/**
* Fills the structure (str) string with the predicted structure according to running the CYK
* algorithm and traceback on the provided sequence string (seq). The parameters should be
* double[3] including the following probabilities:
*
* <p>params[0] = p(S->L) params[1] = p(L->s) params[2] = p(F->LS)
*
* @param seq - The sequence to predict from.
* @param str - The string to fill with the predicted value.
* @param params - The Knudson-Hein Grammar parameters as described above.
* @param verbose - Display output to command line if true.
*/
public static String predictKH(String seq, BigDouble[] params, boolean verbose) {
int size = seq.length();
BigDouble[][][] parr = new BigDouble[size][size][3];
int[][][] tau = new int[size][size][3];
String pred = null;
if (size < 1) {
output.out("Invalid sequence provided: \n\t[" + seq + "]");
return pred;
}
output.out("Predicting secondary structure for \n\t\t[" + seq + "]");
for (int i = 0; i < size; i++)
for (int j = 0; j < size; j++) for (int k = 0; k < 3; k++) tau[i][j][k] = -1;
// Fill Array
kh_CYK(seq, params, parr, tau);
BigDouble prob = parr[0][size - 1][0];
output.out("\t-KH Maximum Probability for sequence is " + prob);
// Trace Back
if (prob.compareTo(0) > 0) {
pred = kh_trace_back(tau);
output.out("\t-KH most likely structure for sequence is\n\t\t[ " + pred + " ]");
}
return pred;
}
public static String predictFromGrammar23S(
String seq, String nat, Grammar grammar, boolean verbose) {
int size = seq.length();
Map<String, BigDouble> parr = new HashMap<String, BigDouble>();
int[][][] tau = new int[size][size][3];
output.out("\t\tTau allocated");
String pred = null;
if (size < 1) {
output.out("\nInvalid sequence provided: \n\t[ " + seq + " ]");
return pred;
}
output.out("\nPredicting secondary structure for \n\t\t[ " + seq + " ]");
long start = System.currentTimeMillis();
for (int i = 0; i < size; i++) {
// Progress Bar
CommandLine.DisplayBar(seq.length(), i, ((long) (System.currentTimeMillis() - start) / 1000));
for (int j = i; j < size; j++) {
String ij = i + ":" + (i + j);
for (int k = 0; k < 3; k++) {
tau[i][j][k] = -1;
parr.put(ij + ":" + k, new BigDouble(0));
}
}
}
CommandLine.DisplayBarFinish();
// Fill Array
if (grammar instanceof PfoldGrammar) {
PfoldGrammar pfold = (PfoldGrammar) grammar;
// _23S_CYK(seq, pfold.getKH_params(), pfold
// .getPfold_paramsUnmatched(), pfold
// .getPfold_paramsBasePairs(), parr, tau);
}
BigDouble prob = parr.get("0:" + (size - 1) + ":" + 0);
output.out("\t-Probability of highest probability parse for sequence is \n\t\t" + prob);
if (prob != null) {
// grammar.recordProbability(prob);
// Trace Back
if (prob.compareTo(0) > 0) {
pred = kh_trace_back(tau);
{
output.out("\t-Highest probability parse generates structure\n\t\t[ " + pred + " ]");
if (nat != null)
output.out(
"\t-FMeasure for predicted structure is\n\t\t" + Compare.getFMeasureBD(nat, pred));
}
}
}
return pred;
}
/**
* Fills the structure (str) string with the predicted structure according to running the CYK
* algorithm and traceback on the provided sequence string (seq). The parameters should be
* double[3] including the following probabilities:
*
* <p>params[0] = p(S->L) params[1] = p(L->s) params[2] = p(F->LS)
*
* @param seq - The sequence to predict from.
* @param str - The string to fill with the predicted value.
* @param params - The Knudson-Hein Grammar parameters as described above.
* @param verbose - Display output to command line if true.
*/
public static String predictFromGrammar(String seq, Grammar grammar) {
int size = seq.length();
BigDecimal[][][] parr = new BigDecimal[size][size][3];
int[][][] tau = new int[size][size][3];
String pred = null;
if (size < 1) {
return pred;
}
for (int i = 0; i < size; i++) {
for (int j = 0; j < size; j++) {
for (int k = 0; k < 3; k++) {
tau[i][j][k] = -1;
parr[i][j][k] = new BigDecimal(0);
}
}
}
if (grammar instanceof PfoldGrammar) {
PfoldGrammar pfold = (PfoldGrammar) grammar;
// BigDouble[] kh_params = new BigDouble[pfold.get_kh_params().length];
// for(int i=0;i<kh_params.length;i++)
// kh_params[i] = new BigDouble(pfold.get_kh_params()[i].doubleValue());
// BigDouble[] Pfold_paramsUnmatched = new
// BigDouble[pfold.getPfold_paramsUnmatched().length];
// for(int i=0;i<Pfold_paramsUnmatched.length;i++)
// Pfold_paramsUnmatched[i] = new
// BigDouble(pfold.getPfold_paramsUnmatched()[i].doubleValue());
// BigDouble[][] Pfold_paramsBasePairs = new
// BigDouble[pfold.getPfold_paramsBasePairs().length][pfold.getPfold_paramsBasePairs()[0].length];
// for(int i=0;i<Pfold_paramsBasePairs.length;i++)
// for(int j=0;j<Pfold_paramsBasePairs[i].length;j++)
// Pfold_paramsBasePairs[i][j] = new
// BigDouble(pfold.getPfold_paramsBasePairs()[i][j].doubleValue());
BigDecimal[] kh_params = new BigDecimal[pfold.get_kh_params().length];
for (int i = 0; i < kh_params.length; i++)
kh_params[i] = new BigDecimal(pfold.get_kh_params()[i].doubleValue());
BigDecimal[] Pfold_paramsUnmatched = new BigDecimal[pfold.getPfold_paramsUnmatched().length];
for (int i = 0; i < Pfold_paramsUnmatched.length; i++)
Pfold_paramsUnmatched[i] =
new BigDecimal(pfold.getPfold_paramsUnmatched()[i].doubleValue());
BigDecimal[][] Pfold_paramsBasePairs =
new BigDecimal[pfold.getPfold_paramsBasePairs().length]
[pfold.getPfold_paramsBasePairs()[0].length];
for (int i = 0; i < Pfold_paramsBasePairs.length; i++)
for (int j = 0; j < Pfold_paramsBasePairs[i].length; j++)
Pfold_paramsBasePairs[i][j] =
new BigDecimal(pfold.getPfold_paramsBasePairs()[i][j].doubleValue());
Pfold_CYK(seq, kh_params, Pfold_paramsUnmatched, Pfold_paramsBasePairs, parr, tau);
}
// BigDouble prob = parr[0][size - 1][0];
BigDecimal prob = parr[0][size - 1][0];
int compVal = prob.compareTo(BigDecimal.ZERO);
if (compVal > 0) pred = kh_trace_back(tau);
else
output.out(
"Probability of highest probability parse is <= 0. ("
+ prob.toString()
+ " :: "
+ compVal
+ ")");
return pred;
}