=hconsensus.length) {
- // happens if sequences calculated over were shorter than alignment width
- consensus.annotations[i]=null;
- continue;
- }
- value = 0;
- if (ignoreGapsInConsensusCalculation)
+ Hashtable hci;
+ if (i >= hconsensus.length || ((hci = hconsensus[i]) == null))
{
- value = ((Float) hconsensus[i].get(AAFrequency.PID_NOGAPS))
- .floatValue();
+ // happens if sequences calculated over were shorter than alignment
+ // width
+ consensus.annotations[i] = null;
+ continue;
}
- else
+ Float fv = (Float) hci
+ .get(ignoreGapsInConsensusCalculation ? PID_NOGAPS : PID_GAPS);
+ if (fv == null)
{
- value = ((Float) hconsensus[i].get(AAFrequency.PID_GAPS))
- .floatValue();
+ consensus.annotations[i] = null;
+ // data has changed below us .. give up and
+ continue;
}
-
- String maxRes = hconsensus[i].get(AAFrequency.MAXRESIDUE).toString();
- String mouseOver = hconsensus[i].get(AAFrequency.MAXRESIDUE) + " ";
+ float value = fv.floatValue();
+ String maxRes = hci.get(AAFrequency.MAXRESIDUE).toString();
+ StringBuilder mouseOver = new StringBuilder(64);
if (maxRes.length() > 1)
{
- mouseOver = "[" + maxRes + "] ";
+ mouseOver.append("[").append(maxRes).append("] ");
maxRes = "+";
}
- int[][] profile = (int[][]) hconsensus[i].get(AAFrequency.PROFILE);
+ else
+ {
+ mouseOver.append(hci.get(AAFrequency.MAXRESIDUE) + " ");
+ }
+ int[][] profile = (int[][]) hci.get(AAFrequency.PROFILE);
+ int sequenceCount = profile[1][0];
+ int nonGappedCount = profile[1][1];
+ int normalisedBy = ignoreGapsInConsensusCalculation ? nonGappedCount
+ : sequenceCount;
if (profile != null && includeAllConsSymbols)
{
- mouseOver = "";
+ mouseOver.setLength(0);
if (alphabet != null)
{
for (int c = 0; c < alphabet.length; c++)
{
- tval = ((float) profile[0][alphabet[c]])
- * 100f
- / (float) profile[1][ignoreGapsInConsensusCalculation ? 1
- : 0];
- mouseOver += ((c == 0) ? "" : "; ") + alphabet[c] + " "
- + ((int) tval) + "%";
+ float tval = profile[0][alphabet[c]] * 100f / normalisedBy;
+ mouseOver
+ .append(((c == 0) ? "" : "; "))
+ .append(alphabet[c])
+ .append(" ")
+ .append(((fmt != null) ? fmt.form(tval) : ((int) tval)))
+ .append("%");
}
}
else
{
- Object[] ca = new Object[profile[0].length];
+ // TODO do this sort once only in calculate()?
+ // char[][] ca = new char[profile[0].length][];
+ char[] ca = new char[profile[0].length];
float[] vl = new float[profile[0].length];
for (int c = 0; c < ca.length; c++)
{
- ca[c] = new char[]
- { (char) c };
- vl[c] = (float) profile[0][c];
+ ca[c] = (char) c;
+ // ca[c] = new char[]
+ // { (char) c };
+ vl[c] = profile[0][c];
}
- ;
- jalview.util.QuickSort.sort(vl, ca);
- for (int p = 0, c = ca.length - 1; profile[0][((char[]) ca[c])[0]] > 0; c--)
+ QuickSort.sort(vl, ca);
+ for (int p = 0, c = ca.length - 1; profile[0][ca[c]] > 0; c--)
{
- if (((char[]) ca[c])[0] != '-')
+ final char residue = ca[c];
+ if (residue != '-')
{
- tval = ((float) profile[0][((char[]) ca[c])[0]])
- * 100f
- / (float) profile[1][ignoreGapsInConsensusCalculation ? 1
- : 0];
- mouseOver += ((p == 0) ? "" : "; ") + ((char[]) ca[c])[0]
- + " " + ((int) tval) + "%";
+ float tval = profile[0][residue] * 100f / normalisedBy;
+ mouseOver
+ .append((((p == 0) ? "" : "; ")))
+ .append(residue)
+ .append(" ")
+ .append(((fmt != null) ? fmt.form(tval)
+ : ((int) tval))).append("%");
p++;
-
}
}
-
}
}
else
{
- mouseOver += ((int) value + "%");
+ mouseOver.append(
+ (((fmt != null) ? fmt.form(value) : ((int) value))))
+ .append("%");
}
- consensus.annotations[i] = new Annotation(maxRes, mouseOver, ' ',
+ consensus.annotations[i] = new Annotation(maxRes,
+ mouseOver.toString(), ' ',
value);
}
}
/**
- * get the sorted profile for the given position of the consensus
+ * Returns a Format designed to show all significant figures for profile
+ * percentages. For less than 100 sequences, returns null (the integer
+ * percentage value will be displayed). For 100-999 sequences, returns "%3.1f"
+ *
+ * @param nseq
+ * @return
+ */
+ protected static Format getPercentageFormat(long nseq)
+ {
+ int scale = 0;
+ while (nseq >= 10)
+ {
+ scale++;
+ nseq /= 10;
+ }
+ return scale <= 1 ? null : new Format("%3." + (scale - 1) + "f");
+ }
+
+ /**
+ * Returns the sorted profile for the given consensus data. The returned array
+ * contains
+ *
+ *
+ * [profileType, numberOfValues, nonGapCount, charValue1, percentage1, charValue2, percentage2, ...]
+ * in descending order of percentage value
+ *
*
* @param hconsensus
+ * the data table from which to extract and sort values
+ * @param ignoreGaps
+ * if true, only non-gapped values are included in percentage
+ * calculations
* @return
*/
public static int[] extractProfile(Hashtable hconsensus,
- boolean ignoreGapsInConsensusCalculation)
+ boolean ignoreGaps)
{
int[] rtnval = new int[64];
int[][] profile = (int[][]) hconsensus.get(AAFrequency.PROFILE);
if (profile == null)
+ {
return null;
- Object[] ca = new Object[profile[0].length];
+ }
+ char[] ca = new char[profile[0].length];
float[] vl = new float[profile[0].length];
for (int c = 0; c < ca.length; c++)
{
- ca[c] = new char[]
- { (char) c };
- vl[c] = (float) profile[0][c];
+ ca[c] = (char) c;
+ vl[c] = profile[0][c];
+ }
+ QuickSort.sort(vl, ca);
+ int nextArrayPos = 2;
+ int totalPercentage = 0;
+ int distinctValuesCount = 0;
+ final int divisor = profile[1][ignoreGaps ? 1 : 0];
+ for (int c = ca.length - 1; profile[0][ca[c]] > 0; c--)
+ {
+ if (ca[c] != '-')
+ {
+ rtnval[nextArrayPos++] = ca[c];
+ final int percentage = (int) (profile[0][ca[c]] * 100f / divisor);
+ rtnval[nextArrayPos++] = percentage;
+ totalPercentage += percentage;
+ distinctValuesCount++;
+ }
+ }
+ rtnval[0] = distinctValuesCount;
+ rtnval[1] = totalPercentage;
+ int[] result = new int[rtnval.length + 1];
+ result[0] = AlignmentAnnotation.SEQUENCE_PROFILE;
+ System.arraycopy(rtnval, 0, result, 1, rtnval.length);
+
+ return result;
+ }
+
+ /**
+ * Extract a sorted extract of cDNA codon profile data. The returned array
+ * contains
+ *
+ *
+ * [profileType, numberOfValues, totalCount, charValue1, percentage1, charValue2, percentage2, ...]
+ * in descending order of percentage value, where the character values encode codon triplets
+ *
+ *
+ * @param hashtable
+ * @return
+ */
+ public static int[] extractCdnaProfile(Hashtable hashtable, boolean ignoreGaps)
+ {
+ // this holds #seqs, #ungapped, and then codon count, indexed by encoded
+ // codon triplet
+ int[] codonCounts = (int[]) hashtable.get(PROFILE);
+ int[] sortedCounts = new int[codonCounts.length - 2];
+ System.arraycopy(codonCounts, 2, sortedCounts, 0,
+ codonCounts.length - 2);
+
+ int[] result = new int[3 + 2 * sortedCounts.length];
+ // first value is just the type of profile data
+ result[0] = AlignmentAnnotation.CDNA_PROFILE;
+
+ char[] codons = new char[sortedCounts.length];
+ for (int i = 0; i < codons.length; i++)
+ {
+ codons[i] = (char) i;
+ }
+ QuickSort.sort(sortedCounts, codons);
+ int totalPercentage = 0;
+ int distinctValuesCount = 0;
+ int j = 3;
+ int divisor = ignoreGaps ? codonCounts[1] : codonCounts[0];
+ for (int i = codons.length - 1; i >= 0; i--)
+ {
+ final int codonCount = sortedCounts[i];
+ if (codonCount == 0)
+ {
+ break; // nothing else of interest here
+ }
+ distinctValuesCount++;
+ result[j++] = codons[i];
+ final int percentage = codonCount * 100 / divisor;
+ result[j++] = percentage;
+ totalPercentage += percentage;
+ }
+ result[2] = totalPercentage;
+
+ /*
+ * Just return the non-zero values
+ */
+ // todo next value is redundant if we limit the array to non-zero counts
+ result[1] = distinctValuesCount;
+ return Arrays.copyOfRange(result, 0, j);
+ }
+
+ /**
+ * Compute a consensus for the cDNA coding for a protein alignment.
+ *
+ * @param alignment
+ * the protein alignment (which should hold mappings to cDNA
+ * sequences)
+ * @param hconsensus
+ * the consensus data stores to be populated (one per column)
+ */
+ public static void calculateCdna(AlignmentI alignment,
+ Hashtable[] hconsensus)
+ {
+ final char gapCharacter = alignment.getGapCharacter();
+ Set mappings = alignment.getCodonFrames();
+ if (mappings == null || mappings.isEmpty())
+ {
+ return;
}
- ;
- jalview.util.QuickSort.sort(vl, ca);
- rtnval[0] = 1;
- for (int c = ca.length - 1; profile[0][((char[]) ca[c])[0]] > 0; c--)
+
+ int cols = alignment.getWidth();
+ for (int col = 0; col < cols; col++)
{
- if (((char[]) ca[c])[0] != '-')
+ // todo would prefer a Java bean for consensus data
+ Hashtable columnHash = new Hashtable();
+ // #seqs, #ungapped seqs, counts indexed by (codon encoded + 1)
+ int[] codonCounts = new int[66];
+ codonCounts[0] = alignment.getSequences().size();
+ int ungappedCount = 0;
+ for (SequenceI seq : alignment.getSequences())
{
- rtnval[rtnval[0]++] = ((char[]) ca[c])[0];
- rtnval[rtnval[0]++] = (int) (((float) profile[0][((char[]) ca[c])[0]]) * 100f / (float) profile[1][ignoreGapsInConsensusCalculation ? 1
- : 0]);
+ if (seq.getCharAt(col) == gapCharacter)
+ {
+ continue;
+ }
+ char[] codon = MappingUtils.findCodonFor(seq, col, mappings);
+ int codonEncoded = CodingUtils.encodeCodon(codon);
+ if (codonEncoded >= 0)
+ {
+ codonCounts[codonEncoded + 2]++;
+ ungappedCount++;
+ }
}
+ codonCounts[1] = ungappedCount;
+ // todo: sort values here, save counts and codons?
+ columnHash.put(PROFILE, codonCounts);
+ hconsensus[col] = columnHash;
+ }
+ }
+
+ /**
+ * Derive displayable cDNA consensus annotation from computed consensus data.
+ *
+ * @param consensusAnnotation
+ * the annotation row to be populated for display
+ * @param consensusData
+ * the computed consensus data
+ * @param showProfileLogo
+ * if true show all symbols present at each position, else only the
+ * modal value
+ * @param nseqs
+ * the number of sequences in the alignment
+ */
+ public static void completeCdnaConsensus(
+ AlignmentAnnotation consensusAnnotation,
+ Hashtable[] consensusData, boolean showProfileLogo, int nseqs)
+ {
+ if (consensusAnnotation == null
+ || consensusAnnotation.annotations == null
+ || consensusAnnotation.annotations.length < consensusData.length)
+ {
+ // called with a bad alignment annotation row - wait for it to be
+ // initialised properly
+ return;
+ }
+
+ for (int col = 0; col < consensusData.length; col++)
+ {
+ Hashtable hci = consensusData[col];
+ if (hci == null)
+ {
+ // gapped protein column?
+ continue;
+ }
+ // array holds #seqs, #ungapped, then codon counts indexed by codon
+ final int[] codonCounts = (int[]) hci.get(PROFILE);
+ int totalCount = 0;
+ StringBuilder mouseOver = new StringBuilder(32);
+
+ /*
+ * First pass - get total count and find the highest
+ */
+ final char[] codons = new char[codonCounts.length - 2];
+ for (int j = 2; j < codonCounts.length; j++)
+ {
+ final int codonCount = codonCounts[j];
+ codons[j - 2] = (char) (j - 2);
+ totalCount += codonCount;
+ }
+
+ /*
+ * Sort array of encoded codons by count ascending - so the modal value
+ * goes to the end; start by copying the count (dropping the first value)
+ */
+ int[] sortedCodonCounts = new int[codonCounts.length - 2];
+ System.arraycopy(codonCounts, 2, sortedCodonCounts, 0,
+ codonCounts.length - 2);
+ QuickSort.sort(sortedCodonCounts, codons);
+
+ int modalCodonEncoded = codons[codons.length - 1];
+ int modalCodonCount = sortedCodonCounts[codons.length - 1];
+ String modalCodon = String.valueOf(CodingUtils
+ .decodeCodon(modalCodonEncoded));
+ if (sortedCodonCounts.length > 1
+ && sortedCodonCounts[codons.length - 2] == modalCodonEncoded)
+ {
+ modalCodon = "+";
+ }
+ float pid = sortedCodonCounts[sortedCodonCounts.length - 1] * 100
+ / (float) totalCount;
+
+ /*
+ * todo ? Replace consensus hashtable with sorted arrays of codons and
+ * counts (non-zero only). Include total count in count array [0].
+ */
+
+ /*
+ * Scan sorted array backwards for most frequent values first.
+ */
+ for (int j = codons.length - 1; j >= 0; j--)
+ {
+ int codonCount = sortedCodonCounts[j];
+ if (codonCount == 0)
+ {
+ break;
+ }
+ int codonEncoded = codons[j];
+ final int pct = codonCount * 100 / totalCount;
+ String codon = String
+ .valueOf(CodingUtils.decodeCodon(codonEncoded));
+ Format fmt = getPercentageFormat(nseqs);
+ String formatted = fmt == null ? Integer.toString(pct) : fmt
+ .form(pct);
+ if (showProfileLogo || codonCount == modalCodonCount)
+ {
+ mouseOver.append(codon).append(": ").append(formatted)
+ .append("% ");
+ }
+ }
+
+ consensusAnnotation.annotations[col] = new Annotation(modalCodon,
+ mouseOver.toString(), ' ', pid);
}
- return rtnval;
}
}