[i][j];\r
+ t = -1;\r
+ }\r
+ }\r
+\r
+ if (E[i][j] >= max)\r
+ {\r
+ max = E[i][j];\r
+ t = 1;\r
+ }\r
+ else if (E[i][j] == max)\r
+ {\r
+ if (prev == 1)\r
+ {\r
+ max = E[i][j];\r
+ t = 1;\r
+ }\r
+ }\r
+\r
+ prev = t;\r
+\r
+ return t;\r
}\r
- return max;\r
- }\r
-\r
- public int[] stringToInt(String s,String type) {\r
- int[] seq1 = new int[s.length()];\r
-\r
- for (int i = 0;i < s.length(); i++) {\r
- String ss = s.substring(i,i+1).toUpperCase();\r
- try {\r
- if (type.equals("pep")) {\r
- seq1[i] = ((Integer)ResidueProperties.aaHash.get(ss)).intValue();\r
- } else if (type.equals("dna")) {\r
- seq1[i] = ((Integer)dnaHash.get(ss)).intValue();\r
+\r
+ /**\r
+ * DOCUMENT ME!\r
+ */\r
+ public void calcScoreMatrix()\r
+ {\r
+ int n = seq1.length;\r
+ int m = seq2.length;\r
+\r
+ // top left hand element\r
+ score[0][0] = lookup[seq1[0]][seq2[0]] * 10;\r
+ E[0][0] = -gapExtend;\r
+ F[0][0] = 0;\r
+\r
+ // Calculate the top row first\r
+ for (int j = 1; j < m; j++)\r
+ {\r
+ // What should these values be? 0 maybe\r
+ E[0][j] = max(score[0][j - 1] - gapOpen, E[0][j - 1] - gapExtend);\r
+ F[0][j] = -gapExtend;\r
+\r
+ score[0][j] = max(lookup[seq1[0]][seq2[j]] * 10, -gapOpen,\r
+ -gapExtend);\r
+\r
+ traceback[0][j] = 1;\r
}\r
- if (seq1[i] > 23) {\r
- seq1[i] = 23;\r
+\r
+ // Now do the left hand column\r
+ for (int i = 1; i < n; i++)\r
+ {\r
+ E[i][0] = -gapOpen;\r
+ F[i][0] = max(score[i - 1][0] - gapOpen, F[i - 1][0] - gapExtend);\r
+\r
+ score[i][0] = max(lookup[seq1[i]][seq2[0]] * 10, E[i][0], F[i][0]);\r
+ traceback[i][0] = -1;\r
}\r
- } catch (Exception e) {\r
- if (type.equals("dna")) {\r
- seq1[i] = 4;\r
- } else {\r
- seq1[i] = 23;\r
+\r
+ // Now do all the other rows\r
+ for (int i = 1; i < n; i++)\r
+ {\r
+ for (int j = 1; j < m; j++)\r
+ {\r
+ E[i][j] = max(score[i][j - 1] - gapOpen, E[i][j - 1] -\r
+ gapExtend);\r
+ F[i][j] = max(score[i - 1][j] - gapOpen, F[i - 1][j] -\r
+ gapExtend);\r
+\r
+ score[i][j] = max(score[i - 1][j - 1] +\r
+ (lookup[seq1[i]][seq2[j]] * 10), E[i][j], F[i][j]);\r
+ traceback[i][j] = findTrace(i, j);\r
+ }\r
}\r
- }\r
}\r
- return seq1;\r
- }\r
\r
- public static void displayMatrix(Graphics g, int[][] mat, int n, int m,int psize) {\r
\r
- int max = -1000;\r
- int min = 1000;\r
\r
- for (int i=0; i < n; i++) {\r
- for (int j=0; j < m; j++) {\r
- if (mat[i][j] >= max) {\r
- max = mat[i][j];\r
+ /**\r
+ * DOCUMENT ME!\r
+ *\r
+ * @param gapChar DOCUMENT ME!\r
+ * @param seq DOCUMENT ME!\r
+ *\r
+ * @return DOCUMENT ME!\r
+ */\r
+ public static String extractGaps(String gapChar, String seq)\r
+ {\r
+ StringTokenizer str = new StringTokenizer(seq, gapChar);\r
+ StringBuffer newString = new StringBuffer();\r
+\r
+ while (str.hasMoreTokens())\r
+ {\r
+ newString.append( str.nextToken() );\r
}\r
- if (mat[i][j] <= min) {\r
- min = mat[i][j];\r
+\r
+ return newString.toString();\r
+ }\r
+\r
+ /**\r
+ * DOCUMENT ME!\r
+ *\r
+ * @param i1 DOCUMENT ME!\r
+ * @param i2 DOCUMENT ME!\r
+ * @param i3 DOCUMENT ME!\r
+ *\r
+ * @return DOCUMENT ME!\r
+ */\r
+ public int max(int i1, int i2, int i3)\r
+ {\r
+ int max = i1;\r
+\r
+ if (i2 > i1)\r
+ {\r
+ max = i2;\r
+ }\r
+\r
+ if (i3 > max)\r
+ {\r
+ max = i3;\r
}\r
- }\r
+\r
+ return max;\r
}\r
- System.out.println(max + " " + min);\r
- for (int i=0; i < n; i++) {\r
- for (int j=0; j < m; j++) {\r
- int x = psize*i;\r
- int y = psize*j;\r
- // System.out.println(mat[i][j]);\r
- float score = (float)(mat[i][j] - min)/(float)(max-min);\r
- g.setColor(new Color(score,0,0));\r
- g.fillRect(x,y,psize,psize);\r
- // System.out.println(x + " " + y + " " + score);\r
- }\r
\r
+ /**\r
+ * DOCUMENT ME!\r
+ *\r
+ * @param i1 DOCUMENT ME!\r
+ * @param i2 DOCUMENT ME!\r
+ *\r
+ * @return DOCUMENT ME!\r
+ */\r
+ public int max(int i1, int i2)\r
+ {\r
+ int max = i1;\r
+\r
+ if (i2 > i1)\r
+ {\r
+ max = i2;\r
+ }\r
+\r
+ return max;\r
}\r
- }\r
\r
+ /**\r
+ * DOCUMENT ME!\r
+ *\r
+ * @param s DOCUMENT ME!\r
+ * @param type DOCUMENT ME!\r
+ *\r
+ * @return DOCUMENT ME!\r
+ */\r
+ public int[] stringToInt(String s, String type)\r
+ {\r
+ int[] seq1 = new int[s.length()];\r
+\r
+ for (int i = 0; i < s.length(); i++)\r
+ {\r
+ String ss = s.substring(i, i + 1).toUpperCase();\r
+\r
+ try\r
+ {\r
+ if (type.equals("pep"))\r
+ {\r
+ seq1[i] = ((Integer) ResidueProperties.aaHash.get(ss)).intValue();\r
+ }\r
+ else if (type.equals("dna"))\r
+ {\r
+ seq1[i] = ((Integer) dnaHash.get(ss)).intValue();\r
+ }\r
+\r
+ if (seq1[i] > 23)\r
+ {\r
+ seq1[i] = 23;\r
+ }\r
+ }\r
+ catch (Exception e)\r
+ {\r
+ if (type.equals("dna"))\r
+ {\r
+ seq1[i] = 4;\r
+ }\r
+ else\r
+ {\r
+ seq1[i] = 23;\r
+ }\r
+ }\r
+ }\r
+\r
+ return seq1;\r
+ }\r
+\r
+ /**\r
+ * DOCUMENT ME!\r
+ *\r
+ * @param g DOCUMENT ME!\r
+ * @param mat DOCUMENT ME!\r
+ * @param n DOCUMENT ME!\r
+ * @param m DOCUMENT ME!\r
+ * @param psize DOCUMENT ME!\r
+ */\r
+ public static void displayMatrix(Graphics g, int[][] mat, int n, int m,\r
+ int psize)\r
+ {\r
+ int max = -1000;\r
+ int min = 1000;\r
+\r
+ for (int i = 0; i < n; i++)\r
+ {\r
+ for (int j = 0; j < m; j++)\r
+ {\r
+ if (mat[i][j] >= max)\r
+ {\r
+ max = mat[i][j];\r
+ }\r
+\r
+ if (mat[i][j] <= min)\r
+ {\r
+ min = mat[i][j];\r
+ }\r
+ }\r
+ }\r
+\r
+ System.out.println(max + " " + min);\r
+\r
+ for (int i = 0; i < n; i++)\r
+ {\r
+ for (int j = 0; j < m; j++)\r
+ {\r
+ int x = psize * i;\r
+ int y = psize * j;\r
+\r
+ // System.out.println(mat[i][j]);\r
+ float score = (float) (mat[i][j] - min) / (float) (max - min);\r
+ g.setColor(new Color(score, 0, 0));\r
+ g.fillRect(x, y, psize, psize);\r
+\r
+ // System.out.println(x + " " + y + " " + score);\r
+ }\r
+ }\r
+ }\r
}\r
git://source.jalview.org / jalview.git / blobdiff