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/rm__reader_8cpp_source.html below:

NCBI C++ ToolKit: src/objtools/readers/rm_reader.cpp Source File

88 #define NCBI_USE_ERRCODE_X Objtools_Rd_RepMask 106  if

(family.empty()) {

265  if

(it ==

m_Map

.

end

())

return false

;

274

vector<string> tokens;

276  while

(! stream.eof()) {

300  if

(line.length() < 6 || line.substr(2, 3) !=

" "

)

continue

;

301  string code

(line.substr(0, 2));

302  string value

(line.substr(5));

305  if

(

code

==

"ID"

) {

317  string

bp(

value

.substr(

value

.rfind(

';'

) + 1));

320

}

else if

(

code

==

"DE"

) {

325

}

else if

(

code

==

"CC"

) {

356

pair<TSpecificity2Taxid::iterator, bool> i_specificity =

358  if

(i_specificity.second) {

360  if

(! i_specificity.first->second) {

362

<<

"RepeatMasker library species failed lookup to taxonomy: " 377  const string

& name)

const 382 template

<

typename

T>

384  const string

& qual,

const T

val)

390

qual_list.push_back(

result

);

411  if

(! family.empty())

val

+= family;

415  if

(! family.empty())

s_SetQual

(qual_list,

"rpt_family"

, family);

427  s_SetQual

(qual_list,

"rpt_family"

, klass);

443  if

(! family.empty())

s_SetQual

(qual_list,

"rpt_family"

, family);

476  m_Ids

.Reset(&generator);

492

imp.

SetKey

(

"repeat_region"

);

495

feat->

SetId

(*

id

);

501

ref->SetId().Assign(*id_it->second);

502

feat->

SetXref

().push_back(ref);

511  bool

standardized(

false

);

523  if

(! standardized) {

542  bool

include_specificity_name(

false

);

546

include_specificity_name = ! specificity_name.empty();

551  tag

->SetDb(

"taxon"

);

552  tag

->SetTag().SetId(specificity);

559

include_specificity_name=

false

;

578

include_rpt_left =

false

;

593  if

(include_rpt_left) {

606  s_SetQual

(qual_list,

"overlapped"

,

true

);

615  s_SetQual

(qual_list,

"rpt_length"

, rpt_length);

618  if

(include_specificity_name) {

627

feat->

SetExts

().push_back(uo);

628

uo->

SetType

().SetStr(

"RepeatMasker"

);

635  if

(include_rpt_left) {

644

uo->

AddField

(

"query_length"

,

static_cast<int>

(

648

uo->

AddField

(

"overlapped"

,

true

);

657

uo->

AddField

(

"rpt_length"

,

static_cast<int>

(rpt_length));

660  if

(include_specificity_name) {

669  if

(qual_list.empty()) feat->

ResetQual

();

674  tag

->SetDb(

"REPBASE"

);

688

ostringstream comment;

689  const char eq

(

'='

), sep(

' '

);

691

comment <<

"source=RepeatMasker"

;

707

<<

"query_range"

<<

eq

;

709  if

(reverse) comment <<

"complement("

;

712  if

(reverse) comment <<

")"

;

742

: m_SeqIdResolver(&seqid_resolver)

743

, m_ToFeat(

flags

, lib, ids)

783  size_t

record_counter = 0;

785  while

( !

lr

.AtEOF() ) {

802  "RepeatMasker Reader: Parse error in record = "

+ line) );

812  "RepeatMasker Reader: Verification error in record = "

+ line) );

823  "RepeatMasker Reader: Aborting file import, " 824  "unable to create feature table for record = "

+ line) );

842  string

labels_1st_line[] = {

"perc"

,

"query"

,

"position"

,

"matching"

,

""

};

843  string

labels_2nd_line[] = {

"score"

,

"div."

,

"del."

,

"ins."

,

"sequence"

,

""

};

846  size_t

current_offset = 0;

848  for

( ; labels_1st_line[

i

] !=

""

; ++

i

) {

849

current_offset =

NStr::FindCase

( line, labels_1st_line[

i

], current_offset );

850  if

(

NPOS

== current_offset ) {

854  if

( labels_1st_line[

i

] ==

""

) {

861  for

( ; labels_2nd_line[

i

] !=

""

; ++

i

) {

862

current_offset =

NStr::FindCase

( line, labels_2nd_line[

i

], current_offset );

863  if

(

NPOS

== current_offset ) {

873  if

(

NStr::StartsWith

(line,

"There were no repetitive sequences detected in "

))

885  if

(e > 0 && s[

b

] ==

'('

) {

887  if

(s[e - 1] ==

')'

) --e;

891

s = s.substr(

b

, e -

b

);

897  const size_t

MIN_VALUE_COUNT = 15;

900

list< string > values;

907

list<string>::iterator it = values.begin();

927  if

(!

id

)

return false

;

935  if

(pos_begin == 0)

return false

;

941  if

(pos_end == 0 || pos_end < pos_begin)

return false

;

960  string

class_family = *it;

966  string

field12 = *it;

974  string

field14 = *it;

995

mask_data.

overlapped

= (it != values.end() && (*it) ==

"*"

);

User-defined methods of the data storage class.

User-defined methods of the data storage class.

User-defined methods of the data storage class.

User-defined methods of the data storage class.

@ eExtreme_Positional

numerical value

User-defined methods of the data storage class.

User-defined methods of the data storage class.

Default implementation of a Seq-id resolver, which knows about FASTA-formatted sequence identifiers.

CSeq_id_Handle ResolveSeqId(const string &id) const

Returns a normalized representation of a sequence identifier, as Seq-id handle.

@Gb_qual.hpp User-defined methods of the data storage class.

*** Import *********************************************** * * Features imported from other databases...

static CObjReaderLineException * Create(EDiagSev eSeverity, unsigned int uLine, const std::string &strMessage, EProblem eProblem=eProblem_GeneralParsingError, const std::string &strSeqId=string(""), const std::string &strFeatureName=string(""), const std::string &strQualifierName=string(""), const std::string &strQualifierValue=string(""), CObjReaderLineException::EErrCode eErrCode=eFormat, const TVecOfLines &vecOfOtherLines=TVecOfLines())

Please use this instead of the constructor because the ctor is protected.

Default implementation for a generator of identifiers, as integers, mashalled as CFeat_id objects.

void ProcessError(CObjReaderLineException &, ILineErrorListener *)

virtual void xAddConversionInfo(CSeq_annot &, ILineErrorListener *)

Class acting as an interface to a RepeatMasker library.

TSpecificity2Taxid m_Specificity2TaxId

CConstIRef< ITaxonomyResolver > m_Taxonomy

bool TestSpecificityMatchesName(TRepeat::TTaxId taxid, const string &name) const

Check if a given taxid's scientific name matches the original specificity string.

void Read(CNcbiIstream &stream)

Reads a library from the RepeatMaskerLib.embl-style input.

bool Get(const string &name, TRepeat &dest) const

Gets information about a given repeat, specified by name.

Implements a concrete class for reading RepeatMasker output from tabular form and rendering it as ASN...

void SetSeqIdResolver(ISeqIdResolver &seqid_resolver)

Use specified delegate for Seq-id resolution.

virtual CRef< CSerialObject > ReadObject(CNcbiIstream &istr, ILineErrorListener *pErrors=nullptr)

Read an object from a given input stream, render it as the most appropriate Genbank object.

virtual bool VerifyData(const SRepeatRegion &mask_data)

CRepeatMaskerReader(TFlags flags=fDefaults, CConstRef< TRepeatLibrary > lib=null, const ISeqIdResolver &seqid_resolver= *(CConstIRef< ISeqIdResolver >(new CFastaIdsResolver)), TIdGenerator &ids= *(CIRef< TIdGenerator >(new COrdinalFeatIdGenerator)))

Implement CReaderBase.

virtual bool ParseRecord(const string &record, SRepeatRegion &mask_data)

virtual bool IsHeaderLine(const string &line)

TConverter & SetConverter()

Delegate for conversion from IRepeatRegion to ASN.1.

virtual bool IsIgnoredLine(const string &line)

void ResetSeqIdResolver()

Use default Seq-id resolution.

virtual CRef< CSeq_annot > ReadSeqAnnot(CNcbiIstream &istr, ILineErrorListener *pErrors=nullptr)

Read an object from a given input stream, render it as a single Seq-annot.

CConstIRef< ISeqIdResolver > m_SeqIdResolver

virtual ~CRepeatMaskerReader(void)

Class which, given an input IRepeatRegion, can generate an appropriate and normalized NCBI ASN....

void ResetRepeatLibrary()

Clear out any repeat library which may be used to add additional attributes to repeats.

CRepeatToFeat(TFlags flags=fDefaults, CConstRef< TRepeatLibrary > lib=null, TIdGenerator &ids= *(CIRef< TIdGenerator >(new COrdinalFeatIdGenerator)))

void SetRepeatLibrary(const TRepeatLibrary &lib)

Set a repeat library which may be used to add additional attributes to repeats.

CRef< CSeq_feat > operator()(const IRepeatRegion &repeat)

Transforms the input repeat into a repeat feature.

void ResetIdGenerator()

Reset the Feature-id generator, do use a default implementation which will generate unique integer lo...

void AssertReferencesResolved()

Asserts that all forward/backward references between any objects visited have now been resolved.

CIRef< TIdGenerator > m_Ids

void SetIdGenerator(TIdGenerator &generator)

Set the Feature-id generator which will be used to assign unique feature IDs.

CConstRef< TRepeatLibrary > m_Library

Deprecated, old API for loading RepeatMasker output.

void Read(CRef< CSeq_annot > annot, TFlags flags=fDefaults, size_t errors=kMax_UInt)

static CRmReader * OpenReader(CNcbiIstream &istr)

CRmReader(CNcbiIstream &istr)

static void CloseReader(CRmReader *reader)

namespace ncbi::objects::

CUser_object & AddField(const string &label, const string &value, EParseField parse=eParse_String)

add a data field to the user object that holds a given value

Abstract base class for lightweight line-by-line reading.

virtual TSeqPos GetRptPosBegin() const =0

virtual TPercent GetPercIns() const =0

virtual TRptId GetRptId() const =0

string GetRptClassFamily() const

Covenience function to get the class and family as one value, the way that RepeatMasker emits them.

virtual bool IsOverlapped() const =0

Flag that there is a higher-scoring match whose domain partly (<80%) includes the domain of this matc...

virtual TPercent GetPercDiv() const =0

virtual TScore GetSwScore() const =0

virtual TSeqPos GetRptPosEnd() const =0

virtual TSeqPos GetRptLeft() const =0

virtual TSeqPos GetSeqLeft() const =0

virtual TPercent GetPercDel() const =0

Interface defining a read-only RepeatMasker repeat feature.

virtual bool IsReverseStrand() const

Convenience functions that gets the strand on the sequence, without dealing with a Seq-loc.

virtual string GetSeqIdString() const

Gets the sequence from the location of the repeat, without dealing with a Seq-loc.

virtual TSeqPos GetSeqPosBegin() const

Convenience function that gets the position start on the sequence, without dealing with a Seq-loc.

virtual TSeqPos GetSeqPosEnd() const

Convenience functions that gets the position end on the sequence, without dealing with a Seq-loc.

virtual CConstRef< CSeq_loc > GetLocation(void) const =0

Gets the location of this repeat.

virtual string GetRptFamily() const =0

Gets repeat family, or empty string if not known.

virtual string GetRptClass() const =0

Gets repeat class, or empty string if not known.

virtual string GetRptName() const =0

Gets repeat name.

ITaxonomyResolver::TTaxId TTaxId

@ fIncludeRepeatId

Store original RepeatMasker repeat_id.

@ fIncludeRepeatSpecificity

Store the specificity from the RepeatMasker library, if provided.

@ fIncludeRepeatPos

Store the repeat position, that is, the interval on the repeat sequence.

@ fRemoveRedundancy

Removes redundant fields.

@ fAllowNonstandardQualifiers

Avoid user objects and instead, put selected information in non-standard and invalid GenBank qualifie...

@ fStandardizeNomenclature

Translate RepeatMasker output to INSDC standard nomenclature for repeats.

@ fIncludeCoreStatistics

Store core statistics, which include the scores of sw_score, perc_div, perc_del, perc_ins,...

@ fIncludeRepeatRepbaseId

Store the RepbaseID from the RepeatMasker library, if provided.

@ fIncludeRepeatLength

Store the repeat length as reported in the library.

@ fSetComment

Selected attributes beyond what is stored in GenBank standard qualifiers will be included as comments...

@ fIncludeExtraStatistics

Store extra statistics, which includes the length of the query (or query_left, equivalently),...

Interface for resolving a sequence identifier given a textual representation.

Structure implementing the IRepeatRegion API as a simple store of data memebers.

string GetRptClass() const

Gets repeat class, or empty string if not known.

TPercent GetPercDiv() const

TPercent GetPercDel() const

string GetSeqIdString() const

Overridden version returns the orginal unparsed sequence identifier, if it was set (non-empty).

bool IsOverlapped() const

Flag that there is a higher-scoring match whose domain partly (<80%) includes the domain of this matc...

TPercent GetPercIns() const

TSeqPos GetSeqLeft() const

string GetRptName() const

Gets repeat name.

string GetRptFamily() const

Gets repeat family, or empty string if not known.

TSeqPos GetRptLeft() const

TScore GetSwScore() const

TTaxId GetRptSpecificity() const

Returns 0, not known.

TSeqPos GetRptPosEnd() const

TSeqPos GetRptLength() const

Gets repeat length, or kInvalidSeqPos if not known.

string GetRptRepbaseId() const

Returns an empty string, not known.

TSeqPos GetRptPosBegin() const

CRef< CSeq_loc > query_location

string GetRptSpecificityName() const

Returns an empty string, not known.

CConstRef< CSeq_loc > GetLocation(void) const

Gets the location of this repeat.

CConstRef< CFeat_id > GetId() const

Gets the more general feature ID for this repeat, which identifies a single repeat,...

Implementation of IRepeat backed by a simple structure.

string GetRptRepbaseId() const

Gets the RepbaseID, or empty string if not known.

string m_RptSpecificityName

string GetRptSpecificityName() const

Gets specificity as a name, or empty string if not known.

TSeqPos GetRptLength() const

Gets repeat length, or kInvalidSeqPos if not known.

TTaxId GetRptSpecificity() const

Gets specificity as a taxonomy ID, or 0 if not known.

container_type::const_iterator const_iterator

container_type::iterator iterator

const_iterator end() const

iterator_bool insert(const value_type &val)

container_type::value_type value_type

const_iterator find(const key_type &key) const

Include a standard set of the NCBI C++ Toolkit most basic headers.

static const char location[]

unsigned int TSeqPos

Type for sequence locations and lengths.

const TSeqPos kInvalidSeqPos

Define special value for invalid sequence position.

#define DIAG_COMPILE_INFO

Make compile time diagnostic information object to use in CNcbiDiag and CException.

#define ERR_POST(message)

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

@ eDiag_Error

Error message.

void Warning(CExceptionArgs_Base &args)

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

Set object to copy of another one.

@ eShallow

Assign/Compare pointers only.

const string AsFastaString(void) const

CConstRef< CSeq_id > GetSeqIdOrNull(void) const

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 CSeq_id_Handle GetHandle(const CSeq_id &id)

Normal way of getting a handle, works for any seq-id.

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

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

bool IsReverseStrand(void) const

Return true if all ranges have reverse strand.

TSeqPos GetStart(ESeqLocExtremes ext) const

Return start and stop positions of the seq-loc.

const CSeq_id * GetId(void) const

Get the id of the location return NULL if has multiple ids or no id at all.

TSeqPos GetStop(ESeqLocExtremes ext) const

void Reset(void)

Reset reference object.

void Reset(void)

Reset reference object.

#define END_NCBI_SCOPE

End previously defined NCBI scope.

#define BEGIN_NCBI_SCOPE

Define ncbi namespace.

CNcbiIstream & NcbiGetlineEOL(CNcbiIstream &is, string &str, string::size_type *count=NULL)

Read from "is" to "str" the next line (taking into account platform specifics of End-of-Line)

IO_PREFIX::istream CNcbiIstream

Portable alias for istream.

NCBI_NS_STD::string::size_type SIZE_TYPE

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

Convert string to int.

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 bool MatchesMask(CTempString str, CTempString mask, ECase use_case=eCase)

Match "str" against the "mask".

static double StringToDouble(const CTempStringEx str, TStringToNumFlags flags=0)

Convert string to double.

static void TruncateSpacesInPlace(string &str, ETrunc where=eTrunc_Both)

Truncate whitespace in a string (in-place)

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

Convert int to string.

static SIZE_TYPE FindCase(const CTempString str, const CTempString pattern, SIZE_TYPE start, SIZE_TYPE end, EOccurrence which=eFirst)

Find the pattern in the specified range of a string using a case sensitive search.

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

Check if a string starts with a specified prefix value.

static bool SplitInTwo(const CTempString str, const CTempString delim, string &str1, string &str2, TSplitFlags flags=0)

Split a string into two pieces using the specified delimiters.

static unsigned int StringToUInt(const CTempString str, TStringToNumFlags flags=0, int base=10)

Convert string to unsigned int.

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 TruncateSpaces(const string &str, ETrunc where=eTrunc_Both)

Truncate whitespace in a 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 IsSetData(void) const

the object itself Check if a value has been assigned to Data data member.

void SetType(TType &value)

Assign a value to Type data member.

TXref & SetXref(void)

Assign a value to Xref data member.

TDbxref & SetDbxref(void)

Assign a value to Dbxref data member.

void SetLocation(TLocation &value)

Assign a value to Location data member.

void SetComment(const TComment &value)

Assign a value to Comment data member.

TExts & SetExts(void)

Assign a value to Exts data member.

void ResetExts(void)

Reset Exts data member.

void SetId(TId &value)

Assign a value to Id data member.

void SetData(TData &value)

Assign a value to Data data member.

vector< CRef< CGb_qual > > TQual

TQual & SetQual(void)

Assign a value to Qual data member.

void ResetQual(void)

Reset Qual data member.

void SetKey(const TKey &value)

Assign a value to Key data member.

void SetData(TData &value)

Assign a value to Data data member.

list< CRef< CSeq_id > > TId

virtual void Reset(void)

Reset the whole object.

list< CRef< CSeq_feat > > TFtable

Lightweight interface for getting lines of data with minimal memory copying.

double value_type

The numeric datatype used by the parser.

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

const GenericPointer< typename T::ValueType > T2 value

Defines NCBI C++ exception handling.

Multi-threading – classes, functions, and features.

Useful/utility classes and methods.

bool eq(T x_, T y_, T round_)

static bool s_StandardizeNomenclature(const IRepeatRegion &repeat, CSeq_feat::TQual &qual_list)

Translate RepeatMasker output to INSDC standard nomenclature for repeats.

static void StripParens(string &s)

static void s_SetQual(CSeq_feat::TQual &qual_list, const string &qual, const T val)

static SLJIT_INLINE sljit_ins lr(sljit_gpr dst, sljit_gpr src)

const value_slice::CValueConvert< value_slice::SRunTimeCP, FROM > Convert(const FROM &value)


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