A RetroSearch Logo

Home - News ( United States | United Kingdom | Italy | Germany ) - Football scores

Search Query:

Showing content from http://www.ncbi.nlm.nih.gov/IEB/ToolBox/CPP_DOC/doxyhtml/seqdb_8cpp_source.html below:

NCBI C++ ToolKit: src/objtools/blast/seqdb_reader/seqdb.cpp Source File

75  "Invalid sequence type specified."

);

122  if

(prot_nucl ==

'-'

) {

161  if

(

dbname

.size() == 0) {

164  "Database name is required."

);

183  if

(

dbname

.size() == 0) {

186  "Database name is required."

);

189  const bool

kUseAtlasLock =

true

;

206  if

(

dbname

.size() == 0) {

209  "Database name is required."

);

232  if

(

dbname

.size() == 0) {

235  "Database name is required."

);

289  if

(

dbname

.size() == 0) {

292  "Database name is required."

);

298  if

(! ids.

Blank

()) {

306  const bool

kUseAtlasLock =

true

;

326  if

(

dbname

.size() == 0) {

329  "Database name is required."

);

332  const bool

kUseAtlasLock =

true

;

350  if

(

dbname

.size() == 0) {

353  "Database name is required."

);

356  const bool

kUseAtlasLock =

true

;

377  if

(

dbname

.size() == 0) {

380  "Database name is required."

);

383  const bool

kUseAtlasLock =

true

;

438  "Internal sequence type is not valid."

);

447

TmpMap gi_to_taxid_tmp;

450

gi_to_taxid.

clear

();

452  ITERATE

( TmpMap, it, gi_to_taxid_tmp ) {

453

gi_to_taxid[it->first] = it->second;

459

vector<TTaxId> & taxids,

481

TmpMap gi_to_taxid_set_tmp;

484

gi_to_taxid_set.clear();

486  ITERATE

( TmpMap, it, gi_to_taxid_set_tmp ) {

487

gi_to_taxid_set[it->first] = it->second;

494

vector<TTaxId>& taxids,

574  int

end_offset)

const 602  "Invalid allocation strategy specified."

);

623  "Invalid allocation strategy specified."

);

648  ITERATE

(vector<string>, vol, vols) {

650

ifstream

f

(fn.c_str(),

ios::in

|ios::binary);

653  f

.seekg(8, ios::beg);

661  if

(retv.

IsEmpty

() || d > retv) {

913  if

(! oids.empty()) {

1030  if

(! oids.empty()) {

1042

vector<string> & paths,

1043

vector<string> * alias_paths,

1078

list< CRef<CSeq_id> > seqids =

GetSeqIDs

(oid);

1085  if

((**seqid).IsGi()) {

1086

gis.push_back((**seqid).GetGi());

1112  Uint8

* total_length,

1113  bool

use_approx)

const 1161

raw.assign(

buffer

, length);

1175  if

(code_from == coding) {

1189 #if ((!defined(NCBI_COMPILER_WORKSHOP) || (NCBI_COMPILER_VERSION > 550)) && \ 1190  (!defined(NCBI_COMPILER_MIPSPRO)) ) 1215  const string

& volname)

1244

vector<int> invalid_algo_ids, available_algo_ids;

1246

invalid_algo_ids.reserve(algorithm_ids.size());

1247  if

(available_algo_ids.empty()) {

1248  copy

(algorithm_ids.begin(), algorithm_ids.end(),

1249

back_inserter(invalid_algo_ids));

1250  return

invalid_algo_ids;

1253  ITERATE

(vector<int>, itr, algorithm_ids) {

1254

vector<int>::const_iterator pos = find(available_algo_ids.begin(),

1255

available_algo_ids.end(), *itr);

1256  if

(pos == available_algo_ids.end()) {

1257

invalid_algo_ids.push_back(*itr);

1260  return

invalid_algo_ids;

1265  string

& program_name,

1278  string

& program_name,

1330  string

retval(

"Unknown"

);

1332  case eProtein

: retval.assign(

"Protein"

);

break

;

1333  case eNucleotide

: retval.assign(

"Nucleotide"

);

break

;

1354  const string

& extn = de.

GetPath

().substr(de.

GetPath

().length() - 3, 1);

1360

oss <<

"\""

<<

value

.m_BlastDbName <<

"\""

;

1362  value

.m_MoleculeType =

1374  string

alias = retval +

".nal"

, index = retval +

".nin"

;

1375

retval = (

CFile

(alias).

Exists

() ? alias : index);

1377  string

alias = retval +

".pal"

, index = retval +

".pin"

;

1378

retval = (

CFile

(alias).

Exists

() ? alias : index);

1400  for

(

size_t i

= 0;

i

< finder.

m_DBs

.size();

i

++) {

1402  if

(path[path.size()-1] !=

'l'

) {

1410  while

(getline(

in

, line)) {

1412

vector<string> tokens;

1414  for

(

size_t

j = 1; j < tokens.size(); j++) {

1415

dbs2remove.

insert

(tokens[j]);

1424

finder.

m_DBs

.end());

1428

vector<SSeqDBInitInfo>

1430  bool

include_alias_files

,

1431  bool

remove_redundant_dbs

)

1434

vector<string> fmasks, dmasks;

1438  if

(dbtype !=

"nucl"

) {

1439

fmasks.push_back(

"*.pin"

);

1440  if

(include_alias_files) {

1441

fmasks.push_back(

"*.pal"

);

1444  if

(dbtype !=

"prot"

) {

1445

fmasks.push_back(

"*.nin"

);

1446  if

(include_alias_files) {

1447

fmasks.push_back(

"*.nal"

);

1450

dmasks.push_back(

"*"

);

1457  if

(remove_redundant_dbs) {

1461  return

dbfinder.

m_DBs

;

1466

vector<string> paths;

1472

vector<string> extn;

1475  string

blastdb_dirname;

1477  ITERATE

(vector<string>, path, paths) {

1478  ITERATE

(vector<string>, ext, extn) {

1480  if

(

file

.Exists()) {

1485

blastdb_dirname =

file

.GetDir();

1488

<<

file

.GetPath());

1495  if

(paths.size() > 1) {

1496  _ASSERT

( !blastdb_dirname.empty() );

1498

vector<string> dblist;

1500  if

(dblist.size() > 1) {

1502

oss <<

"Cannot compute disk usage for multiple BLASTDBs (i.e.: '" 1503

<<

dbname

<<

"') at once. Please try again using one BLASTDB " 1508  for

(

const auto

& ext: extn) {

1510  if

(

file

.Exists()) {

1517

<<

file

.GetPath());

1537  _ASSERT

(

"Unknown molecule for BLAST DB"

!= 0);

1544  int

num_files_removed = 0;

1545

vector<string> db_files, alias_files;

1548

vector<string> extn;

1550

vector<string> lmdb_extn;

1554

oss << dbpath <<

"."

<< *

lmdb

;

1556  if

(

CFile

(fname).Remove()) {

1558

num_files_removed++;

1561  unsigned int

index = 0;

1564  while

(

CFile

(vfname).Remove()) {

1575  ITERATE

(vector<string>,

f

, db_files) {

1576  ITERATE

(vector<string>, e, extn) {

1578

oss << *

f

<<

"."

<< *e;

1580  if

(

CFile

(fname).Remove()) {

1582

num_files_removed++;

1586  ITERATE

(vector<string>,

f

, alias_files) {

1589

num_files_removed++;

1592  return static_cast<bool>

(num_files_removed != 0);

1612

vector<string> paths;

1613

vector<string> alias;

1621  ITERATE

(vector<string>,

a

, alias) {

1624  string

fn = user_path + af.

GetName

();

1625

db_files.push_back(fn);

1635

vector<string> extn;

1639  const string

kExtnMol(1, is_protein ?

'p'

:

'n'

);

1640  const string

index_ext = kExtnMol +

"in"

;

1641  const string

seq_ext = kExtnMol +

"sq"

;

1643  ITERATE

(vector<string>, path, paths) {

1644  ITERATE

(vector<string>, ext, extn) {

1646  if

(

file

.Exists()) {

1647  string f

= user_path +

file

.GetName();

1648

db_files.push_back(

f

);

1651

disk_bytes += length;

1652  if

((*ext == index_ext) || (*ext == seq_ext)) {

1653

cached_bytes += length;

1657

<<

file

.GetPath());

1664

vector<string> lmdb_list;

1667  ITERATE

(vector<string>,

l

, lmdb_list) {

1669  if

(

file

.Exists()) {

1670  string f

= user_path +

file

.GetName();

1671

db_files.push_back(

f

);

1674

disk_bytes += length;

1678  static const char

* v5_exts[]={

"os"

,

"ot"

,

"tf"

,

"to"

,

NULL

};

1679  for

(

const char

** p=v5_exts; *p !=

NULL

; p++) {

1680  CFile

v(

file

.GetDir() +

file

.GetBase() +

"."

+ kExtnMol + (*p));

1682  string

vf = user_path + v.

GetName

();

1683

db_files.push_back(vf);

1702  sort

(db_files.begin(), db_files.end());

1709  Uint8

total_length = 0;

1712

vector<string> dblist;

1716  if

(off != string::npos ) {

1717

(*itr).erase(0, off+1);

1722

m->SetDbname(dbnames);

1727

m->SetNumber_of_letters(total_length);

1728

m->SetNumber_of_sequences(num_seqs);

1732

m->SetLast_updated(date.

AsString

(timeFmt));

1734  Int8

disk_bytes(0), cached_bytes(0);

1736

m->SetBytes_total(disk_bytes);

1737

m->SetBytes_to_cache(cached_bytes);

1744  if

((tax_ids.

size

() > 1) || ((tax_ids.

size

() == 1) && (0 != *tax_ids.

begin

()))){

1745

m->SetNumber_of_taxids(

static_cast<int>

(tax_ids.

size

()));

void remove_if(Container &c, Predicate *__pred)

`Blob' Class for SeqDB (and WriteDB).

Functor class for FindFilesInDir.

void operator()(CDirEntry &de)

vector< SSeqDBInitInfo > m_DBs

string GetFileName(size_t idx)

Auxiliary function to get the original file name found by this object.

void SetFrame(const string &frame)

void Log(const string &name, const char *value, CDebugDumpFormatter::EValueType type=CDebugDumpFormatter::eValue, const string &comment=kEmptyStr)

CNcbiOstrstreamToString class helps convert CNcbiOstrstream to a string Sample usage:

static SIZE_TYPE Convert(const CTempString &src, TCoding src_coding, TSeqPos pos, TSeqPos length, string &dst, TCoding dst_coding)

static const string GenerateSearchPath()

Generate search path.

SeqDB ID list for performing boolean set operations.

bool Blank() const

Check if an ID list is blank.

bool IsPositive()

Checks whether a positive GI list was produced.

CRef< CSeqDBNegativeList > GetNegativeList()

Retrieve a negative GI list.

CRef< CSeqDBGiList > GetPositiveList()

Retrieve a positive GI list.

void GetTaxIDs(int oid, map< TGi, TTaxId > &gi_to_taxid, bool persist)

Get gi to taxid map for an OID.

int GetSequence(int oid, const char **buffer)

Get the sequence data for a sequence.

void GetDBTaxIds(set< TTaxId > &tax_ids)

Get all unique tax ids from db.

char GetSeqType() const

Get the sequence type.

int GetOidAtOffset(int first_seq, Uint8 residue) const

Find the OID corresponding to the offset given in residues, into the database as a whole.

int GetMinLength() const

Returns the length of the smallest sequence in the database.

void GetLeafTaxIDs(int oid, map< TGi, set< TTaxId > > &gi_to_taxid_set, bool persist)

Get gi to taxid map for an OID.

void AccessionsToOids(const vector< string > &accs, vector< blastdb::TOid > &oids)

EBlastDbVersion GetBlastDbVersion() const

Return blast db version.

int GetMaskAlgorithmId(const string &algo_name)

Get the numeric ID for a algorithm name.

int GetColumnId(const string &title)

Get an ID number for a given column title.

bool GiToOidwFilterCheck(TGi gi, int &oid)

GiToOis is meant to simply return oid for a gi if one exisits This method finds the oid and checks if...

void GetColumnBlob(int col_id, int oid, bool keep, CBlastDbBlob &blob)

Fetch the data blob for the given column and oid.

void SetIterationRange(int oid_begin, int oid_end)

Set Iteration Range.

int GetNumOIDs() const

Returns the size of the (possibly sparse) OID range.

void GetMaskAlgorithmDetails(int algorithm_id, string &program, string &program_name, string &algo_opts)

Get information about one type of masking available here.

void SetNumberOfThreads(int num_threads, bool force_mt=false)

Invoke the garbage collector to free up memory.

TGi GetSeqGI(int oid)

Look up for the GI of a sequence.

int GetMaxLength() const

Returns the length of the largest sequence in the database.

void TaxIdsToOids(set< TTaxId > &tax_ids, vector< blastdb::TOid > &rv)

Get Oid list for input tax ids.

void RetAmbigSeq(const char **buffer) const

Returns any resources associated with the sequence.

void FlushOffsetRangeCache()

Flush all offset ranges cached.

void GetAllTaxIDs(int oid, set< TTaxId > &taxids)

Get all tax ids (leaf and non-leaf for an oid.

Uint8 GetVolumeLength() const

Returns the sum of the lengths of all volumes.

void GetLMDBFileNames(vector< string > &lmdb_list) const

const string & GetDBNameList() const

Get list of database names.

CRef< CBioseq > GetBioseq(int oid, TGi target_gi, const CSeq_id *target_seq_id, bool seqdata)

Get a CBioseq for a sequence.

static void FindVolumePaths(const string &dbname, char prot_nucl, vector< string > &paths, vector< string > *alias_paths, bool recursive, bool expand_links)

Find volume paths.

void ListColumns(vector< string > &titles)

List columns titles found in this database.

int GetAmbigPartialSeq(int oid, char **buffer, int nucl_code, ESeqDBAllocType alloc_type, CSeqDB::TSequenceRanges *partial_ranges, CSeqDB::TSequenceRanges *masks) const

bool OidToPig(int oid, int &pig) const

Translate a PIG to an OID.

void GetAliasFileValues(TAliasFileValues &afv)

Get Name/Value Data From Alias Files.

Uint8 GetExactTotalLength()

Returns the exact sum of the lengths of all available sequences.

void GetTaxIdsForSeqId(const CSeq_id &seq_id, vector< TTaxId > &taxids)

bool IdsToOids(CSeqDBGiList &id_list)

Get OIDs from an ID list.

CRef< CSeq_data > GetSeqData(int oid, TSeqPos begin, TSeqPos end) const

Fetch data as a CSeq_data object.

const map< string, string > & GetColumnMetaData(int column_id)

Get all metadata for the specified column.

static void GetTaxInfo(TTaxId taxid, SSeqDBTaxInfo &info)

Get taxonomy information.

string GetAvailableMaskAlgorithmDescriptions()

Returns a formatted string with the list of available masking algorithms in this database for display...

bool CheckOrFindOID(int &next_oid)

Find an included OID, incrementing next_oid if necessary.

void SeqidToOids(const CSeq_id &seqid, vector< int > &oids, bool multi)

Translate a CSeq-id to a list of OIDs.

int GetNumSeqsStats() const

Returns the number of sequences available.

int GetAmbigSeq(int oid, char **buffer, int nucl_code, SSeqDBSlice *region, ESeqDBAllocType strategy, CSeqDB::TSequenceRanges *masks=NULL) const

Get a pointer to a range of sequence data with ambiguities.

void GetAvailableMaskAlgorithms(vector< int > &algorithms)

Get a list of algorithm IDs for which mask data exists.

int GetSeqLengthApprox(int oid) const

Get the approximate sequence length.

void GetTaxIdsForOids(const vector< blastdb::TOid > &oids, set< TTaxId > &tax_ids)

void GetTotals(ESummaryType sumtype, int *oid_count, Uint8 *total_length, bool use_approx)

Returns the sum of the sequence lengths.

void RetSequence(const char **buffer)

Returns any resources associated with the sequence.

const CSeqDBGiList * GetGiList() const

Get GI list attached to this database.

void GetMaskData(int oid, int algo_id, CSeqDB::TSequenceRanges &ranges)

Get masked ranges of a sequence.

bool TiToOid(Int8 ti, int &oid)

Translate a TI to an OID.

Uint8 GetTotalLengthStats() const

Returns the sum of the lengths of all available sequences.

void SetVolsMemBit(int mbit)

Set the membership bit of all volumes.

bool GiToOid(TGi gi, int &oid) const

Translate a GI to an OID.

void SetOffsetRanges(int oid, const TRangeList &offset_ranges, bool append_ranges, bool cache_data)

Apply a range of offsets to a database sequence.

list< CRef< CSeq_id > > GetSeqIDs(int oid)

Gets a list of sequence identifiers.

void AccessionToOids(const string &acc, vector< int > &oids)

Find OIDs matching the specified string.

CRef< CBlast_def_line_set > GetHdr(int oid)

Get the sequence header data.

int GetNumSeqs() const

Returns the number of sequences available.

Uint8 GetTotalLength() const

Returns the sum of the lengths of all available sequences.

bool PigToOid(int pig, int &oid) const

Translate a PIG to an OID.

CSeqDB::EOidListType GetNextOIDChunk(int &begin_chunk, int &end_chunk, int oid_size, vector< int > &oid_list, int *oid_state)

Return a chunk of OIDs, and update the OID bookmark.

string GetDate() const

Returns the construction date of the database.

string GetTitle() const

Returns the database title.

void ResetInternalChunkBookmark()

Restart chunk iteration at the beginning of the database.

int GetSeqLength(int oid) const

Get the sequence length.

CSeqDBIdSet GetIdSet()

Get IdSet list attached to this database.

bool OidToGi(int oid, TGi &gi)

Translate a GI to an OID.

CSeqDBIter & operator++()

Increment operator.

int m_Length

The length of this OID.

const CSeqDB * m_DB

The CSeqDB object which this object iterates over.

CSeqDBIter & operator=(const CSeqDBIter &)

Copy one iterator to another.

void x_RetSeq()

Release hold on current sequence.

const char * m_Data

The sequence data for this OID.

int m_OID

The OID this iterator is currently accessing.

void x_GetSeq()

Get data pointer and length for the current sequence.

CSeqDBIter(const CSeqDBIter &)

Construct one iterator from another.

void GetColumnBlob(int col_id, int oid, CBlastDbBlob &blob)

Fetch the data blob for the given column and oid.

bool IdsToOids(CSeqDBGiList &id_list) const

Get OIDs from an ID list.

int GetMinLength() const

Returns the length of the shortest sequence in the database.

void GetDBTaxIds(set< TTaxId > &tax_ids) const

Get all unique tax ids from db.

static void FindVolumePaths(const string &dbname, ESeqType seqtype, vector< string > &paths, vector< string > *alias_paths=NULL, bool recursive=true, bool expand_links=true)

Find volume paths.

bool OidToPig(int oid, int &pig) const

Translate an OID to a PIG.

void TaxIdsToOids(set< TTaxId > &tax_ids, vector< blastdb::TOid > &rv) const

Get Oid list for input tax ids.

bool GiToOidwFilterCheck(TGi gi, int &oid) const

Translate a GI To an OID with filter check.

Uint8 GetTotalLength() const

Returns the sum of the lengths of all available sequences.

static string ESeqType2String(ESeqType type)

Converts a CSeqDB sequence type into a human readable string.

void GetGis(int oid, vector< TGi > &gis, bool append=false) const

Gets a list of GIs for an OID.

bool PigToOid(int pig, int &oid) const

Translate a PIG to an OID.

void SetIterationRange(int oid_begin, int oid_end)

Set Iteration Range.

void GetSequenceAsString(int oid, CSeqUtil::ECoding coding, string &output, TSeqRange range=TSeqRange()) const

Get a sequence in a given encoding.

int GetNumOIDs() const

Returns the size of the (possibly sparse) OID range.

static string GenerateSearchPath()

Returns the default BLAST database search path configured for this local installation of BLAST.

TGi GetSeqGI(int oid) const

Returns the first Gi (if any) of the sequence.

vector< int > ValidateMaskAlgorithms(const vector< int > &algorithm_ids)

Validates the algorithm IDs passed to this function, returning a vector of those algorithm IDs not pr...

Uint8 GetVolumeLength() const

Returns the sum of the lengths of all volumes.

void GetAvailableMaskAlgorithms(vector< int > &algorithms)

Get a list of algorithm IDs for which mask data exists.

bool OidToGi(int oid, TGi &gi) const

Translate an OID to a GI.

const string & GetDBNameList() const

Get list of database names.

list< CRef< CSeq_id > > GetSeqIDs(int oid) const

Gets a list of sequence identifiers.

static const char * kOidNotFound

String containing the error message in exceptions thrown when a given OID cannot be found.

Int8 GetDiskUsage() const

Retrieve the disk usage in bytes for this BLAST database.

string GetAvailableMaskAlgorithmDescriptions()

Returns a formatted string with the list of available masking algorithms in this database for display...

void ResetInternalChunkBookmark()

Resets this object's internal chunk bookmark, which is used when the oid_state argument to GetNextOID...

EOidListType

Indicates how block of OIDs was returned.

CRef< CSeq_data > GetSeqData(int oid, TSeqPos begin, TSeqPos end) const

Fetch data as a CSeq_data object.

bool GiToPig(TGi gi, int &pig) const

Translate a GI to a PIG.

void GetAliasFileValues(TAliasFileValues &afv)

Get Name/Value Data From Alias Files.

void RemoveOffsetRanges(int oid)

Remove any offset ranges for the given OID.

int GetMaxLength() const

Returns the length of the largest sequence in the database.

int GetSeqLength(int oid) const

Returns the sequence length in base pairs or residues.

bool PigToGi(int pig, TGi &gi) const

Translate a PIG to a GI.

ESeqType GetSequenceType() const

Returns the type of database opened - protein or nucleotide.

const CSeqDBGiList * GetGiList() const

Get GI list attached to this database.

ESeqType

Sequence types (eUnknown tries protein, then nucleotide).

bool SeqidToOid(const CSeq_id &seqid, int &oid) const

Translate a Seq-id to any matching OID.

void RetAmbigSeq(const char **buffer) const

Returns any resources associated with the sequence.

int GetOidAtOffset(int first_seq, Uint8 residue) const

Find the sequence closest to the given offset into the database.

void SetOffsetRanges(int oid, const TRangeList &offset_ranges, bool append_ranges, bool cache_data)

Apply a range of offsets to a database sequence.

CRef< CBioseq > GetBioseq(int oid, TGi target_gi=ZERO_GI, const CSeq_id *target_seq_id=NULL) const

Get a CBioseq for a sequence.

int GetAmbigPartialSeq(int oid, char **buffer, int nucl_code, ESeqDBAllocType strategy, TSequenceRanges *partial_ranges, TSequenceRanges *masks=NULL) const

CRef< CBioseq > GetBioseqNoData(int oid, TGi target_gi=ZERO_GI, const CSeq_id *target_seq_id=NULL) const

Get a CBioseq for a sequence without sequence data.

void GetTaxIDs(int oid, map< TGi, TTaxId > &gi_to_taxid, bool persist=false) const

Get taxid for an OID.

void SetVolsMemBit(int mbit)

Set the membership of all volumes.

void GetTaxIdsForOids(const vector< blastdb::TOid > &oids, set< TTaxId > &tax_ids) const

void GetMaskAlgorithmDetails(int algorithm_id, objects::EBlast_filter_program &program, string &program_name, string &algo_opts)

Get information about one type of masking available here.

void GetTotals(ESummaryType sumtype, int *oid_count, Uint8 *total_length, bool use_approx=true) const

Returns the sum of the sequence lengths.

void RetSequence(const char **buffer) const

Returns any resources associated with the sequence.

string GetTitle() const

Returns the database title.

int GetNumSeqs() const

Returns the number of sequences available.

void GetTaxIdsForSeqId(const CSeq_id &seq_id, vector< TTaxId > &taxids)

Get all tax ids for a seq id.

EOidListType GetNextOIDChunk(int &begin_chunk, int &end_chunk, int oid_size, vector< int > &oid_list, int *oid_state=NULL)

Return a chunk of OIDs, and update the OID bookmark.

void x_GetDBFilesMetaData(Int8 &disk_bytes, Int8 &cached_bytes, vector< string > &db_files, const string &user_path) const

CRef< CBlast_db_metadata > GetDBMetaData(string user_path=kEmptyStr)

void GetAllTaxIDs(int oid, set< TTaxId > &taxids) const

Get all tax ids for an oid.

int GetSequence(int oid, const char **buffer) const

Get a pointer to raw sequence data.

void AccessionToOids(const string &acc, vector< int > &oids) const

Translate an Accession to a list of OIDs.

void ListColumns(vector< string > &titles)

List columns titles found in this database.

void GetTaxIdsForAccession(const string &accs, vector< TTaxId > &taxids)

Get all tax ids for an accessions.

bool CheckOrFindOID(int &next_oid) const

Find an included OID, incrementing next_oid if necessary.

string GetDate() const

Returns the construction date of the database.

int GetNumSeqsStats() const

Returns the number of sequences available.

ESummaryType

Types of summary information available.

@ eFilteredAll

Values from alias files, or summation over all included sequences.

int GetColumnId(const string &title)

Get an ID number for a given column title.

void SeqidToOids(const CSeq_id &seqid, vector< int > &oids) const

Translate a Seq-id to a list of OIDs.

int GetMaskAlgorithmId(const string &algo_name) const

Get the numeric algorithm ID for a string.

int GetAmbigSeqAlloc(int oid, char **buffer, int nucl_code, ESeqDBAllocType strategy, TSequenceRanges *masks=NULL) const

Get a pointer to sequence data with ambiguities.

bool TiToOid(Int8 ti, int &oid) const

Translate a TI to an OID.

CSeqDBIter Begin() const

Returns a sequence iterator.

const string & GetColumnValue(int column_id, const string &key)

Look up the value for a specific column metadata key.

static const char * kBlastDbDateFormat

Format string for the date returned by CSeqDB::GetDate.

CRef< CBioseq > GiToBioseq(TGi gi) const

Get a CBioseq for a given GI.

class CSeqDBImpl * m_Impl

Implementation details are hidden. (See seqdbimpl.hpp).

EBlastDbVersion GetBlastDbVersion() const

Return blast db version.

void SetNumberOfThreads(int num_threads, bool force_mt=false)

Setting the number of threads.

static void GetTaxInfo(TTaxId taxid, SSeqDBTaxInfo &info)

Get taxonomy information.

Uint8 GetExactTotalLength()

Returns the exact sum of the lengths of all available sequences.

CRef< CBlast_def_line_set > GetHdr(int oid) const

Get the ASN.1 header for the sequence.

void DebugDump(CDebugDumpContext ddc, unsigned int depth) const

Dump debug information for this object.

void AccessionsToOids(const vector< string > &accs, vector< blastdb::TOid > &oids) const

Uint8 GetTotalLengthStats() const

Returns the sum of the lengths of all available sequences.

int GetSeqLengthApprox(int oid) const

Returns an unbiased, approximate sequence length.

CRef< CBioseq > SeqidToBioseq(const CSeq_id &seqid) const

Get a CBioseq for a given Seq-id.

CSeqDB()

No-argument Constructor.

CRef< CBioseq > PigToBioseq(int pig) const

Get a CBioseq for a given PIG.

int GetAmbigSeq(int oid, const char **buffer, int nucl_code) const

Get a pointer to sequence data with ambiguities.

void GetMaskData(int oid, const vector< int > &algo_ids, TSequenceRanges &ranges)

Get masked ranges of a sequence.

bool GiToOid(TGi gi, int &oid) const

Translate a GI to an OID.

const map< string, string > & GetColumnMetaData(int column_id)

Get all metadata for the specified column.

CSeqDBIdSet GetIdSet() const

Get IdSet list attached to this database.

void FlushOffsetRangeCache()

Flush all offset ranges cached.

void GetLeafTaxIDs(int oid, map< TGi, set< TTaxId > > &gi_to_taxid_set, bool persist=false) const

Get taxid for an OID.

Functor object for s_RemoveAliasComponents where the path name is matched in SSeqDBInitInfo.

PathFinder(const string &p)

bool operator()(const SSeqDBInitInfo &value) const

iterator_bool insert(const value_type &val)

const_iterator begin() const

static unsigned char depth[2 *(256+1+29)+1]

static void DLIST_NAME() append(DLIST_LIST_TYPE *list, DLIST_TYPE *item)

static SQLCHAR output[256]

unsigned int TSeqPos

Type for sequence locations and lengths.

#define ITERATE(Type, Var, Cont)

ITERATE macro to sequence through container elements.

#define NON_CONST_ITERATE(Type, Var, Cont)

Non constant version of ITERATE macro.

SStrictId_Tax::TId TTaxId

Taxon id type.

#define ERR_POST(message)

Error posting with file, line number information but without error codes.

#define LOG_POST(message)

This macro is deprecated and it's strongly recomended to move in all projects (except tests) to macro...

void Error(CExceptionArgs_Base &args)

#define NCBI_THROW(exception_class, err_code, message)

Generic macro to throw an exception, given the exception class, error code and message string.

void Trace(CExceptionArgs_Base &args)

void Info(CExceptionArgs_Base &args)

void FindFilesInDir(const CDir &dir, const vector< string > &masks, const vector< string > &masks_subdir, TFindFunc &find_func, TFindFiles flags=fFF_Default)

Find files in the specified directory.

Int8 GetLength(void) const

Get size of file.

virtual bool Remove(TRemoveFlags flags=eRecursive) const

Remove a directory entry.

static string MakePath(const string &dir=kEmptyStr, const string &base=kEmptyStr, const string &ext=kEmptyStr)

Assemble a path from basic components.

EFindFiles

File finding flags.

static char GetPathSeparator(void)

Get path separator symbol specific for the current platform.

string GetName(void) const

Get the base entry name with extension (if any).

const string & GetPath(void) const

Get entry path.

virtual bool Exists(void) const

Check existence of file.

@ fFF_Recursive

descend into sub-dirs

@ fParse_RawText

Try to ID raw non-numeric accessions.

@ fParse_ValidLocal

Treat otherwise unidentified strings as raw accessions, provided that they pass rudimentary validatio...

virtual void DebugDump(CDebugDumpContext ddc, unsigned int depth) const

Define method for dumping debug information.

TObjectType * GetPointerOrNull(void) THROWS_NONE

Get pointer value.

int32_t Int4

4-byte (32-bit) signed integer

uint32_t Uint4

4-byte (32-bit) unsigned integer

int64_t Int8

8-byte (64-bit) signed integer

uint64_t Uint8

8-byte (64-bit) unsigned integer

bool NotEmpty(void) const

position_type GetToOpen(void) const

#define END_NCBI_SCOPE

End previously defined NCBI scope.

#define BEGIN_NCBI_SCOPE

Define ncbi namespace.

IO_PREFIX::ifstream CNcbiIfstream

Portable alias for ifstream.

static list< string > & Split(const CTempString str, const CTempString delim, list< string > &arr, TSplitFlags flags=0, vector< SIZE_TYPE > *token_pos=NULL)

Split a string using specified delimiters.

static TNumeric StringToNumeric(const CTempString str, TStringToNumFlags flags=0, int base=10)

Convert string to a numeric value.

static string IntToString(int value, TNumToStringFlags flags=0, int base=10)

Convert int to string.

static SIZE_TYPE Find(const CTempString str, const CTempString pattern, ECase use_case=eCase, EDirection direction=eForwardSearch, SIZE_TYPE occurrence=0)

Find the pattern in the string.

static string Join(const TContainer &arr, const CTempString &delim)

Join strings using the specified delimiter.

static string & Replace(const string &src, const string &search, const string &replace, string &dst, SIZE_TYPE start_pos=0, SIZE_TYPE max_replace=0, SIZE_TYPE *num_replace=0)

Replace occurrences of a substring within a string.

static bool StartsWith(const CTempString str, const CTempString start, ECase use_case=eCase)

Check if a string starts with a specified prefix value.

@ fConvErr_NoThrow

Do not throw an exception on error.

@ fSplit_Tokenize

All delimiters are merged and trimmed, to get non-empty tokens only.

@ fSplit_MergeDelimiters

Merge adjacent delimiters.

@ eNocase

Case insensitive compare.

string AsString(const CTimeFormat &format=kEmptyStr, TSeconds out_tz=eCurrentTimeZone) const

Transform time to string.

bool IsEmpty(void) const

Is time object empty (date and time)?

static CTimeFormat GetPredefined(EPredefined fmt, TFlags flags=fDefault)

Get predefined format.

@ eISO8601_DateTimeSec

Y-M-DTh:m:s (eg 1997-07-16T19:20:30)

strategy

Block allocation strategies.

EBlast_filter_program

This defines the possible sequence filtering algorithms to be used in a BLAST database.

TFrom GetFrom(void) const

Get the From member data.

char * dbname(DBPROCESS *dbproc)

Get name of current database.

unsigned int

A callback function used to compare two keys in a database.

constexpr auto sort(_Init &&init)

constexpr bool empty(list< Ts... >) noexcept

const struct ncbi::grid::netcache::search::fields::SIZE size

const struct ncbi::grid::netcache::search::fields::KEY key

const GenericPointer< typename T::ValueType > T2 value

std::istream & in(std::istream &in_, double &x_)

void copy(Njn::Matrix< S > *matrix_, const Njn::Matrix< T > &matrix0_)

CSeqDB::ESeqType ParseMoleculeTypeString(const string &s)

Convert a string to a CSeqDB ESeqType object.

static char s_GetSeqTypeChar(CSeqDB::ESeqType seqtype)

Helper function to translate enumerated type to character.

static CSeqDBImpl * s_SeqDBInit(const string &dbname, char prot_nucl, int oid_begin, int oid_end, bool use_atlas_lock, CSeqDBGiList *gi_list=NULL, CSeqDBNegativeList *neg_list=NULL, CSeqDBIdSet idset=CSeqDBIdSet())

Helper function to build private implementation object.

bool DeleteBlastDb(const string &dbpath, CSeqDB::ESeqType seq_type)

Deletes all files associated with a BLAST database.

vector< SSeqDBInitInfo > FindBlastDBs(const string &path, const string &dbtype, bool recurse, bool include_alias_files, bool remove_redundant_dbs)

Find BLAST DBs in the directory specified.

static void s_RemoveAliasComponents(CBlastDbFinder &finder)

Defines BLAST database access classes.

ESeqDBAllocType

Certain methods have an "Alloc" version.

void SeqDB_GetLMDBFileExtensions(bool db_is_protein, vector< string > &extn)

Retrieves file extensions for BLAST LMDB files.

const int kSeqDBNuclNcbiNA8

Used to request ambiguities in Ncbi/NA8 format.

void SeqDB_GetFileExtensions(bool db_is_protein, vector< string > &extensions, EBlastDbVersion dbver=eBDB_Version4)

Retrieves a list of all supported file extensions for BLAST databases.

EBlastDbVersion

BLAST database version.

This file defines several SeqDB utility functions related to byte order and file system portability.

void SeqDB_CombineAndQuote(const vector< string > &dbs, string &dbname)

Combine and quote list of database names.

const U & SeqDB_MapFind(const std::map< T, U > &m, const T &k, const U &dflt)

Find a map value or return a default.

T SeqDB_GetStdOrd(const T *stdord_obj)

Read a network order integer value.

The top level of the private implementation layer for SeqDB.

static SLJIT_INLINE sljit_ins l(sljit_gpr r, sljit_s32 d, sljit_gpr x, sljit_gpr b)

List of sequence offset ranges.

Structure to define basic information to initialize a BLAST DB.

OID-Range type to simplify interfaces.


RetroSearch is an open source project built by @garambo | Open a GitHub Issue

Search and Browse the WWW like it's 1997 | Search results from DuckDuckGo

HTML: 3.2 | Encoding: UTF-8 | Version: 0.7.4