void BiPolerCaller::simulateData(bool setDeleteTag)
{
deletionQV = rndPct(gen) <= lowDelPct ? lowDelPhred : highDelPhred;
insrtionQV = rndPct(gen) <= lowInsPct ? lowInsPhred : highInsPhred;
substitutionQV = rndPct(gen) <= lowSubPct ? lowSubPhred : highSubPhred;
const long double pDeletion = phred2prob(deletionQV);
const long double pInsertion = phred2prob(insrtionQV);
const long double pSubstitution = phred2prob(substitutionQV);
const long double pError = pDeletion + pInsertion + pSubstitution;
qualityValue = prob2phred(pError);
if (rndProb(gen) < pError)
{
const long double pErrorType = rndProb(gen);
if (pErrorType < pDeletion/pError)
deletionError();
else if (pErrorType < (pDeletion + pInsertion)/pError)
insertionError(setDeleteTag);
else
substitutinoError(setDeleteTag);
}
else
{
baseCall = t[locus];
if (setDeleteTag) deletionTag = randomBase(baseCall);
substitutionTag = randomBase(baseCall);
}
}
// This is Equation 2 (tranformed to log) from
// A statistical framework for SNP calling ... , Heng Li, Bioinformatics, 2011
// http://bioinformatics.oxfordjournals.org/content/27/21/2987.full
double Caller::genotype_log_likelihood(const BasePileup& bp,
const vector<pair<int, int> >& base_offsets,
double g, char first, char second) {
double m = 2.; // always assume two alleles
double log_likelihood = log(0.25); // 1 / m^2, where m = ploidy = 2;
const string& bases = bp.bases();
const string& quals = bp.qualities();
double perr;
for (int i = 0; i < base_offsets.size(); ++i) {
char base = Pileups::extract_match(bp, base_offsets[i].first);
char qual = base_offsets[i].second >= 0 ? quals[base_offsets[i].second] : _default_quality;
perr = phred2prob(qual);
if (base == first) {
log_likelihood += safe_log((m - g) * perr + g * (1. - perr));
} else if (base == second) {
log_likelihood += safe_log((m - g) * (1. - perr) + g * perr);
} else {
log_likelihood += safe_log(perr * perr);
}
}
return log_likelihood;
}
BiPolerCaller::BiPolerCaller(const string &t
, char lowDelPhred
, char highDelPhred
, char lowInsPhred
, char highInsPhred
, char lowSubPhred
, char highSubPhred
, short lowDelPct
, short lowInsPct
, short lowSubPct
, unsigned int locus)
: BaseCaller(t, locus)
,lowDelPhred(lowDelPhred)
,highDelPhred(highDelPhred)
,lowInsPhred(lowInsPhred)
,highInsPhred(highInsPhred)
,lowSubPhred(lowSubPhred)
,highSubPhred(highSubPhred)
,lowDelPct(lowDelPct)
,lowInsPct(lowInsPct)
,lowSubPct(lowSubPct)
,probLowDel(phred2prob(lowDelPhred))
,probHighDel(phred2prob(highDelPhred))
,probLowIns(phred2prob(lowInsPhred))
,probHighIns(phred2prob(highInsPhred))
,probLowSub(phred2prob(lowSubPhred))
,probHighSub(phred2prob(highSubPhred))
,rndPct(1,100)
{
simulateData();
}