SiteContainer* SiteContainerTools::getCompleteSites(const SiteContainer& sites)
{
vector<string> seqNames = sites.getSequencesNames();
VectorSiteContainer* noGapCont = new VectorSiteContainer(seqNames.size(), sites.getAlphabet());
noGapCont->setSequencesNames(seqNames, false);
CompleteSiteContainerIterator csi(sites);
while (csi.hasMoreSites())
{
noGapCont->addSite(*csi.nextSite());
}
return noGapCont;
}
SiteContainer* SiteContainerTools::removeGapSites(const SiteContainer& sites, double maxFreqGaps)
{
vector<string> seqNames = sites.getSequencesNames();
VectorSiteContainer* noGapCont = new VectorSiteContainer(seqNames.size(), sites.getAlphabet());
noGapCont->setSequencesNames(seqNames, false);
for (unsigned int i = 0; i < sites.getNumberOfSites(); ++i) {
map<int, double> freq;
SiteTools::getFrequencies(sites.getSite(i), freq);
if (freq[-1] <= maxFreqGaps)
noGapCont->addSite(sites.getSite(i), false);
}
return noGapCont;
}
SiteContainer* SiteContainerTools::removeGapOrUnresolvedOnlySites(const SiteContainer& sites)
{
vector<string> seqNames = sites.getSequencesNames();
VectorSiteContainer* noGapCont = new VectorSiteContainer(seqNames.size(), sites.getAlphabet());
noGapCont->setSequencesNames(seqNames, false);
for (unsigned int i = 0; i < sites.getNumberOfSites(); i++)
{
const Site* site = &sites.getSite(i);
if (!SiteTools::isGapOrUnresolvedOnly(*site))
noGapCont->addSite(*site, false);
}
return noGapCont;
}
SiteContainer* SiteContainerTools::getSelectedSites(
const SiteContainer& sequences,
const SiteSelection& selection)
{
vector<string> seqNames = sequences.getSequencesNames();
VectorSiteContainer* sc = new VectorSiteContainer(seqNames.size(), sequences.getAlphabet());
sc->setSequencesNames(seqNames, false);
for (unsigned int i = 0; i < selection.size(); i++)
{
sc->addSite(sequences.getSite(selection[i]), false);
// We do not check names, we suppose that the container passed as an argument is correct.
// WARNING: what if selection contains many times the same indice? ...
}
sc->setGeneralComments(sequences.getGeneralComments());
return sc;
}
SiteContainer* SiteContainerTools::removeStopCodonSites(const SiteContainer& sites) throw (AlphabetException)
{
const CodonAlphabet* pca = dynamic_cast<const CodonAlphabet*>(sites.getAlphabet());
if (!pca)
throw AlphabetException("Not a Codon Alphabet", sites.getAlphabet());
vector<string> seqNames = sites.getSequencesNames();
VectorSiteContainer* noStopCont = new VectorSiteContainer(seqNames.size(), sites.getAlphabet());
noStopCont->setSequencesNames(seqNames, false);
for (unsigned int i = 0; i < sites.getNumberOfSites(); i++)
{
const Site* site = &sites.getSite(i);
if (!SiteTools::hasStopCodon(*site))
noStopCont->addSite(*site, false);
}
return noStopCont;
}
SiteContainer* SiteContainerTools::resolveDottedAlignment(
const SiteContainer& dottedAln,
const Alphabet* resolvedAlphabet) throw (AlphabetException, Exception)
{
if (!AlphabetTools::isDefaultAlphabet(dottedAln.getAlphabet()))
throw AlphabetException("SiteContainerTools::resolveDottedAlignment. Alignment alphabet should of class 'DefaultAlphabet'.", dottedAln.getAlphabet());
// First we look for the reference sequence:
size_t n = dottedAln.getNumberOfSequences();
if (n == 0)
throw Exception("SiteContainerTools::resolveDottedAlignment. Input alignment contains no sequence.");
const Sequence* refSeq = 0;
for (size_t i = 0; i < n; ++i) // Test each sequence
{
const Sequence* seq = &dottedAln.getSequence(i);
bool isRef = true;
for (unsigned int j = 0; isRef && j < seq->size(); ++j) // For each site in the sequence
{
if (seq->getChar(j) == ".")
isRef = false;
}
if (isRef) // We found the reference sequence!
{
refSeq = new BasicSequence(*seq);
}
}
if (!refSeq)
throw Exception("SiteContainerTools::resolveDottedAlignment. No reference sequence was found in the input alignment.");
// Now we build a new VectorSiteContainer:
VectorSiteContainer* sites = new VectorSiteContainer(n, resolvedAlphabet);
// We add each site one by one:
size_t m = dottedAln.getNumberOfSites();
string state;
for (unsigned int i = 0; i < m; ++i)
{
string resolved = refSeq->getChar(i);
const Site* site = &dottedAln.getSite(i);
Site resolvedSite(resolvedAlphabet, site->getPosition());
for (unsigned int j = 0; j < n; j++)
{
state = site->getChar(j);
if (state == ".")
{
state = resolved;
}
resolvedSite.addElement(state);
}
// Add the new site:
sites->addSite(resolvedSite);
}
// Seq sequence names:
sites->setSequencesNames(dottedAln.getSequencesNames());
// Delete the copied sequence:
delete refSeq;
// Return result:
return sites;
}