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/vdb2blast__util_8cpp_source.html below:

NCBI C++ ToolKit: src/algo/blast/vdb/vdb2blast_util.cpp Source File

70  virtual

list< CRef<objects::CSeq_id> >

GetId

(

Uint4

oid)

const

;

84  virtual size_t Size

()

const

;

93  const

vector<TSeqRange>& target_ranges,

118

: m_sraBlastUtil(sraBlastUtil)

134

list< CRef<CSeq_id> > listIds;

135

listIds.push_back(seqIdVDB);

142

list< CRef<CSeq_id> > listIds =

GetId

(oid);

143  ASSERT

(!listIds.empty());

146

seqLoc->

SetWhole

().Assign(**listIds.begin());

173  const

vector<TSeqRange>& target_ranges,

197  copy

(vecSRARunAccessions.begin(),

198

vecSRARunAccessions.end(),

199

inserter(string_set, string_set.

begin

()));

200

vecSRARunAccessions.clear();

202

back_inserter(vecSRARunAccessions));

207  "Failed to process SRA accession list: "

+

m_strAllRuns

);

210  if

(vecSRARunAccessions.empty())

222

vector<string> vecSRARunAccessions;

226  Uint4

numRuns = vecSRARunAccessions.size();

227  char

** vdbRunAccessions =

new char

*[numRuns];

228  for

(

Uint4

iRun = 0; iRun < numRuns; iRun++) {

229  if

(!vecSRARunAccessions[iRun].

empty

()) {

230

vdbRunAccessions[iRun] =

231  strdup

(vecSRARunAccessions[iRun].c_str());

243  for

(

Uint4

iRun = 0; iRun < numRuns; iRun++)

245  if

(isRunExcluded[iRun]) {

246

excluded_runs += vdbRunAccessions[iRun];

247

excluded_runs +=

" "

;

250  delete

[] vdbRunAccessions;

251  delete

[] isRunExcluded;

253  if

(rc > 0 && excluded_runs !=

kEmptyStr

) {

257  "Error opening the following db(s): "

+ excluded_runs);

263  "Failed to construct the VDB BlastSeqSrc object"

);

269  string

strErrMsg(errMsg);

273  "VDB BlastSeqSrc construction failed: "

+ strErrMsg);

283  bool

bIncludeFilteredReads):

284

m_bOwnSeqSrc(bOwnSeqSrc), m_strAllRuns(strAllRuns), m_isCSRAUtil(bCSRA),

285

m_IncludeFilteredReads(bIncludeFilteredReads)

328  size_t

first_digit_pos =

id

.find_first_of(

kDigits

);

329  if

((first_digit_pos > 3) && (first_digit_pos <= 6)) {

330  if

(

id

.find_first_not_of(

kDigits

, first_digit_pos) == std::string::npos) {

345  const char

* readName =

NULL

;

352  "Incomplete SeqID for SRA sequence"

);

364  "Empty VDB tag in SeqID"

);

366

readName = nameStr.c_str();

378  "Failed to get the OID for the VDB tag: "

+

string

(readName));

387  if

(!db_name.empty())

390  string tmp

= db_name;

391  if

(last_pos != string::npos) {

392  tmp

= db_name.substr(last_pos +1);

396  if

(

tmp

.find_first_not_of(

kDigits

, 4) == std::string::npos)

424  const string

gnl_tag(

"gnl|SRA|"

);

425  string

strId =

string

(nameRun);

430

list<CRef<CSeq_id> > ids;

434

strId = gnl_tag + strId;

437

seqId.

Reset

(bs_id);

442

strId = gnl_tag + strId;

450

strId = gnl_tag + strId;

492  "Failed to read the VDB sequence string for OID="

+

497  if

(!cstrSeq || strlen(cstrSeq) == 0)

500  "Got an empty VDB sequence string for OID="

+

513

bioseqResult->

SetInst

(*seqInst);

516

bioseqResult->

SetId

().push_back(seqId);

522

bioseqResult->

SetDescr

().Set().push_back(descTitle);

557  "2na reader has not been initialized"

);

574  "Failed to read the VDB sequence string for OID="

+

585  if

(!cstrSeq || strlen(cstrSeq) == 0)

588  "Got an empty VDB sequence string for OID="

+

596  "Failed to seq id for OID="

+

609

bioseqResult->

SetInst

(*seqInst);

612

bioseqResult->

SetId

().push_back(

id

);

618

bioseqResult->

SetDescr

().Set().push_back(descTitle);

626  if

(alnSet.

Empty

())

629

CSeq_align_set::Tdata::const_iterator itAln;

630  for

(itAln = alnSet->

Get

().begin(); itAln != alnSet->

Get

().end(); itAln++)

635

subjId->

Assign

(subjIdFromAln);

652

CBlastFormatUtil::SDbInfo dbInfo;

653

dbInfo.is_protein =

false

;

655

dbInfo.definition = dbInfo.name;

658

vecDbInfo.push_back(dbInfo);

662

m_bOwnSeqSrc(

true

), m_strAllRuns(strAllRuns), m_isCSRAUtil(bCSRA), m_IncludeFilteredReads(

false

)

683  Uint8

& max_seq_length,

Uint8

& av_seq_length,

bool

getRefStats)

702  unsigned int

numRuns = vdbs.size();

704  char

** vdbRunAccessions =

new char

*[numRuns];

705  for

(

Uint4

iRun = 0; iRun < numRuns; iRun++) {

706  if

(!vdbs[iRun].

empty

()) {

707

vdbRunAccessions[iRun] =

strdup

(vdbs[iRun].c_str());

714  SRABlastSeqSrcInit

((

const char

**)vdbRunAccessions, numRuns,

false

, isRunExcluded.

get

(), &rc,

false

,

false

);

718  for

(

Uint4

iRun = 0; iRun < numRuns; iRun++)

720  if

(isRunExcluded.

get

()[iRun]) {

721

cannot_open += vdbs[iRun]+

" "

;

723  free

(vdbRunAccessions[iRun]);

725  delete

[] vdbRunAccessions;

737  if

(status != 0 || mgr ==

NULL

) {

740

status = VdbBlastMgrKLogHandlerSetStdErr(mgr);

742

status = VdbBlastMgrKLogLibHandlerSetStdErr(mgr);

745

status = VdbBlastMgrKLogLevelSetWarn(mgr);

790

ref_num_seqs = vdbData->

numSeqs

;

803  if

( srr_len !=

NPOS

) {

829  return

(num_thread == 0? 1: num_thread);

834  int

num_ids = ids.size();

844  for

(

Uint8

j=0; j < num_seqs; j++) {

874  unsigned int

num_wgs = 0;

878  for

(

unsigned int i

=0;

i

< dbs.size();

i

++) {

880  string tmp

= dbs[

i

];

881  if

(last_pos != string::npos) {

882  tmp

= dbs[

i

].substr(last_pos +1);

890  if

(num_wgs == dbs.size()) {

Int4 BlastSeqSrcGetSeqLen(const BlastSeqSrc *seq_src, void *oid)

Retrieve sequence length (number of residues/bases)

char * BlastSeqSrcGetInitError(const BlastSeqSrc *seq_src)

Function to retrieve NULL terminated string containing the description of an initialization error or ...

Int4 BlastSeqSrcGetNumSeqs(const BlastSeqSrc *seq_src)

Get the number of sequences contained in the sequence source.

Int8 BlastSeqSrcGetTotLen(const BlastSeqSrc *seq_src)

Get the total length of all sequences in the sequence source.

BlastSeqSrc * BlastSeqSrcFree(BlastSeqSrc *seq_src)

Frees the BlastSeqSrc structure by invoking the destructor function set by the user-defined construct...

const char * BlastSeqSrcGetName(const BlastSeqSrc *seq_src)

Get the Blast Sequence source name (e.g.

NCBI_XBLAST_EXPORT void * _BlastSeqSrcImpl_GetDataStructure(const BlastSeqSrc *var)

const CSeq_id & GetSeq_id(TDim row) const

Get seq-id (the first one if segments have different ids).

static Uint8 GetTotalPhysicalMemorySize(void)

Return the amount of actual/total physical memory, in bytes.

BlastSeqSrc * GetSRASeqSrc()

Return the stored SRA BlastSeqSrc object.

static Uint4 GetMaxNumCSRAThread(void)

void FillVDBInfo(vector< CBlastFormatUtil::SDbInfo > &vecDbInfo)

Populate the DB info list with information on open SRA runs.

string m_strAllRuns

Space-delimited list of opened SRA run accessions.

Uint4 GetOIDFromVDBSeqId(CRef< objects::CSeq_id > seqId)

Get the ordinal number (OID) for the given SRA sequence.

void x_GetSRARunAccessions(vector< string > &vecSRARunAccessions)

Tokenize the stored whitespace-delimited string of SRA runs.

bool m_bOwnSeqSrc

Release the BlastSeqSrc object in destructor.

CRef< objects::CSeq_id > GetVDBSeqIdFromOID(Uint4 oid)

Get the SRA sequence SeqID given its ordinal number (OID).

bool m_IncludeFilteredReads

static void GetVDBStats(const string &strAllRuns, Uint8 &num_seqs, Uint8 &length, bool getRefStats=false)

Fucntion to get around the OID (blastseqsrc) limit So num of seqs > int4 can be returned.

static Uint4 SetupVDBManager()

*Note* Call this in main thread first, if you are going to instantiate this object or use any of the ...

BlastSeqSrc * m_seqSrc

Pointer to a properly initialized SRA BlastSeqSrc.

CRef< objects::CBioseq > CreateBioseqFromVDBSeqId(CRef< objects::CSeq_id > seqId)

Construct a Bioseq object for the given SRA sequence.

static bool IsCSRA(const string &db_name)

static void CheckVDBs(const vector< string > &vdbs)

Function to check a list of dbs if they can be opened Throw an exception if any of the db cannot be o...

BlastSeqSrc * x_MakeVDBSeqSrc()

Construct an SRA BlastSeqSrc object from the given strings.

void GetOidsFromSeqIds_WGS(const vector< string > &ids, vector< int > &oids)

CRef< blast::IBlastSeqInfoSrc > GetSRASeqInfoSrc()

Return the SRA BlastSeqInfoSrc object (create if none exists).

virtual ~CVDBBlastUtil()

Destructor.

CVDBBlastUtil(const string &strAllRuns, bool bOwnSeqSrc=false, bool bCSRA=false, bool bIncludeFilteredReads=false)

Constructor that creates and stores the SRA BlastSeqSrc object.

static bool IsSRA(const string &db_name)

static void GetAllStats(const string &strAllRuns, Uint8 &num_seqs, Uint8 &length, Uint8 &ref_num_seqs, Uint8 &ref_length)

static void ReleaseVDBManager()

Call this release vdb manager if SetupManger has been explicitly called in the main thread.

void AddSubjectsToScope(CRef< CScope > scope, CConstRef< CSeq_align_set > alnSet)

Populate the CScope object with subject sequence Bioseqs.

CRef< objects::CBioseq > CreateBioseqFromOid(Uint8 oid)

static IDType VDBIdType(const CSeq_id &id)

virtual bool HasGiList() const

Returns true if the subject is restricted by a GI list, always returns false in this implementation.

virtual CConstRef< objects::CSeq_loc > GetSeqLoc(Uint4 oid) const

Method to retrieve the sequence location given its ordinal number.

CVDBSeqInfoSrc(CRef< CVDBBlastUtil > sraBlastUtil)

Constructor taking a CVDBBlastUtil object.

virtual bool GetMasks(Uint4 oid, const vector< TSeqRange > &target_ranges, TMaskedSubjRegions &retval) const

Retrieves the subject masks for the corresponding oid, always returns false in this implementation.

virtual size_t Size() const

Returns the size of the underlying container of sequences.

virtual list< CRef< objects::CSeq_id > > GetId(Uint4 oid) const

Method to retrieve a sequence identifier given its ordinal number.

virtual bool CanReturnPartialSequence() const

Return true if the implementation can return anything besides a seq-loc for the entire sequence.

virtual ~CVDBSeqInfoSrc()

Destructor.

virtual Uint4 GetLength(Uint4 oid) const

Method to retrieve a sequence length given its ordinal number.

CRef< CVDBBlastUtil > m_sraBlastUtil

The CVDBBlastUtil object that takes care of various conversions.

Collection of masked regions for a single query sequence.

const_iterator begin() const

const_iterator end() const

#define VDBSRC_OVERFLOW_RV

void VDBSRC_ReleaseErrorMsg(TVDBErrMsg *vdbErrMsg)

Release the Error message.

void VDBSRC_InitEmptyErrorMsg(TVDBErrMsg *vdbErrMsg)

Initialize an empty Error message (No Error).

void VDBSRC_FormatErrorMsg(char **errMsg, const TVDBErrMsg *vdbErrMsg)

Format the error message as a single human-readable string.

bool Empty(const CNcbiOstrstream &src)

static const char * str(char *buf, int n)

element_type * get(void) const

Get pointer.

#define ERR_POST(message)

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

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 Warning(CExceptionArgs_Base &args)

static char GetPathSeparator(void)

Get path separator symbol specific for the current platform.

static EAccessionInfo IdentifyAccession(const CTempString &accession, TParseFlags flags=fParse_AnyRaw)

Deduces information from a bare accession a la WHICH_db_accession; may report false negatives on prop...

string GetSeqIdString(bool with_version=false) const

Return seqid string with optional version for text seqid type.

virtual void Assign(const CSerialObject &source, ESerialRecursionMode how=eRecursive)

Optimized implementation of CSerialObject::Assign, which is not so efficient.

static SIZE_TYPE ParseFastaIds(CBioseq::TId &ids, const CTempString &s, bool allow_partial_failure=false)

Parse an entire set of |-delimited FASTA-style IDs, appending the results to IDS.

static int Score(const CRef< CSeq_id > &id)

Wrappers for use with FindBestChoice from <corelib/ncbiutil.hpp>

@ fParse_RawText

Try to ID raw non-numeric accessions.

CBioseq_Handle AddBioseq(CBioseq &bioseq, TPriority pri=kPriority_Default, EExist action=eExist_Throw)

Add bioseq, return bioseq handle.

@ kPriority_Default

Use default priority for added data.

bool Empty(void) const THROWS_NONE

Check if CConstRef is empty – not pointing to any object which means having a null value.

void Reset(void)

Reset reference object.

bool NotEmpty(void) const THROWS_NONE

Check if CRef is not empty – pointing to an object and has a non-null value.

int32_t Int4

4-byte (32-bit) signed integer

uint32_t Uint4

4-byte (32-bit) unsigned integer

uint64_t Uint8

8-byte (64-bit) unsigned integer

#define END_NCBI_SCOPE

End previously defined NCBI scope.

#define BEGIN_NCBI_SCOPE

Define ncbi namespace.

NCBI_NS_STD::string::size_type SIZE_TYPE

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 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 & 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 string UIntToString(unsigned int value, TNumToStringFlags flags=0, int base=10)

Convert UInt to string.

static bool EqualNocase(const CTempString s1, SIZE_TYPE pos, SIZE_TYPE n, const char *s2)

Case-insensitive equality of a substring with another string.

static string UInt8ToString(Uint8 value, TNumToStringFlags flags=0, int base=10)

Convert UInt8 to string.

@ fSplit_Tokenize

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

C::value_type FindBestChoice(const C &container, F score_func)

Find the best choice (lowest score) for values in a container.

bool IsStr(void) const

Check if variant Str is selected.

const TTag & GetTag(void) const

Get the Tag member data.

bool CanGetDb(void) const

Check if it is safe to call GetDb method.

bool CanGetTag(void) const

Check if it is safe to call GetTag method.

const TDb & GetDb(void) const

Get the Db member data.

const TStr & GetStr(void) const

Get the variant data.

const Tdata & Get(void) const

Get the member data.

bool IsGeneral(void) const

Check if variant General is selected.

E_Choice Which(void) const

Which variant is currently selected.

const TGeneral & GetGeneral(void) const

Get the variant data.

@ e_General

for other databases

TId & SetId(void)

Assign a value to Id data member.

TTitle & SetTitle(void)

Select the variant.

TLength GetLength(void) const

Get the Length member data.

void SetInst(TInst &value)

Assign a value to Inst data member.

void SetDescr(TDescr &value)

Assign a value to Descr data member.

void SetRepr(TRepr value)

Assign a value to Repr data member.

void SetLength(TLength value)

Assign a value to Length data member.

void SetSeq_data(TSeq_data &value)

Assign a value to Seq_data data member.

void SetMol(TMol value)

Assign a value to Mol data member.

@ eRepr_raw

continuous sequence

@ e_Iupacna

IUPAC 1 letter nuc acid code.

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

Defines to provide correct exporting from DLLs in some configurations.

Uint1 Boolean

bool replacment for C

#define FALSE

bool replacment for C indicating false.

#define ASSERT

macro for assert.

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

BlastSeqSrc * SRABlastSeqSrcInit(const char **sraRunAccessions, Uint4 numRuns, Boolean isProtein, Boolean *excluded_runs, Uint4 *status, Boolean isCSRA, Boolean include_filtered_reads)

Allocate and initialize the SRA BlastSeqSrc object.

Complete type definition of Blast Sequence Source ADT.

Structure providing top-level VDB data access.

TVDB2naICReader * reader_2na

char * names

Names of the VDB data represented by this object (usually this will include all the SVDB run accessio...

Structure describing the error messages the library can generate.

Boolean isError

True if the object describes an error.

static const char kDigits[]

static bool s_IsWGSId(const string &id)

Boolean VDBSRC_GetOIDFromReadName(TVDBData *vdbData, const char *nameRun, Int4 *oid)

Get the sequence OID given its SRA-specific sequence information.

uint64_t VDBSRC_GetAvgSeqLen(TVDBData *vdbData)

Get the average sequence length in the open SRA data.

int VDBSRC_IsCSRA(const char *run)

Return 1 if run is csra, 0 if not and -1 for error.

VdbBlastMgr * VDBSRC_GetVDBManager(uint32_t *status)

This will call VdbBlastInit and intiailize a singleton for VDBBlastMgr This needs to be called in the...

void VDBSRC_ReleaseVDBManager()

This needs to be called if VDBSRC_GetVDBManager has been called in the main thread.

uint64_t VDBSRC_GetTotSeqLen(TVDBData *vdbData)

Get the total sequence length in the open SRA data.

void VDBSRC_MakeCSRASeqSrcFromSRASeqSrc(TVDBData *vdbData, TVDBErrMsg *vdbErrMsg, Boolean getStats)

uint64_t VDBSRC_GetMaxSeqLen(TVDBData *vdbData)

Get the maximum sequence length in the open SRA data.

Boolean VDBSRC_GetReadNameForOID(TVDBData *vdbData, Int4 oid, char *name_buffer, size_t buf_size)

Get the SRA-specific sequence information for the given OID.

File contains internal structures and functions for reading VDB databases.

#define VDB_2NA_CHUNK_BUF_SIZE

Boolean VDBSRC_Get4naSequenceAsString(TVDBData *vdbData, uint64_t oid, char **seqIupacna, TVDBErrMsg *vdbErrMsg)

Access and convert the selected sequence to a human-readable string.

Boolean VDBSRC_Get2naSequenceAsString(TVDBData *vdbData, uint64_t oid, char **seqIupacna, TVDBErrMsg *vdbErrMsg)


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