\r
import jalview.datamodel.*;\r
\r
-import jalview.gui.*;\r
-\r
import java.util.*;\r
\r
\r
-public class Conservation {\r
+/**\r
+ * Calculates conservation values for a given set of sequences\r
+ *\r
+ * @author $author$\r
+ * @version $Revision$\r
+ */\r
+public class Conservation\r
+{\r
Vector sequences;\r
int start;\r
int end;\r
int maxLength = 0; // used by quality calcs\r
boolean seqNumsChanged = false; // updated after any change via calcSeqNum;\r
Vector total = new Vector();\r
+\r
+ /** Stores calculated quality values */\r
public Vector quality;\r
+\r
+ /** Stores maximum and minimum values of quality values */\r
public Double[] qualityRange = new Double[2];\r
String consString = "";\r
Sequence consSequence;\r
String name = "";\r
int[][] cons2;\r
\r
+ /**\r
+ * Creates a new Conservation object.\r
+ *\r
+ * @param name Name of conservation\r
+ * @param propHash DOCUMENT ME!\r
+ * @param threshold to count the residues in residueHash(). commonly used value is 3\r
+ * @param sequences sequences to be used in calculation\r
+ * @param start start residue position\r
+ * @param end end residue position\r
+ */\r
public Conservation(String name, Hashtable propHash, int threshold,\r
- Vector sequences, int start, int end) {\r
+ Vector sequences, int start, int end)\r
+ {\r
this.name = name;\r
this.propHash = propHash;\r
this.threshold = threshold;\r
calcSeqNums();\r
}\r
\r
- private void calcSeqNums() {\r
- for (int i = 0; i < sequences.size(); i++) {\r
+ /**\r
+ * DOCUMENT ME!\r
+ */\r
+ private void calcSeqNums()\r
+ {\r
+ int i=0, iSize=sequences.size();\r
+ for (i=0; i < iSize; i++)\r
+ {\r
calcSeqNum(i);\r
}\r
}\r
\r
- private void calcSeqNum(int i) {\r
+ /**\r
+ * DOCUMENT ME!\r
+ *\r
+ * @param i DOCUMENT ME!\r
+ */\r
+ private void calcSeqNum(int i)\r
+ {\r
String sq = null; // for dumb jbuilder not-inited exception warning\r
int[] sqnum = null;\r
\r
- if ((i > -1) && (i < sequences.size())) {\r
+ if ((i > -1) && (i < sequences.size()))\r
+ {\r
sq = ((SequenceI) sequences.elementAt(i)).getSequence();\r
\r
- if (seqNums.size() <= i) {\r
+ if (seqNums.size() <= i)\r
+ {\r
seqNums.addElement(new int[sq.length() + 1]);\r
}\r
\r
- if (sq.hashCode() != ((int[]) seqNums.elementAt(i))[0]) {\r
+ if (sq.hashCode() != ((int[]) seqNums.elementAt(i))[0])\r
+ {\r
int j;\r
int len;\r
seqNumsChanged = true;\r
sq = ((SequenceI) sequences.elementAt(i)).getSequence();\r
len = sq.length();\r
\r
- if (maxLength < len) {\r
+ if (maxLength < len)\r
+ {\r
maxLength = len;\r
}\r
\r
sqnum = new int[len + 1]; // better to always make a new array - sequence can change its length\r
sqnum[0] = sq.hashCode();\r
\r
- for (j = 1; j <= len; j++) {\r
- sqnum[j] = ((Integer) jalview.schemes.ResidueProperties.aaHash.get(new String(\r
- sq.substring(j - 1, j)))).intValue(); // yuk\r
+ for (j = 1; j <= len; j++)\r
+ {\r
+ sqnum[j] = ((Integer) jalview.schemes.ResidueProperties.aaHash.get(String.valueOf(\r
+ sq.charAt(j - 1)))).intValue(); // yuk - JBPNote - case taken care of in aaHash\r
}\r
\r
seqNums.setElementAt(sqnum, i);\r
}\r
- } else {\r
+ }\r
+ else\r
+ {\r
// JBPNote INFO level debug\r
System.err.println(\r
"ERROR: calcSeqNum called with out of range sequence index for Alignment\n");\r
}\r
}\r
\r
- public void calculate() {\r
- for (int i = start; i <= end; i++) {\r
- Hashtable resultHash = null;\r
- Hashtable residueHash = null;\r
-\r
+ /**\r
+ * Calculates the conservation values for given set of sequences\r
+ */\r
+ public void calculate()\r
+ {\r
+ Hashtable resultHash, residueHash, ht;\r
+ int count, thresh, j, jSize = sequences.size();\r
+ String sequence, res, type;\r
+ Enumeration enumeration, enumeration2;\r
+\r
+ for (int i = start; i <= end; i++)\r
+ {\r
resultHash = new Hashtable();\r
residueHash = new Hashtable();\r
\r
- for (int j = 0; j < sequences.size(); j++) {\r
+ for (j = 0; j < jSize; j++)\r
+ {\r
// JBPNote - have to make sure elements of the sequences vector\r
// are tested like this everywhere...\r
- if (sequences.elementAt(j) instanceof Sequence) {\r
- Sequence s = (Sequence) sequences.elementAt(j);\r
+ if (sequences.elementAt(j) instanceof Sequence)\r
+ {\r
+ sequence = ((Sequence) sequences.elementAt(j)).getSequence();\r
\r
- if (s.getLength() > i) {\r
- String res = s.getSequence().substring(i, i + 1);\r
+ if (sequence.length() > i)\r
+ {\r
+ res = String.valueOf(Character.toUpperCase(sequence.charAt(i)));\r
\r
- if (residueHash.containsKey(res)) {\r
- int count = ((Integer) residueHash.get(res)).intValue();\r
+ if (residueHash.containsKey(res))\r
+ {\r
+ count = ((Integer) residueHash.get(res)).intValue();\r
count++;\r
residueHash.put(res, new Integer(count));\r
- } else {\r
+ }\r
+ else\r
+ {\r
residueHash.put(res, new Integer(1));\r
}\r
- } else {\r
- if (residueHash.containsKey("-")) {\r
- int count = ((Integer) residueHash.get("-")).intValue();\r
+ }\r
+ else\r
+ {\r
+ if (residueHash.containsKey("-"))\r
+ {\r
+ count = ((Integer) residueHash.get("-")).intValue();\r
count++;\r
residueHash.put("-", new Integer(count));\r
- } else {\r
+ }\r
+ else\r
+ {\r
residueHash.put("-", new Integer(1));\r
}\r
}\r
}\r
\r
//What is the count threshold to count the residues in residueHash()\r
- int thresh = (threshold * (sequences.size())) / 100;\r
+ thresh = (threshold * (sequences.size())) / 100;\r
\r
//loop over all the found residues\r
- Enumeration e = residueHash.keys();\r
+ enumeration = residueHash.keys();\r
\r
- while (e.hasMoreElements()) {\r
- String res = (String) e.nextElement();\r
+ while (enumeration.hasMoreElements())\r
+ {\r
+ res = (String) enumeration.nextElement();\r
\r
- if (((Integer) residueHash.get(res)).intValue() > thresh) {\r
+ if (((Integer) residueHash.get(res)).intValue() > thresh)\r
+ {\r
//Now loop over the properties\r
- Enumeration e2 = propHash.keys();\r
+ enumeration2 = propHash.keys();\r
\r
- while (e2.hasMoreElements()) {\r
- String type = (String) e2.nextElement();\r
- Hashtable ht = (Hashtable) propHash.get(type);\r
+ while (enumeration2.hasMoreElements())\r
+ {\r
+ type = (String) enumeration2.nextElement();\r
+ ht = (Hashtable) propHash.get(type);\r
\r
//Have we ticked this before?\r
- if (!resultHash.containsKey(type)) {\r
- if (ht.containsKey(res)) {\r
+ if (!resultHash.containsKey(type))\r
+ {\r
+ if (ht.containsKey(res))\r
+ {\r
resultHash.put(type, ht.get(res));\r
- } else {\r
+ }\r
+ else\r
+ {\r
resultHash.put(type, ht.get("-"));\r
}\r
- } else if (((Integer) resultHash.get(type)).equals(\r
- (Integer) ht.get(res)) == false) {\r
+ }\r
+ else if (((Integer) resultHash.get(type)).equals(\r
+ (Integer) ht.get(res)) == false)\r
+ {\r
resultHash.put(type, new Integer(-1));\r
}\r
}\r
}\r
}\r
\r
- public int countGaps(int j) {\r
- int count = 0;\r
-\r
- for (int i = 0; i < sequences.size(); i++) {\r
- if ((j + 1) > ((Sequence) sequences.elementAt(i)).getSequence()\r
- .length()) {\r
- count++;\r
-\r
- continue;\r
- }\r
-\r
- char c = ((Sequence) sequences.elementAt(i)).getSequence().charAt(j);\r
-\r
- if (jalview.util.Comparison.isGap((c))) {\r
- count++;\r
- }\r
- }\r
-\r
- return count;\r
- }\r
\r
/***\r
* countConsNGaps\r
* returns gap count in int[0], and conserved residue count in int[1]\r
*/\r
- public int[] countConsNGaps(int j) {\r
+ public int[] countConsNGaps(int j)\r
+ {\r
int count = 0;\r
int cons = 0;\r
int nres = 0;\r
int[] r = new int[2];\r
char f = '$';\r
+ int i, iSize = sequences.size();\r
+ char c;\r
\r
- for (int i = 0; i < sequences.size(); i++) {\r
- if (j >= ((Sequence) sequences.elementAt(i)).getSequence().length()) {\r
+ for (i = 0; i < iSize; i++)\r
+ {\r
+ if (j >= ((Sequence) sequences.elementAt(i)).getLength())\r
+ {\r
count++;\r
-\r
continue;\r
}\r
\r
- char c = ((Sequence) sequences.elementAt(i)).getSequence().charAt(j);\r
+ c = ((Sequence) sequences.elementAt(i)).getCharAt(j); // gaps do not have upper/lower case\r
\r
- if (jalview.util.Comparison.isGap((c))) {\r
+ if (jalview.util.Comparison.isGap((c)))\r
+ {\r
count++;\r
- } else {\r
+ }\r
+ else\r
+ {\r
nres++;\r
\r
- if (nres == 1) {\r
+ if (nres == 1)\r
+ {\r
f = c;\r
cons++;\r
- } else if (f == c) {\r
+ }\r
+ else if (f == c)\r
+ {\r
cons++;\r
}\r
}\r
return r;\r
}\r
\r
- public void verdict(boolean consflag, float percentageGaps) {\r
- String consString = "";\r
-\r
- for (int i = start; i <= end; i++) {\r
- int[] gapcons = countConsNGaps(i);\r
- boolean cons = (gapcons[0] == 1) ? true : false;\r
- int totGaps = gapcons[1];\r
- float pgaps = ((float) totGaps * 100) / (float) sequences.size();\r
-\r
- // System.out.println("percentage gaps = "+pgaps+"\n");\r
- if (percentageGaps > pgaps) {\r
- Hashtable resultHash = (Hashtable) total.elementAt(i - start);\r
+ /**\r
+ * Calculates the conservation sequence\r
+ *\r
+ * @param consflag if true, poitiveve conservation; false calculates negative conservation\r
+ * @param percentageGaps commonly used value is 25\r
+ */\r
+ public void verdict(boolean consflag, float percentageGaps)\r
+ {\r
+ StringBuffer consString = new StringBuffer();\r
+ String type;\r
+ Integer result;\r
+ int[] gapcons;\r
+ int totGaps, count;\r
+ float pgaps;\r
+ Hashtable resultHash ;\r
+ Enumeration enumeration;\r
+\r
+\r
+ for (int i = start; i <= end; i++)\r
+ {\r
+ gapcons = countConsNGaps(i);\r
+ totGaps = gapcons[1];\r
+ pgaps = ((float) totGaps * 100) / (float) sequences.size();\r
+\r
+ if (percentageGaps > pgaps)\r
+ {\r
+ resultHash = (Hashtable) total.elementAt(i - start);\r
\r
//Now find the verdict\r
- int count = 0;\r
- Enumeration e3 = resultHash.keys();\r
+ count = 0;\r
+ enumeration = resultHash.keys();\r
\r
- while (e3.hasMoreElements()) {\r
- String type = (String) e3.nextElement();\r
- Integer result = (Integer) resultHash.get(type);\r
+ while (enumeration.hasMoreElements())\r
+ {\r
+ type = (String) enumeration.nextElement();\r
+ result = (Integer) resultHash.get(type);\r
\r
//Do we want to count +ve conservation or +ve and -ve cons.?\r
- if (consflag) {\r
- if (result.intValue() == 1) {\r
+ if (consflag)\r
+ {\r
+ if (result.intValue() == 1)\r
+ {\r
count++;\r
}\r
- } else {\r
- if (result.intValue() != -1) {\r
+ }\r
+ else\r
+ {\r
+ if (result.intValue() != -1)\r
+ {\r
count++;\r
}\r
}\r
}\r
\r
- if (count < 10) {\r
- consString = consString + String.valueOf(count); // Conserved props!=Identity\r
- } else {\r
- consString = consString + ((gapcons[0] == 1) ? "*" : "+");\r
+ if (count < 10)\r
+ {\r
+ consString.append(count); // Conserved props!=Identity\r
+ }\r
+ else\r
+ {\r
+ consString.append((gapcons[0] == 1) ? "*" : "+");\r
}\r
- } else {\r
- consString = consString + "-";\r
+ }\r
+ else\r
+ {\r
+ consString.append("-");\r
}\r
}\r
\r
- consSequence = new Sequence(name, consString, start, end);\r
+ consSequence = new Sequence(name, consString.toString(), start, end);\r
}\r
\r
- public Sequence getConsSequence() {\r
+ /**\r
+ *\r
+ *\r
+ * @return Conservation sequence\r
+ */\r
+ public Sequence getConsSequence()\r
+ {\r
return consSequence;\r
}\r
\r
// From Alignment.java in jalview118\r
- public void findQuality() {\r
+ public void findQuality()\r
+ {\r
findQuality(0, maxLength - 1);\r
}\r
\r
- private void percentIdentity2() {\r
+ /**\r
+ * DOCUMENT ME!\r
+ */\r
+ private void percentIdentity2()\r
+ {\r
calcSeqNums(); // updates maxLength, too.\r
\r
- if ((cons2 == null) || seqNumsChanged) {\r
+ if ((cons2 == null) || seqNumsChanged)\r
+ {\r
cons2 = new int[maxLength][24];\r
\r
// Initialize the array\r
- for (int j = 0; j < 24; j++) {\r
- for (int i = 0; i < maxLength; i++) {\r
+ for (int j = 0; j < 24; j++)\r
+ {\r
+ for (int i = 0; i < maxLength; i++)\r
+ {\r
cons2[i][j] = 0;\r
}\r
}\r
int[] sqnum;\r
int j = 0;\r
\r
- while (j < sequences.size()) {\r
+ while (j < sequences.size())\r
+ {\r
sqnum = (int[]) seqNums.elementAt(j);\r
\r
- for (int i = 1; i < sqnum.length; i++) {\r
+ for (int i = 1; i < sqnum.length; i++)\r
+ {\r
cons2[i - 1][sqnum[i]]++;\r
}\r
\r
- for (int i = sqnum.length - 1; i < maxLength; i++) {\r
+ for (int i = sqnum.length - 1; i < maxLength; i++)\r
+ {\r
cons2[i][23]++; // gap count\r
}\r
\r
}\r
}\r
\r
- public void findQuality(int start, int end) {\r
+ /**\r
+ * Calculates the quality of the set of sequences\r
+ *\r
+ * @param start Start residue\r
+ * @param end End residue\r
+ */\r
+ public void findQuality(int start, int end)\r
+ {\r
quality = new Vector();\r
\r
double max = -10000;\r
- String s = "";\r
int[][] BLOSUM62 = jalview.schemes.ResidueProperties.getBLOSUM62();\r
\r
//Loop over columns // JBPNote Profiling info\r
\r
int size = seqNums.size();\r
int[] lengths = new int[size];\r
+ double tot, bigtot, sr, tmp;\r
+ double [] x, xx;\r
+ int l, j, i, ii, seqNum;\r
\r
- for (int l = 0; l < size; l++)\r
+ for (l = 0; l < size; l++)\r
lengths[l] = ((int[]) seqNums.elementAt(l)).length - 1;\r
\r
- for (int j = start; j <= end; j++) {\r
- double bigtot = 0;\r
+\r
+ for (j = start; j <= end; j++)\r
+ {\r
+ bigtot = 0;\r
\r
// First Xr = depends on column only\r
- double[] x = new double[24];\r
+ x = new double[24];\r
\r
- for (int ii = 0; ii < 24; ii++) {\r
+ for (ii = 0; ii < 24; ii++)\r
+ {\r
x[ii] = 0;\r
\r
- try {\r
- for (int i2 = 0; i2 < 24; i2++) {\r
+ try\r
+ {\r
+ for (int i2 = 0; i2 < 24; i2++)\r
+ {\r
x[ii] += (((double) cons2[j][i2] * BLOSUM62[ii][i2]) +\r
4);\r
}\r
- } catch (Exception e) {\r
+ }\r
+ catch (Exception e)\r
+ {\r
System.err.println("Exception during quality calculation.");\r
e.printStackTrace();\r
}\r
}\r
\r
// Now calculate D for each position and sum\r
- for (int k = 0; k < size; k++) {\r
- double tot = 0;\r
- double[] xx = new double[24];\r
- int seqNum = (j < lengths[k])\r
+ for (int k = 0; k < size; k++)\r
+ {\r
+ tot = 0;\r
+ xx = new double[24];\r
+ seqNum = (j < lengths[k])\r
? ((int[]) seqNums.elementAt(k))[j + 1] : 23; // Sequence, or gap at the end\r
\r
// This is a loop over r\r
- for (int i = 0; i < 23; i++) {\r
- double sr = 0;\r
+ for (i = 0; i < 23; i++)\r
+ {\r
+ sr = 0;\r
\r
- try {\r
+ try\r
+ {\r
sr = (double) BLOSUM62[i][seqNum] + 4;\r
- } catch (Exception e) {\r
+ }\r
+ catch (Exception e)\r
+ {\r
System.out.println("Exception in sr: " + e);\r
e.printStackTrace();\r
}\r
}\r
\r
// This is the quality for one column\r
- if (max < bigtot) {\r
+ if (max < bigtot)\r
+ {\r
max = bigtot;\r
}\r
\r
// bigtot = bigtot * (size-cons2[j][23])/size;\r
quality.addElement(new Double(bigtot));\r
\r
- s += "-";\r
\r
// Need to normalize by gaps\r
}\r
\r
double newmax = -10000;\r
\r
- for (int j = start; j <= end; j++) {\r
- double tmp = ((Double) quality.elementAt(j)).doubleValue();\r
+ for (j = start; j <= end; j++)\r
+ {\r
+ tmp = ((Double) quality.elementAt(j)).doubleValue();\r
tmp = ((max - tmp) * (size - cons2[j][23])) / size;\r
\r
// System.out.println(tmp+ " " + j);\r
quality.setElementAt(new Double(tmp), j);\r
\r
- if (tmp > newmax) {\r
+ if (tmp > newmax)\r
+ {\r
newmax = tmp;\r
}\r
}\r
git://source.jalview.org / jalview.git / blobdiff