* This class is used extensively in calculating alignment colourschemes that
* depend on the amount of conservation in each alignment column.
*
- * @author $author$
- * @version $Revision$
*/
public class AAFrequency
{
- public static final String PROFILE = "P";
-
- private static final String AMINO = "amino";
-
- private static final String DNA = "DNA";
-
- private static final String RNA = "RNA";
+ private static final double LOG2 = Math.log(2);
- /*
- * Quick look-up of String value of char 'A' to 'Z'
- */
- private static final String[] CHARS = new String['Z' - 'A' + 1];
-
- static
- {
- for (char c = 'A'; c <= 'Z'; c++)
- {
- CHARS[c - 'A'] = String.valueOf(c);
- }
- }
+ public static final String PROFILE = "P";
public static final ProfilesI calculate(List<SequenceI> list, int start,
int end)
}
}
-
-
/**
* Calculate the consensus symbol(s) for each column in the given range.
*
information.annotations[i] = null;
return 0;
}
-
- HiddenMarkovModel hmm;
SequenceI hmmSeq = information.sequenceRef;
- hmm = hmmSeq.getHMM();
+ HiddenMarkovModel hmm = hmmSeq.getHMM();
- Float value = getInformationContent(i, hmm);
+ float value = hmm.getInformationContent(i);
if (value > max)
{
}
/**
- * Returns the information content at a specified column.
+ * Returns the sorted HMM profile for the given column of the alignment. The
+ * returned array contains
*
- * @param column
- * Index of the column, starting from 0.
- * @return
- */
- public static float getInformationContent(int column,
- HiddenMarkovModel hmm)
- {
- float informationContent = 0f;
-
- for (char symbol : hmm.getSymbols())
- {
- float freq = 0f;
- freq = ResidueProperties.backgroundFrequencies
- .get(hmm.getAlphabetType()).get(symbol);
- Double hmmProb = hmm.getMatchEmissionProbability(column, symbol);
- float prob = hmmProb.floatValue();
- informationContent += prob * (Math.log(prob / freq) / Math.log(2));
-
- }
-
- return informationContent;
- }
-
- /**
- * Produces a HMM profile for a column in an alignment
+ * <pre>
+ * [profileType=0, numberOfValues, 100, charValue1, percentage1, charValue2, percentage2, ...]
+ * in descending order of percentage value
+ * </pre>
*
- * @param aa
- * Alignment annotation for which the profile is being calculated.
+ * @param hmm
* @param column
- * Column in the alignment the profile is being made for.
* @param removeBelowBackground
- * Boolean indicating whether to ignore residues with probabilities
- * less than their background frequencies.
+ * if true, ignores residues with probability less than their
+ * background frequency
+ * @param infoHeight
+ * if true, uses the log ratio 'information' measure to scale the
+ * value
* @return
*/
public static int[] extractHMMProfile(HiddenMarkovModel hmm, int column,
boolean removeBelowBackground, boolean infoHeight)
{
-
- if (hmm != null)
+ if (hmm == null)
{
- int size = hmm.getNumberOfSymbols();
- char symbols[] = new char[size];
- int values[] = new int[size];
- List<Character> charList = hmm.getSymbols();
- Integer totalCount = 0;
+ return null;
+ }
+ int size = hmm.getNumberOfSymbols();
+ char symbols[] = new char[size];
+ int values[] = new int[size];
+ List<Character> charList = hmm.getSymbols();
+ int totalCount = 0;
- for (int i = 0; i < size; i++)
- {
- char symbol = charList.get(i);
- symbols[i] = symbol;
- int value = getAnalogueCount(hmm, column, symbol,
- removeBelowBackground, infoHeight);
- values[i] = value;
- totalCount += value;
- }
+ for (int i = 0; i < size; i++)
+ {
+ char symbol = charList.get(i);
+ symbols[i] = symbol;
+ int value = getAnalogueCount(hmm, column, symbol,
+ removeBelowBackground, infoHeight);
+ values[i] = value;
+ totalCount += value;
+ }
- QuickSort.sort(values, symbols);
+ /*
+ * sort symbols by increasing emission probability
+ */
+ QuickSort.sort(values, symbols);
- int[] profile = new int[3 + size * 2];
+ int[] profile = new int[3 + size * 2];
- profile[0] = AlignmentAnnotation.SEQUENCE_PROFILE;
- profile[1] = size;
- profile[2] = 100;
+ profile[0] = AlignmentAnnotation.SEQUENCE_PROFILE;
+ profile[1] = size;
+ profile[2] = 100;
- if (totalCount != 0)
+ /*
+ * order symbol/count profile by decreasing emission probability
+ */
+ if (totalCount != 0)
+ {
+ int arrayPos = 3;
+ for (int k = size - 1; k >= 0; k--)
{
- int arrayPos = 3;
- for (int k = size - 1; k >= 0; k--)
+ Float percentage;
+ int value = values[k];
+ if (removeBelowBackground)
{
- Float percentage;
- Integer value = values[k];
- if (removeBelowBackground)
- {
- percentage = (value.floatValue() / totalCount.floatValue())
- * 100;
- }
- else
- {
- percentage = value.floatValue() / 100f;
- }
- int intPercent = Math.round(percentage);
- profile[arrayPos] = symbols[k];
- profile[arrayPos + 1] = intPercent;
- arrayPos += 2;
+ percentage = ((float) value) / totalCount * 100f;
}
+ else
+ {
+ percentage = value / 100f;
+ }
+ int intPercent = Math.round(percentage);
+ profile[arrayPos] = symbols[k];
+ profile[arrayPos + 1] = intPercent;
+ arrayPos += 2;
}
- return profile;
}
- return null;
+ return profile;
}
/**
* Converts the emission probability of a residue at a column in the alignment
- * to a 'count' to allow for processing by the annotation renderer.
+ * to a 'count', suitable for rendering as an annotation value
*
* @param hmm
* @param column
- * @param removeBelowBackground
- * When true, this method returns 0 for any symbols with a match
- * emission probability less than the background frequency.
* @param symbol
+ * @param removeBelowBackground
+ * if true, returns 0 for any symbol with a match emission
+ * probability less than the background frequency
+ * @infoHeight if true, uses the log ratio 'information content' to scale the
+ * value
* @return
*/
static int getAnalogueCount(HiddenMarkovModel hmm, int column,
char symbol, boolean removeBelowBackground, boolean infoHeight)
{
- Double value;
-
- value = hmm.getMatchEmissionProbability(column, symbol);
- double freq;
-
- freq = ResidueProperties.backgroundFrequencies
+ double value = hmm.getMatchEmissionProbability(column, symbol);
+ double freq = ResidueProperties.backgroundFrequencies
.get(hmm.getAlphabetType()).get(symbol);
if (value < freq && removeBelowBackground)
{
if (infoHeight)
{
- value = value * (Math.log(value / freq) / Math.log(2));
+ value = value * (Math.log(value / freq) / LOG2);
}
- value = value * 10000;
- return Math.round(value.floatValue());
+ value = value * 10000d;
+ return Math.round((float) value);
}
}