/// Constructor
CQueryFactoryInfo::CQueryFactoryInfo(CRef<IQueryFactory> query_factory,
EBlastProgramType program)
: m_IsProt(Blast_SubjectIsProtein(program) ? true : false), m_MaxLength(0),
m_MinLength(1), m_AvgLength(0), m_QuerySource(0), m_NumSeqs(0)
{
CRef<IRemoteQueryData> query_data(query_factory->MakeRemoteQueryData());
CRef<CBioseq_set> bss(query_data->GetBioseqSet());
_ASSERT(bss.NotEmpty());
m_QuerySource.Reset(new CBlastQuerySourceBioseqSet(*bss, m_IsProt));
if ( !m_QuerySource ) {
NCBI_THROW(CBlastException, eSeqSrcInit,
"Failed to initialize sequences for IQueryFactory");
}
// TODO support for m_MinLength
SetupSubjects_OMF(*m_QuerySource, program, &m_SeqBlkVector, &m_MaxLength);
m_NumSeqs = static_cast<Uint4>(m_QuerySource->Size());
_ASSERT(!m_SeqBlkVector.empty());
}
/// Constructor
CMultiSeqInfo::CMultiSeqInfo(TSeqLocVector& seq_vector,
EBlastProgramType program,
bool dbscan_mode)
{
m_ibIsProt = Blast_SubjectIsProtein(program) ? true : false;
m_DbScanMode = dbscan_mode;
m_iTotalLength=0;
// Fix subject location for tblast[nx].
if (Blast_SubjectIsTranslated(program))
{
TSeqLocVector temp_slv;
vector<Int2> strand_v;
ITERATE(TSeqLocVector, iter, seq_vector)
{
strand_v.push_back((Int2) (*iter).seqloc->GetStrand());
CRef<CSeq_loc> sl(new CSeq_loc);
sl->Assign(*((*iter).seqloc));
sl->SetStrand(eNa_strand_both);
if ((*iter).mask)
{
CRef<CSeq_loc> mask_sl(new CSeq_loc);
mask_sl->Assign(*((*iter).mask));
SSeqLoc sseq_loc(*sl, *((*iter).scope), *mask_sl);
temp_slv.push_back(sseq_loc);
}
else
{
SSeqLoc sseq_loc(*sl, *((*iter).scope));
temp_slv.push_back(sseq_loc);
}
}
SetupSubjects(temp_slv, program, &m_ivSeqBlkVec, &m_iMaxLength);
int index=0;
ITERATE(vector<Int2>, s_iter, strand_v)
{
m_ivSeqBlkVec[index++]->subject_strand = *s_iter;
}
}
CQueryFactoryInfo::CQueryFactoryInfo(const TSeqLocVector& subj_seqs,
EBlastProgramType program)
: m_IsProt(Blast_SubjectIsProtein(program) ? true : false), m_MaxLength(0),
m_MinLength(1), m_AvgLength(0), m_QuerySource(0), m_NumSeqs(subj_seqs.size())
{
// Fix subject location for tblast[nx].
if (Blast_SubjectIsTranslated(program))
{
TSeqLocVector temp_slv;
vector<Int2> strand_v;
ITERATE(TSeqLocVector, iter, subj_seqs)
{
strand_v.push_back((Int2) (*iter).seqloc->GetStrand());
CRef<CSeq_loc> sl(new CSeq_loc);
sl->Assign(*((*iter).seqloc));
sl->SetStrand(eNa_strand_both);
if ((*iter).mask)
{
CRef<CSeq_loc> mask_sl(new CSeq_loc);
mask_sl->Assign(*((*iter).mask));
SSeqLoc sseq_loc(*sl, *((*iter).scope), *mask_sl);
temp_slv.push_back(sseq_loc);
}
else
{
SSeqLoc sseq_loc(*sl, *((*iter).scope));
temp_slv.push_back(sseq_loc);
}
}
SetupSubjects(temp_slv, program, &m_SeqBlkVector, &m_MaxLength);
int index=0;
ITERATE(vector<Int2>, s_iter, strand_v)
{
m_SeqBlkVector[index++]->subject_strand = *s_iter;
}
}
void
InitializeSubject(CRef<blast::CBlastDatabaseArgs> db_args,
CRef<blast::CBlastOptionsHandle> opts_hndl,
bool is_remote_search,
CRef<blast::CLocalDbAdapter>& db_adapter,
CRef<objects::CScope>& scope)
{
db_adapter.Reset();
_ASSERT(db_args.NotEmpty());
CRef<CSearchDatabase> search_db = db_args->GetSearchDatabase();
// Initialize the scope...
if (is_remote_search) {
const bool is_protein =
Blast_SubjectIsProtein(opts_hndl->GetOptions().GetProgramType())
? true : false;
SDataLoaderConfig config(is_protein);
if (search_db.NotEmpty() && search_db->GetDatabaseName() != "n/a") {
config.m_BlastDbName = search_db->GetDatabaseName();
}
CBlastScopeSource scope_src(config);
// configure scope to fetch sequences remotely for formatting
if (scope.NotEmpty()) {
scope_src.AddDataLoaders(scope);
} else {
scope = scope_src.NewScope();
}
} else {
if (scope.Empty()) {
scope.Reset(new CScope(*CObjectManager::GetInstance()));
}
}
_ASSERT(scope.NotEmpty());
// ... and then the subjects
CRef<IQueryFactory> subjects;
if ( (subjects = db_args->GetSubjects(scope)) ) {
_ASSERT(search_db.Empty());
db_adapter.Reset(new CLocalDbAdapter(subjects, opts_hndl));
} else {
_ASSERT(search_db.NotEmpty());
try {
// Try to open the BLAST database even for remote searches, as if
// it is available locally, it will be better to fetch the
// sequence data for formatting from this (local) source
CRef<CSeqDB> seqdb = search_db->GetSeqDb();
db_adapter.Reset(new CLocalDbAdapter(*search_db));
scope->AddDataLoader(RegisterOMDataLoader(seqdb));
} catch (const CSeqDBException&) {
// The BLAST database couldn't be found, report this for local
// searches, but for remote searches go on.
if (is_remote_search ) {
db_adapter.Reset(new CLocalDbAdapter(*search_db));
} else {
throw;
}
}
}
/// Set the BLASTDB data loader as the default data loader (if applicable)
if (search_db.NotEmpty()) {
string dbloader_name =
s_FindBlastDbDataLoaderName(search_db->GetDatabaseName(),
search_db->IsProtein());
if ( !dbloader_name.empty() ) {
// FIXME: will this work with multiple BLAST DBs?
scope->AddDataLoader(dbloader_name,
CBlastDatabaseArgs::kSubjectsDataLoaderPriority);
_TRACE("Setting " << dbloader_name << " priority to "
<< (int)CBlastDatabaseArgs::kSubjectsDataLoaderPriority
<< " for subjects");
}
}
}