Vector sequences;\r
int start;\r
int end;\r
-\r
- Vector total = new Vector();\r
-\r
- String consString = "";\r
-\r
- DrawableSequence consSequence;\r
- Hashtable propHash;\r
- int threshold;\r
+ Vector seqNums; // vector of int vectors where first is sequence checksum\r
+ int maxLength=0; // used by quality calcs\r
+ boolean seqNumsChanged = false; // updated after any change via calcSeqNum;\r
+ private void calcSeqNums() {\r
+ for (int i=0; i<sequences.size(); i++) {\r
+ calcSeqNum(i);\r
+ }\r
+ }\r
+ private void calcSeqNum(int i) {\r
+ String sq=null; // for dumb jbuilder not-inited exception warning\r
+ int[] sqnum = null;\r
+ if (i>-1 && i<sequences.size()) {\r
+ sq = ((SequenceI)sequences.elementAt(i)).getSequence();\r
+ if (seqNums.size()<=i) {\r
+ seqNums.addElement(new int[sq.length()+1]);\r
+ }\r
+ if (sq.hashCode()!=((int[])seqNums.elementAt(i))[0])\r
+ {\r
+ int j, len;\r
+ seqNumsChanged = true;\r
+ sq = ((SequenceI) sequences.elementAt(i)).getSequence();\r
+ len = sq.length();\r
+ if (maxLength < len)\r
+ maxLength = len;\r
+ sqnum = new int[len + 1]; // better to always make a new array - sequence can change its length\r
+ sqnum[0] = sq.hashCode();\r
+ for (j = 1; j <= len; j++)\r
+ {\r
+ sqnum[j] = ( (Integer) jalview.schemes.ResidueProperties.aaHash.get(new\r
+ String(sq.substring(j - 1, j)))).intValue(); // yuk\r
+ }\r
+ seqNums.setElementAt(sqnum, i);\r
+ }\r
+ } else {\r
+ // JBPNote INFO level debug\r
+ System.out.println("calcSeqNum called with out of range sequence index for Alignment\n");\r
+ }\r
+ }\r
+ Vector total = new Vector();\r
+ public Vector quality;\r
+ public Double[] qualityRange = new Double[2];\r
+ String consString = "";\r
+\r
+ Sequence consSequence;\r
+ Hashtable propHash;\r
+ int threshold;\r
\r
String name = "";\r
\r
this.sequences = sequences;\r
this.start = start;\r
this.end = end;\r
+ seqNums = new Vector(sequences.size());\r
+ calcSeqNums();\r
}\r
\r
\r
residueHash = new Hashtable();\r
\r
for (int j=0; j < sequences.size(); 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
\r
}\r
}\r
}\r
+\r
total.addElement(resultHash);\r
}\r
}\r
\r
- public int countGaps(int j) {\r
+ public int countGaps(int j)\r
+ {\r
int count = 0;\r
\r
- for (int i = 0; i < sequences.size();i++) {\r
+ for (int i = 0; i < sequences.size();i++)\r
+ {\r
if( j+1 > ((Sequence)sequences.elementAt(i)).getSequence().length())\r
{ count++; continue;}\r
\r
- String tmp = ((Sequence)sequences.elementAt(i)).getSequence().substring(j,j+1);\r
- if (tmp.equals(" ") || tmp.equals(".") || tmp.equals("-")) {\r
+ char c = ((Sequence)sequences.elementAt(i)).getSequence().charAt(j);\r
+ if (jalview.util.Comparison.isGap((c)))\r
count++;\r
- }\r
+\r
}\r
return count;\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
+ {\r
+ int count = 0;\r
+ int cons=0;\r
+ int nres = 0;\r
+ int[] r = new int[2];\r
+ char f='$';\r
+ for (int i = 0; i < sequences.size();i++)\r
+ {\r
+ if( j >= ((Sequence)sequences.elementAt(i)).getSequence().length())\r
+ { count++;\r
+ continue;}\r
+\r
+ char c = ((Sequence)sequences.elementAt(i)).getSequence().charAt(j);\r
+ if (jalview.util.Comparison.isGap((c)))\r
+ count++;\r
+ else {\r
+ nres++;\r
+ if (nres==1) {\r
+ f = c;\r
+ cons++;\r
+ } else\r
+ if (f == c) {\r
+ cons++;\r
+ }\r
+ }\r
+ }\r
+ r[0] = (nres==cons) ? 1 : 0;\r
+ r[1] = count;\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 totGaps = countGaps(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
{\r
Hashtable resultHash = (Hashtable)total.elementAt(i-start);\r
}\r
\r
if (count < 10)\r
- consString = consString + String.valueOf(count);\r
+ consString = consString + String.valueOf(count);// Conserved props!=Identity\r
else\r
- consString = consString + "*";\r
+ consString = consString + ((gapcons[0]==1) ? "*" : "+");\r
\r
}\r
else\r
}\r
}\r
\r
- consSequence = new DrawableSequence(name,consString,start,end);\r
+ consSequence = new Sequence(name,consString,start,end);\r
}\r
\r
- public jalview.gui.DrawableSequence getConsSequence() {\r
+ public Sequence getConsSequence() {\r
return consSequence;\r
}\r
\r
+ // From Alignment.java in jalview118\r
+\r
+ public void findQuality() {\r
+ findQuality(0,maxLength-1);\r
+ }\r
+\r
+ int[][] cons2;\r
+\r
+ private void percentIdentity2() {\r
+ calcSeqNums(); // updates maxLength, too.\r
+ if (cons2==null || seqNumsChanged) {\r
+ cons2 = new int[maxLength][24];\r
+ // Initialize the array\r
+ for (int j=0;j<24;j++) {\r
+ for (int i=0; i < maxLength;i++) {\r
+ cons2[i][j] = 0;\r
+ }\r
+ }\r
+\r
+ int sqnum[];\r
+ int j = 0;\r
+ while(j < sequences.size()) {\r
+ sqnum=(int[])seqNums.elementAt(j);\r
+ for (int i = 1; i < sqnum.length; i++) {\r
+ cons2[i-1][sqnum[i]]++;\r
+ }\r
+ for (int i=sqnum.length-1; i<maxLength; i++) {\r
+ cons2[i][23]++; // gap count\r
+ }\r
+ j++;\r
+ }\r
+\r
+ // unnecessary ?\r
+ /* for (int i=start; i <= end; i++) {\r
+ int max = -1000;\r
+ int maxi = -1;\r
+ int maxj = -1;\r
+\r
+ for (int j=0;j<24;j++) {\r
+ if (cons2[i][j] > max) {\r
+ max = cons2[i][j];\r
+ maxi = i;\r
+ maxj = j;\r
+ }\r
+\r
+ }\r
+ } */\r
+ }\r
+\r
+}\r
+\r
+\r
+public void findQuality(int start, int end) {\r
+ quality = new Vector();\r
+ double max = -10000;\r
+ String s = "";\r
+ int[][] BLOSUM62 = jalview.schemes.ResidueProperties.getBLOSUM62();\r
+ //Loop over columns // JBPNote Profiling info\r
+ // long ts = System.currentTimeMillis();\r
+ //long te = System.currentTimeMillis();\r
+ percentIdentity2();\r
+\r
+ int size = seqNums.size();\r
+ int[] lengths = new int[size];\r
+\r
+ for (int 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
+ // First Xr = depends on column only\r
+ double x[] = new double[24];\r
+\r
+ for (int ii=0; ii < 24; ii++) {\r
+ x[ii] = 0;\r
+ try {\r
+ for (int i2=0; i2 < 24; i2++) {\r
+ x[ii] += (double)cons2[j][i2] * BLOSUM62[ii][i2]+4;\r
+ }\r
+ } catch (Exception e) {\r
+ System.out.println("Exception : " + e);\r
+ }\r
+ //System.out.println("X " + ii + " " + x[ii]);\r
+ x[ii] /= (size);\r
+ //System.out.println("X " + ii + " " + x[ii]);\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 =\r
+ (j<lengths[k])\r
+ ? ((int[]) seqNums.elementAt(k))[j+1]\r
+ : 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
+ try {\r
+ sr = (double)BLOSUM62[i][seqNum]+4;\r
+ } catch (Exception e) {\r
+ System.out.println("Exception in sr: " + e);\r
+ }\r
+ //Calculate X with another loop over residues\r
+\r
+ // System.out.println("Xi " + i + " " + x[i] + " " + sr);\r
+ xx[i] = x[i] - sr;\r
+\r
+ tot += xx[i]*xx[i];\r
+ }\r
+ bigtot += Math.sqrt(tot);\r
+ }\r
+ // This is the quality for one column\r
+ if (max < bigtot) {max = bigtot;}\r
+ // bigtot = bigtot * (size-cons2[j][23])/size;\r
+\r
+ quality.addElement(new Double(bigtot));\r
+\r
+ s += "-";\r
+\r
+ // Need to normalize by gaps\r
+ }\r
+ double newmax=-10000;\r
+ for (int j=start; j <= end; j++) {\r
+ double tmp = ((Double)quality.elementAt(j)).doubleValue();\r
+ tmp = (max - tmp)*(size-cons2[j][23])/size;\r
+ // System.out.println(tmp+ " " + j);\r
+ quality.setElementAt(new Double(tmp),j);\r
+ if (tmp>newmax)\r
+ newmax = tmp;\r
+ }\r
+ // System.out.println("Quality " + s);\r
+ qualityRange[0] = new Double(0);\r
+ qualityRange[1] = new Double(newmax);\r
+ }\r
+\r
+\r
}\r
git://source.jalview.org / jalview.git / blobdiff