CRef<CSeq_entry> CAlnReader::GetSeqEntry()
{
if (m_Entry) {
return m_Entry;
} else if ( !m_ReadDone ) {
NCBI_THROW2(CObjReaderParseException, eFormat,
"CAlnReader::GetSeqEntry(): "
"Seq_entry is not available until after Read()", 0);
}
m_Entry = new CSeq_entry();
CRef<CSeq_annot> seq_annot (new CSeq_annot);
seq_annot->SetData().SetAlign().push_back(GetSeqAlign());
m_Entry->SetSet().SetClass(CBioseq_set::eClass_pop_set);
m_Entry->SetSet().SetAnnot().push_back(seq_annot);
CBioseq_set::TSeq_set& seq_set = m_Entry->SetSet().SetSeq_set();
typedef CDense_seg::TDim TNumrow;
for (TNumrow row_i = 0; row_i < m_Dim; row_i++) {
const string& seq_str = m_SeqVec[row_i];
const size_t& seq_str_len = seq_str.size();
CRef<CSeq_entry> seq_entry (new CSeq_entry);
// seq-id(s)
CBioseq::TId& ids = seq_entry->SetSeq().SetId();
CSeq_id::ParseFastaIds(ids, m_Ids[row_i], true);
if (ids.empty()) {
ids.push_back(CRef<CSeq_id>(new CSeq_id(CSeq_id::e_Local,
m_Ids[row_i])));
}
// mol
CSeq_inst::EMol mol = CSeq_inst::eMol_not_set;
CSeq_id::EAccessionInfo ai = ids.front()->IdentifyAccession();
if (ai & CSeq_id::fAcc_nuc) {
mol = CSeq_inst::eMol_na;
} else if (ai & CSeq_id::fAcc_prot) {
mol = CSeq_inst::eMol_aa;
} else {
switch (CFormatGuess::SequenceType(seq_str.data(), seq_str_len)) {
case CFormatGuess::eNucleotide: mol = CSeq_inst::eMol_na; break;
case CFormatGuess::eProtein: mol = CSeq_inst::eMol_aa; break;
default: break;
}
}
// seq-inst
CRef<CSeq_inst> seq_inst (new CSeq_inst);
seq_entry->SetSeq().SetInst(*seq_inst);
seq_set.push_back(seq_entry);
// repr
seq_inst->SetRepr(CSeq_inst::eRepr_raw);
// mol
seq_inst->SetMol(mol);
// len
_ASSERT(seq_str_len == m_SeqLen[row_i]);
seq_inst->SetLength(seq_str_len);
// data
CSeq_data& data = seq_inst->SetSeq_data();
if (mol == CSeq_inst::eMol_aa) {
data.SetIupacaa().Set(seq_str);
} else {
data.SetIupacna().Set(seq_str);
CSeqportUtil::Pack(&data);
}
}
return m_Entry;
}
CRef<CBioseq_set> CMakeCdrProds::MakeCdrProds(CRef<CSeq_annot> annot,
CBioseq_Handle handle)
{
CRef<CBioseq_set> bioseq_set(new CBioseq_set);
if (!annot->GetData().IsFtable()) {
// Is this the right thing to do?
// Could throw, or could return null CRef instead.
return bioseq_set;
}
list<CRef<CSeq_feat> >& ftable = annot->SetData().SetFtable();
NON_CONST_ITERATE (list<CRef<CSeq_feat> >, feat, ftable) {
if (!(*feat)->GetData().IsCdregion()) {
// not interested if not a Cdregion
continue;
}
if ((*feat)->IsSetProduct()) {
// already has a product; don't make new one
continue;
}
string prot;
CSeqTranslator::Translate(**feat, handle.GetScope(), prot);
CRef<CSeq_data> seq_data(new CSeq_data(prot,
CSeq_data::e_Iupacaa));
CRef<CSeq_inst> seq_inst(new CSeq_inst);
seq_inst->SetSeq_data(*seq_data);
seq_inst->SetRepr(CSeq_inst_Base::eRepr_raw);
seq_inst->SetMol(CSeq_inst_Base::eMol_aa);
seq_inst->SetLength(prot.size());
CRef<CBioseq> bio_seq(new CBioseq);
string num = NStr::NumericToString(sm_Counter.Add(1));
// pad to five digits
if (num.size() < 5) {
num.insert(SIZE_TYPE(0), 5 - num.size(), '0');
}
string acc = "tp" + num;
string full_acc = "lcl|" + acc;
CRef<CSeq_id> id(new CSeq_id(full_acc));
bio_seq->SetId().push_back(id);
// a title
CRef<CSeqdesc> title(new CSeqdesc);
title->SetTitle(string("Translation product ") + acc);
bio_seq->SetDescr().Set().push_back(title);
// Mol_type
CRef<CSeqdesc> mol_type(new CSeqdesc);
mol_type->SetMol_type( eGIBB_mol_peptide);
bio_seq->SetDescr().Set().push_back(mol_type);
// set the instance
bio_seq->SetInst(*seq_inst);
// wrap this Bio_seq in an entry
CRef<CSeq_entry> seq_entry(new CSeq_entry);
seq_entry->SetSeq(*bio_seq);
// add this entry to our Bioseq_set
bioseq_set->SetSeq_set().push_back(seq_entry);
// record it as product in the annot we're handed
CRef<CSeq_loc> prod_loc(new CSeq_loc);
prod_loc->SetWhole(*id);
(*feat)->SetProduct(*prod_loc);
}
return bioseq_set;
}
