utilities.hpp File Reference

#include <corelib/ncbistd.hpp>
#include <corelib/ncbistr.hpp>
#include <objects/seqfeat/Seq_feat.hpp>
#include <objects/seqfeat/SeqFeatData.hpp>
#include <objects/seqfeat/Org_ref.hpp>
#include <objects/seqset/Bioseq_set.hpp>
#include <objects/biblio/Id_pat.hpp>
#include <objects/biblio/Auth_list.hpp>
#include <objmgr/seq_vector.hpp>
#include <serial/iterator.hpp>
#include <vector>
#include <list>

Include dependency graph for utilities.hpp:

This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Go to the SVN repository for this file.

Typedefs
typedef const CSeq_feat::TDbxref	TDbtags

Enumerations
enum	EAccessionFormatError {   eAccessionFormat_valid = 0 , eAccessionFormat_no_start_letters , eAccessionFormat_wrong_number_of_digits , eAccessionFormat_null ,   eAccessionFormat_too_long , eAccessionFormat_missing_version , eAccessionFormat_bad_version }
enum	EDateValid {   eDateValid_valid = 0x0 , eDateValid_bad_str = 0x01 , eDateValid_bad_year = 0x02 , eDateValid_bad_month = 0x04 ,   eDateValid_bad_day = 0x08 , eDateValid_bad_season = 0x10 , eDateValid_bad_other = 0x20 , eDateValid_empty_date = 0x40 }
enum	EBioseqEndIsType { eBioseqEndIsType_None = 0 , eBioseqEndIsType_Last , eBioseqEndIsType_All }
enum	EDuplicateFeatureType {   eDuplicate_Not = 0 , eDuplicate_Duplicate , eDuplicate_SameIntervalDifferentLabel , eDuplicate_DuplicateDifferentTable ,   eDuplicate_SameIntervalDifferentLabelDifferentTable }

Functions
bool	IsClassInEntry (const CSeq_entry &se, CBioseq_set::EClass clss)
bool	IsDeltaOrFarSeg (const CSeq_loc &loc, CScope *scope)
bool	IsBlankStringList (const list< string > &str_list)
TGi	GetGIForSeqId (const CSeq_id &id)
CScope::TIds	GetSeqIdsForGI (TGi gi)
bool	IsFarLocation (const CSeq_loc &loc, const CSeq_entry_Handle &seh)
CSeqVector	GetSequenceFromLoc (const CSeq_loc &loc, CScope &scope, CBioseq_Handle::EVectorCoding coding=CBioseq_Handle::eCoding_Iupac)
CSeqVector	GetSequenceFromFeature (const CSeq_feat &feat, CScope &scope, CBioseq_Handle::EVectorCoding coding=CBioseq_Handle::eCoding_Iupac, bool product=false)
string	GetSequenceStringFromLoc (const CSeq_loc &loc, CScope &scope)
bool	IsResidue (unsigned char residue)
CConstRef< CSeq_id >	GetReportableSeqIdForAlignment (const CSeq_align &align, CScope &scope)
string	GetAccessionFromObjects (const CSerialObject *obj, const CSeq_entry *ctx, CScope &scope, int *version)
string	GetAccessionFromBioseq (const CBioseq &bioseq, int *version)
string	GetAccessionFromBioseqSet (const CBioseq_set &bsst, int *version)
CBioseq_set_Handle	GetSetParent (const CBioseq_set_Handle &set, CBioseq_set::TClass set_class)
CBioseq_set_Handle	GetSetParent (const CBioseq_Handle &bioseq, CBioseq_set::TClass set_class)
CBioseq_set_Handle	GetGenProdSetParent (const CBioseq_set_Handle &set)
CBioseq_set_Handle	GetGenProdSetParent (const CBioseq_Handle &set)
CBioseq_set_Handle	GetNucProtSetParent (const CBioseq_Handle &bioseq)
CBioseq_Handle	GetNucBioseq (const CBioseq_set_Handle &bioseq_set)
CBioseq_Handle	GetNucBioseq (const CBioseq_Handle &bioseq)
EAccessionFormatError	ValidateAccessionString (const string &accession, bool require_version)
bool	s_IdXrefsAreReciprocal (const CSeq_feat &cds, const CSeq_feat &mrna)
bool	s_FeatureIdsMatch (const CFeat_id &f1, const CFeat_id &f2)
bool	s_StringHasPMID (const string &str)
bool	HasBadCharacter (const string &str)
bool	EndsWithBadCharacter (const string &str)
int	CheckDate (const CDate &date, bool require_full_date=false)
bool	IsDateInPast (const CDate &date)
string	GetDateErrorDescription (int flags)
bool	IsBioseqTSA (const CBioseq &seq, CScope *scope)
string	GetValidatorLocationLabel (const CSeq_loc &loc, CScope &scope)
void	AppendBioseqLabel (string &str, const CBioseq &sq, bool supress_context)
string	GetBioseqIdLabel (const CBioseq &sq)
bool	HasECnumberPattern (const string &str)
bool	SeqIsPatent (const CBioseq &seq)
bool	SeqIsPatent (const CBioseq_Handle &seq)
bool	s_PartialAtGapOrNs (CScope *scope, const CSeq_loc &loc, unsigned int tag, bool only_gap=false)
CBioseq_Handle	BioseqHandleFromLocation (CScope *m_Scope, const CSeq_loc &loc)
void	CheckBioseqEndsForNAndGap (const CBioseq_Handle &bsh, EBioseqEndIsType &begin_n, EBioseqEndIsType &begin_gap, EBioseqEndIsType &end_n, EBioseqEndIsType &end_gap, bool &begin_ambig, bool &end_ambig)
bool	ShouldCheckForNsAndGap (const CBioseq_Handle &bsh)
void	CheckBioseqEndsForNAndGap (const CSeqVector &vec, EBioseqEndIsType &begin_n, EBioseqEndIsType &begin_gap, EBioseqEndIsType &end_n, EBioseqEndIsType &end_gap, bool &begin_ambig, bool &end_ambig)
bool	IsDicistronicGene (const CSeq_feat_Handle &f)
	Indicates whether feature is a dicistronic gene. More...
bool	IsDicistronic (const CSeq_feat_Handle &f)
EDuplicateFeatureType	IsDuplicate (const CSeq_feat_Handle &f1, const CSeq_feat_Handle &f2, bool check_partials=false, bool case_sensitive=false)
	Reports how two features duplicate each other. More...
bool	IsLocFullLength (const CSeq_loc &loc, const CBioseq_Handle &bsh)
bool	PartialsSame (const CSeq_loc &loc1, const CSeq_loc &loc2)
bool	IsSpecificHostValid (const string &host, string &error_msg)
	returns true and error_msg will be empty, if specific host is valid returns true and error_msg will be "Host is empty", if specific host is empty returns false if specific host is invalid More...
string	FixSpecificHost (const string &host)
	returns the corrected specific host, if the specific host is invalid and can be corrected returns an empty string, if the specific host is invalid and cannot be corrected returns the original value except the preceding/trailing spaces, if the specific host is valid More...
bool	IsCommonName (const CT3Data &data)
bool	HasMisSpellFlag (const CT3Data &data)
bool	FindMatchInOrgRef (const string &str, const COrg_ref &org)
string	SpecificHostValueToCheck (const string &val)
bool	IsLikelyTaxname (const string &val)
string	InterpretSpecificHostResult (const string &host, const CT3Reply &reply, const string &orig_host=kEmptyStr)
void	AdjustSpecificHostForTaxServer (string &spec_host)
void	ConvertToEntrezTerm (string &title)
string	TranslateCodingRegionForValidation (const CSeq_feat &feat, CScope &scope, bool &alt_start)
bool	ReportTranslationErrors (const string &except_text)
bool	HasBadStartCodon (const CSeq_feat &feat, CScope &scope, bool ignore_exceptions)
bool	HasBadStartCodon (const CSeq_loc &loc, const string &transl_prot)
size_t	CountInternalStopCodons (const string &transl_prot)
bool	HasInternalStop (const CSeq_feat &feat, CScope &scope, bool ignore_exceptions)
CRef< CSeqVector >	MakeSeqVectorForResidueCounting (const CBioseq_Handle &bsh)
bool	HasBadProteinStart (const CSeqVector &sv)
bool	HasBadProteinStart (const CSeq_feat &cds, CScope &scope)
size_t	CountProteinStops (const CSeqVector &sv)
bool	HasStopInProtein (const CSeq_feat &feat, CScope &scope)
void	FeatureHasEnds (const CSeq_feat &feat, CScope *scope, bool &no_beg, bool &no_end)
CBioseq_Handle	GetCDSProductSequence (const CSeq_feat &feat, CScope *scope, const CTSE_Handle &tse, bool far_fetch, bool &is_far)
vector< TSeqPos >	GetMismatches (const CSeq_feat &feat, const CBioseq_Handle &prot_handle, const string &transl_prot)
vector< TSeqPos >	GetMismatches (const CSeq_feat &feat, const CSeqVector &prot_vec, const string &transl_prot)
void	CalculateEffectiveTranslationLengths (const string &transl_prot, const CSeqVector &prot_vec, size_t &len, size_t &prot_len)
bool	HasNoStop (const CSeq_feat &feat, CScope *scope)
bool	IsSequenceFetchable (const CSeq_id &id, CScope *scope=nullptr)
bool	IsSequenceFetchable (const string &seq_id, CScope *scope=nullptr)
bool	DoesFeatureHaveUnnecessaryException (const CSeq_feat &feat, CScope &scope)
bool	IsNTNCNWACAccession (const string &acc)
bool	IsNTNCNWACAccession (const CSeq_id &id)
bool	IsNTNCNWACAccession (const CBioseq &seq)
bool	IsNG (const CSeq_id &id)
bool	IsNG (const CBioseq &seq)
bool	IsTemporary (const CSeq_id &id)
bool	IsOrganelle (int genome)
bool	IsOrganelle (const CBioseq_Handle &seq)
bool	IsInOrganelleSmallGenomeSet (const CSeq_id &id, CScope &scope)
bool	BadMultipleSequenceLocation (const CSeq_loc &loc, CScope &scope)
bool	ConsistentWithA (Char ch)
bool	ConsistentWithC (Char ch)
bool	ConsistentWithG (Char ch)
bool	ConsistentWithT (Char ch)
bool	IsBadSubmissionFirstName (const string &first)
bool	IsBadSubmissionLastName (const string &last)
bool	g_IsMasterAccession (const CSeq_id &id)

Typedef Documentation

TDbtags

typedef const CSeq_feat::TDbxref TDbtags

Definition at line 199 of file utilities.hpp.

Enumeration Type Documentation

EAccessionFormatError

enum EAccessionFormatError

Enumerator
eAccessionFormat_valid
eAccessionFormat_no_start_letters
eAccessionFormat_wrong_number_of_digits
eAccessionFormat_null
eAccessionFormat_too_long
eAccessionFormat_missing_version
eAccessionFormat_bad_version

Definition at line 105 of file utilities.hpp.

EBioseqEndIsType

enum EBioseqEndIsType

Enumerator
eBioseqEndIsType_None
eBioseqEndIsType_Last
eBioseqEndIsType_All

Definition at line 156 of file utilities.hpp.

EDateValid

enum EDateValid

Enumerator
eDateValid_valid
eDateValid_bad_str
eDateValid_bad_year
eDateValid_bad_month
eDateValid_bad_day
eDateValid_bad_season
eDateValid_bad_other
eDateValid_empty_date

Definition at line 122 of file utilities.hpp.

EDuplicateFeatureType

enum EDuplicateFeatureType

Enumerator
eDuplicate_Not
eDuplicate_Duplicate
eDuplicate_SameIntervalDifferentLabel
eDuplicate_DuplicateDifferentTable
eDuplicate_SameIntervalDifferentLabelDifferentTable

Definition at line 191 of file utilities.hpp.

Function Documentation

AdjustSpecificHostForTaxServer()

void AdjustSpecificHostForTaxServer

(

string &

spec_host

)

Definition at line 2032 of file utilities.cpp.

References ArraySize(), i, NStr::ReplaceInPlace(), sIgnoreHostWordList, and NStr::TruncateSpacesInPlace().

Referenced by SpecificHostValueToCheck().

AppendBioseqLabel()

void AppendBioseqLabel	(	string &	str,
		const CBioseq &	sq,
		bool	supress_context
	)

Definition at line 1064 of file utilities.cpp.

References ENUM_METHOD_NAME, CEnumeratedTypeValues::FindName(), GetBioseqIdLabel(), CBioseq_Base::GetInst(), CSeq_inst_Base::GetLength(), CSeq_inst_Base::GetMol(), CSeq_inst_Base::GetRepr(), NStr::IntToString(), CSeq_inst_Base::IsSetLength(), str(), and string.

Referenced by CValidErrorFormat::GetDescriptorLabel(), CValidError_imp::PostErr(), and CValidError_bioseq::x_ReportOverlappingPeptidePair().

BadMultipleSequenceLocation()

bool BadMultipleSequenceLocation	(	const CSeq_loc &	loc,
		CScope &	scope
	)

Definition at line 3080 of file utilities.cpp.

References CSeq_loc_CI::GetSeq_id(), IsInOrganelleSmallGenomeSet(), IsSameBioseq(), and CSeq_id::Match().

Referenced by CRemoveUnindexedFeatures::apply(), CValidError_imp::BadMultipleSequenceLocation(), and CValidError_imp::ValidateSeqLocIds().

BioseqHandleFromLocation()

CBioseq_Handle BioseqHandleFromLocation	(	CScope *	m_Scope,
		const CSeq_loc &	loc
	)

Definition at line 1277 of file utilities.cpp.

References CScope::eGetBioseq_All, CScope::GetBioseqHandle(), CSeq_id_Handle::GetHandle(), and m_Scope.

Referenced by CValidator::CCacheImpl::GetBioseqHandleFromLocation(), GetSequenceStringFromLoc(), IsDeltaOrFarSeg(), s_AreLinkedToDifferentFeats(), and s_GetSeq_featAcc().

CalculateEffectiveTranslationLengths()

void CalculateEffectiveTranslationLengths	(	const string &	transl_prot,
		const CSeqVector &	prot_vec,
		size_t &	len,
		size_t &	prot_len
	)

Definition at line 2597 of file utilities.cpp.

References NStr::EndsWith(), len, and CSeqVector::size().

Referenced by CCDSTranslationProblems::CalculateTranslationProblems(), GetMismatches(), and CCDSTranslationProblems::x_GetTranslationMismatches().

CheckBioseqEndsForNAndGap() [1/2]

void CheckBioseqEndsForNAndGap	(	const CBioseq_Handle &	bsh,
		EBioseqEndIsType &	begin_n,
		EBioseqEndIsType &	begin_gap,
		EBioseqEndIsType &	end_n,
		EBioseqEndIsType &	end_gap,
		bool &	begin_ambig,
		bool &	end_ambig
	)

Definition at line 1422 of file utilities.cpp.

References CheckBioseqEndsForNAndGap(), eBioseqEndIsType_None, CBioseq_Handle::eCoding_Iupac, CBioseq_Handle::GetSeqVector(), and ShouldCheckForNsAndGap().

Referenced by BOOST_AUTO_TEST_CASE(), and CValidError_bioseq::ValidateNsAndGaps().

CheckBioseqEndsForNAndGap() [2/2]

void CheckBioseqEndsForNAndGap	(	const CSeqVector &	vec,
		EBioseqEndIsType &	begin_n,
		EBioseqEndIsType &	begin_gap,
		EBioseqEndIsType &	end_n,
		EBioseqEndIsType &	end_gap,
		bool &	begin_ambig,
		bool &	end_ambig
	)

Definition at line 1317 of file utilities.cpp.

References eBioseqEndIsType_All, eBioseqEndIsType_Last, eBioseqEndIsType_None, i, CSeqVector::IsInGap(), CSeqVector::IsNucleotide(), s_PosIsNNotGap(), and CSeqVector::size().

Referenced by CheckBioseqEndsForNAndGap().

CheckDate()

int CheckDate	(	const CDate &	date,
		bool	require_full_date = false
	)

Definition at line 780 of file utilities.cpp.

References eDateValid_bad_day, eDateValid_bad_month, eDateValid_bad_other, eDateValid_bad_season, eDateValid_bad_str, eDateValid_bad_year, eDateValid_valid, NStr::Equal(), CDate_Base::GetStd(), CDate_Base::GetStr(), isalpha(), NStr::IsBlank(), CDate_Base::IsStd(), and CDate_Base::IsStr().

Referenced by CValidError_imp::ValidateCitSub(), CValidError_imp::ValidatePubArticleNoPMID(), CValidError_imp::ValidatePubGen(), and CValidError_desc::ValidateSeqDesc().

ConsistentWithA()

bool ConsistentWithA

(

Char

ch

)

Definition at line 2871 of file utilities.cpp.

References NULL.

Referenced by CheckAdjacentSpliceSites(), CheckSpliceSite(), and CCDSTranslationProblems::x_Is5AtEndSpliceSiteOrGap().

ConsistentWithC()

bool ConsistentWithC

(

Char

ch

)

Definition at line 2876 of file utilities.cpp.

References NULL.

Referenced by CheckAdjacentSpliceSites(), CheckSpliceSite(), and CCDSTranslationProblems::x_Is5AtEndSpliceSiteOrGap().

ConsistentWithG()

bool ConsistentWithG

(

Char

ch

)

Definition at line 2881 of file utilities.cpp.

References NULL.

Referenced by CheckAdjacentSpliceSites(), CheckSpliceSite(), and CCDSTranslationProblems::x_Is5AtEndSpliceSiteOrGap().

ConsistentWithT()

bool ConsistentWithT

(

Char

ch

)

Definition at line 2886 of file utilities.cpp.

References NULL.

Referenced by CheckAdjacentSpliceSites(), CheckSpliceSite(), and CCDSTranslationProblems::x_Is5AtEndSpliceSiteOrGap().

ConvertToEntrezTerm()

void ConvertToEntrezTerm

(

string &

title

)

Definition at line 2220 of file utilities.cpp.

References isspace(), s_ConvertChar(), and NStr::TruncateSpacesInPlace().

Referenced by pub_report::CSkipPubJournalHook::IsJournalMissing(), and CDoiLookup::LookupIsojta().

CountInternalStopCodons()

size_t CountInternalStopCodons

(

const string &

transl_prot

)

Definition at line 2393 of file utilities.cpp.

References NStr::IsBlank(), and ITERATE.

Referenced by CCDSTranslationProblems::CalculateTranslationProblems(), GetInternalStopErrorMessage(), GetStartCodonErrorMessage(), and HasInternalStop().

CountProteinStops()

size_t CountProteinStops

(

const CSeqVector &

sv

)

Definition at line 2504 of file utilities.cpp.

Referenced by HasStopInProtein(), and CValidError_bioseq::ValidateRawConst().

DoesFeatureHaveUnnecessaryException()

bool DoesFeatureHaveUnnecessaryException	(	const CSeq_feat &	feat,
		CScope &	scope
	)

Definition at line 2939 of file utilities.cpp.

References CSpliceProblems::CalculateSpliceProblems(), DoesCodingRegionHaveUnnecessaryException(), DoesmRNAHaveUnnecessaryException(), CSeqFeatData::eSubtype_mRNA, CScope::GetBioseqHandle(), CSeq_feat_Base::GetData(), CSeq_feat_Base::GetLocation(), CSeqFeatData::GetSubtype(), CSeqFeatData_Base::IsCdregion(), CSpliceProblems::IsExceptionUnnecessary(), IsPseudo(), CSeq_feat_Base::IsSetData(), CSeq_feat_Base::IsSetExcept_text(), and CSeq_feat_Base::IsSetLocation().

Referenced by CRemoveUnnecessaryExceptions::apply(), and BOOST_AUTO_TEST_CASE().

EndsWithBadCharacter()

bool EndsWithBadCharacter

(

const string &

str

)

Definition at line 768 of file utilities.cpp.

References NStr::EndsWith(), and str().

Referenced by CSingleFeatValidator::ValidateCharactersInField().

FeatureHasEnds()

void FeatureHasEnds	(	const CSeq_feat &	feat,
		CScope *	scope,
		bool &	no_beg,
		bool &	no_end
	)

Definition at line 2546 of file utilities.cpp.

References eSeqlocPartial_Start, eSeqlocPartial_Stop, CSeq_feat_Base::GetLocation(), CSeq_feat_Base::GetProduct(), CSeq_feat_Base::IsSetProduct(), and SeqLocPartialCheck().

Referenced by CCDSTranslationProblems::CalculateTranslationProblems(), GetMismatches(), HasNoStop(), CCDSTranslationProblems::x_GetCdTransErrors(), and CCDSTranslationProblems::x_GetTranslationMismatches().

FindMatchInOrgRef()

bool FindMatchInOrgRef	(	const string &	str,
		const COrg_ref &	org
	)

Definition at line 1985 of file utilities.cpp.

References NStr::EqualCase(), NStr::EqualNocase(), COrgMod_Base::eSubtype_gb_synonym, COrgMod_Base::eSubtype_old_name, FOR_EACH_SYN_ON_ORGREF, COrg_ref_Base::GetCommon(), COrgName_Base::GetMod(), COrg_ref_Base::GetOrgname(), COrg_ref_Base::GetTaxname(), NStr::IsBlank(), COrg_ref_Base::IsSetCommon(), COrgName_Base::IsSetMod(), COrg_ref_Base::IsSetOrgname(), COrg_ref_Base::IsSetTaxname(), match(), and str().

Referenced by CSpecificHostRequest::AddReply(), and InterpretSpecificHostResult().

FixSpecificHost()

string FixSpecificHost

(

const string &

host

)

returns the corrected specific host, if the specific host is invalid and can be corrected returns an empty string, if the specific host is invalid and cannot be corrected returns the original value except the preceding/trailing spaces, if the specific host is valid

Definition at line 2189 of file utilities.cpp.

References val.

Referenced by BOOST_AUTO_TEST_CASE().

g_IsMasterAccession()

bool g_IsMasterAccession

(

const CSeq_id &

id

)

Definition at line 2999 of file utilities.cpp.

References CSeq_id_Base::e_Ddbj, CSeq_id_Base::e_Embl, CSeq_id_Base::e_Genbank, CSeq_id_Base::e_Other, CSeq_id_Base::e_Tpg, NStr::EndsWith(), GetAccession(), and s_IsGenbankMasterAccession().

Referenced by CValidError_bioseq::IsMaster(), and s_IsMaster().

GetAccessionFromBioseq()

string GetAccessionFromBioseq	(	const CBioseq &	bioseq,
		int *	version
	)

Definition at line 427 of file utilities.cpp.

References s_GetBioseqAcc(), and dtl::version.

Referenced by CValidError_imp::PostErr(), and CTbl2AsnApp::ProcessHugeFile().

GetAccessionFromBioseqSet()

string GetAccessionFromBioseqSet	(	const CBioseq_set &	bsst,
		int *	version
	)

Definition at line 433 of file utilities.cpp.

References kEmptyStr, s_GetBioseqAcc(), s_GetSeqFromSet(), and dtl::version.

Referenced by CValidErrorFormat::GetBioseqSetLabel(), CValidError_imp::PostErr(), and CTbl2AsnApp::ProcessHugeFile().

GetAccessionFromObjects()

string GetAccessionFromObjects	(	const CSerialObject *	obj,
		const CSeq_entry *	ctx,
		CScope &	scope,
		int *	version
	)

Definition at line 443 of file utilities.cpp.

References ctx, CScope::GetBioseqHandle(), CBioseq_set_Handle::GetCompleteBioseq_set(), CSeq_loc::GetId(), CSeq_graph_Base::GetLoc(), CSeq_annot_Handle::GetParentEntry(), GetReportableSeqIdForAlignment(), CSeq_entry_Base::GetSeq(), CSeq_entry_Handle::GetSeq(), CScope::GetSeq_annotHandle(), CScope::GetSeq_entryHandle(), CSeq_entry_Base::GetSet(), CSeq_entry_Handle::GetSet(), CSerialObject::GetThisTypeInfo(), NStr::IsBlank(), CSeq_entry_Base::IsSeq(), CSeq_entry_Handle::IsSeq(), CSeq_entry_Base::IsSet(), CSeq_entry_Handle::IsSet(), s_GetAccessionForSeqdesc(), s_GetBioseqAcc(), s_GetSeq_featAcc(), s_GetSeqFromSet(), and dtl::version.

Referenced by CValidError_imp::PostErr(), CTable2AsnValidator::Validate(), and CValidError_align::x_ValidateDim().

GetBioseqIdLabel()

string GetBioseqIdLabel

(

const CBioseq &

sq

)

Definition at line 981 of file utilities.cpp.

References CBioseq_Base::GetId().

Referenced by AppendBioseqLabel(), GetValidatorLocationLabel(), and CSingleFeatValidator::ValidateSplice().

GetCDSProductSequence()

CBioseq_Handle GetCDSProductSequence	(	const CSeq_feat &	feat,
		CScope *	scope,
		const CTSE_Handle &	tse,
		bool	far_fetch,
		bool &	is_far
	)

Definition at line 2574 of file utilities.cpp.

References CScope::GetBioseqHandle(), CScope::GetBioseqHandleFromTSE(), GetId(), CSeq_feat_Base::GetProduct(), and CSeq_feat_Base::IsSetProduct().

Referenced by HasNoStop().

GetDateErrorDescription()

string GetDateErrorDescription

(

int

flags

)

Definition at line 855 of file utilities.cpp.

References eDateValid_bad_day, eDateValid_bad_month, eDateValid_bad_other, eDateValid_bad_season, eDateValid_bad_str, eDateValid_bad_year, eDateValid_empty_date, and flags.

Referenced by CValidError_imp::PostBadDateError().

GetGenProdSetParent() [1/2]

CBioseq_set_Handle GetGenProdSetParent

(

const CBioseq_Handle &

set

)

Definition at line 575 of file utilities.cpp.

References CBioseq_set_Base::eClass_gen_prod_set, and GetSetParent().

GetGenProdSetParent() [2/2]

CBioseq_set_Handle GetGenProdSetParent

(

const CBioseq_set_Handle &

set

)

Definition at line 570 of file utilities.cpp.

References CBioseq_set_Base::eClass_gen_prod_set, and GetSetParent().

Referenced by CValidError_bioseqset::ValidateNucProtSet(), and CSingleFeatValidator::x_ValidateExceptText().

GetGIForSeqId()

TGi GetGIForSeqId

(

const CSeq_id &

id

)

Definition at line 125 of file utilities.cpp.

References CScope::AddDefaults(), CScope::GetGi(), CSeq_id_Handle::GetHandle(), CObjectManager::GetInstance(), and ZERO_GI.

GetMismatches() [1/2]

vector<TSeqPos> GetMismatches	(	const CSeq_feat &	feat,
		const CBioseq_Handle &	prot_handle,
		const string &	transl_prot
	)

Definition at line 2658 of file utilities.cpp.

References CSeq_data_Base::e_Ncbieaa, GetMismatches(), CBioseq_Handle::GetSeqVector(), CBioseq_Handle::IsAa(), and CSeqVector::SetCoding().

GetMismatches() [2/2]

vector<TSeqPos> GetMismatches	(	const CSeq_feat &	feat,
		const CSeqVector &	prot_vec,
		const string &	transl_prot
	)

Definition at line 2626 of file utilities.cpp.

References CalculateEffectiveTranslationLengths(), FeatureHasEnds(), CSeq_feat_Base::GetPartial(), CSeqVector::GetScope(), i, CSeq_feat_Base::IsSetPartial(), and len.

Referenced by GetMismatches(), and HasNoStop().

GetNucBioseq() [1/2]

CBioseq_Handle GetNucBioseq

(

const CBioseq_Handle &

bioseq

)

Definition at line 608 of file utilities.cpp.

References GetNucBioseq(), CBioseq_Handle::GetParentEntry(), CSeq_entry_Handle::GetParentEntry(), CSeq_entry_Handle::GetSet(), CBioseq_Handle::IsNucleotide(), CSeq_entry_Handle::IsSet(), and nuc.

GetNucBioseq() [2/2]

CBioseq_Handle GetNucBioseq

(

const CBioseq_set_Handle &

bioseq_set

)

Definition at line 587 of file utilities.cpp.

References CSeq_inst_Base::eMol_na, CBioseq_set_Handle::GetParentEntry(), CSeq_entry_Handle::GetParentEntry(), CSeq_entry_Handle::GetSet(), CSeq_entry_Handle::IsSet(), and nuc.

Referenced by CValidError_imp::AddBioseqWithNoPub(), GetNucBioseq(), and CValidError_bioseq::ValidateBioseqContext().

GetNucProtSetParent()

CBioseq_set_Handle GetNucProtSetParent

(

const CBioseq_Handle &

bioseq

)

Definition at line 581 of file utilities.cpp.

References CBioseq_set_Base::eClass_nuc_prot, and GetSetParent().

Referenced by CValidError_bioseq::ValidateBioseqContext().

GetReportableSeqIdForAlignment()

CConstRef<CSeq_id> GetReportableSeqIdForAlignment	(	const CSeq_align &	align,
		CScope &	scope
	)

Definition at line 399 of file utilities.cpp.

References CScope::GetBioseqHandle(), CSeq_align_Base::C_Segs::GetDendiag(), CSeq_align_Base::GetDim(), CSeq_align_Base::GetSegs(), CSeq_align::GetSeq_id(), i, IsBioseqInSameSeqEntryAsAlign(), CSeq_align_Base::C_Segs::IsDendiag(), CSeq_align_Base::IsSetDim(), CSeq_align_Base::IsSetSegs(), and CSeq_align_Base::C_Segs::IsStd().

Referenced by GetAccessionFromObjects(), CValidError_imp::PostErr(), and CValidError_align::x_ValidateDim().

GetSeqIdsForGI()

CScope::TIds GetSeqIdsForGI

(

TGi

gi

)

Definition at line 142 of file utilities.cpp.

References CScope::AddDefaults(), CScope::GetIds(), CObjectManager::GetInstance(), and CSeq_id_Base::SetGi().

Referenced by CValidError_bioseq::ValidateIDSetAgainstDb().

GetSequenceFromFeature()

CSeqVector GetSequenceFromFeature	(	const CSeq_feat &	feat,
		CScope &	scope,
		CBioseq_Handle::EVectorCoding	coding = CBioseq_Handle::eCoding_Iupac,
		bool	product = false
	)

Definition at line 214 of file utilities.cpp.

References CSeq_feat_Base::CanGetLocation(), CSeq_feat_Base::CanGetProduct(), CSeq_feat_Base::GetLocation(), CSeq_feat_Base::GetProduct(), and GetSequenceFromLoc().

Referenced by CGapFeatValidator::Validate(), and CSingleFeatValidator::x_ValidateRptUnitVal().

GetSequenceFromLoc()

CSeqVector GetSequenceFromLoc	(	const CSeq_loc &	loc,
		CScope &	scope,
		CBioseq_Handle::EVectorCoding	coding = CBioseq_Handle::eCoding_Iupac
	)

Definition at line 203 of file utilities.cpp.

References CSeqMap::CreateSeqMapForSeq_loc(), and eNa_strand_plus.

Referenced by GetSequenceFromFeature(), CSingleFeatValidator::x_CalculateLocationGaps(), and CSingleFeatValidator::x_IsMostlyNs().

GetSequenceStringFromLoc()

string GetSequenceStringFromLoc	(	const CSeq_loc &	loc,
		CScope &	scope
	)

Definition at line 175 of file utilities.cpp.

References BioseqHandleFromLocation(), CFastaOstream::fAssembleParts, CFastaOstream::fInstantiateGaps, kEmptyStr, NStr::ReplaceInPlace(), CFastaOstream::SetFlag(), and CFastaOstream::WriteSequence().

Referenced by CGapFeatValidator::Validate(), CValidError_bioseq::ValidateRawConst(), and CRNAValidator::x_ValidateTrnaCodons().

GetSetParent() [1/2]

CBioseq_set_Handle GetSetParent	(	const CBioseq_Handle &	bioseq,
		CBioseq_set::TClass	set_class
	)

Definition at line 551 of file utilities.cpp.

References CBioseq_set_Handle::GetClass(), CBioseq_Handle::GetParentEntry(), CSeq_entry_Handle::GetParentEntry(), CSeq_entry_Handle::GetSet(), GetSetParent(), CSeq_entry_Handle::IsSet(), and CBioseq_set_Handle::IsSetClass().

GetSetParent() [2/2]

CBioseq_set_Handle GetSetParent	(	const CBioseq_set_Handle &	set,
		CBioseq_set::TClass	set_class
	)

Definition at line 532 of file utilities.cpp.

References CBioseq_set_Handle::GetClass(), CSeq_entry_Handle::GetParentEntry(), CSeq_entry_Handle::GetSet(), CSeq_entry_Handle::IsSet(), and CBioseq_set_Handle::IsSetClass().

Referenced by GetGenProdSetParent(), GetNucProtSetParent(), and GetSetParent().

GetValidatorLocationLabel()

string GetValidatorLocationLabel	(	const CSeq_loc &	loc,
		CScope &	scope
	)

Definition at line 958 of file utilities.cpp.

References CSeq_loc::Assign(), CScope::GetBioseqHandle(), GetBioseqIdLabel(), CBioseq_Handle::GetCompleteBioseq(), CSeq_loc::GetLabel(), CSeq_loc_Base::GetWhole(), NStr::IsBlank(), CSeq_loc_Base::IsWhole(), NStr::ReplaceInPlace(), and UpdateToBestId().

Referenced by CValidErrorFormat::GetFeatureLocationLabel(), CValidErrorFormat::GetFeatureProductLocLabel(), CCdregionValidator::MapToNTCoords(), CValidError_imp::ValidateSeqLoc(), CValidError_imp::x_CheckLoc(), CCdregionValidator::x_ValidateCodebreak(), and CValidError_graph::x_ValidateGraphLocation().

HasBadCharacter()

bool HasBadCharacter

(

const string &

str

)

Definition at line 755 of file utilities.cpp.

References NStr::Find(), and str().

Referenced by CSingleFeatValidator::ValidateCharactersInField().

HasBadProteinStart() [1/2]

bool HasBadProteinStart	(	const CSeq_feat &	cds,
		CScope &	scope
	)

Definition at line 2481 of file utilities.cpp.

References CScope::GetBioseqHandle(), CSeq_feat_Base::GetData(), CSeq_feat_Base::GetProduct(), HasBadProteinStart(), CBioseq_Handle::IsAa(), CSeqFeatData_Base::IsCdregion(), CSeq_feat_Base::IsSetData(), CSeq_feat_Base::IsSetProduct(), and MakeSeqVectorForResidueCounting().

HasBadProteinStart() [2/2]

bool HasBadProteinStart

(

const CSeqVector &

sv

)

Definition at line 2467 of file utilities.cpp.

References CSeqVector::IsInGap(), and CSeqVector::size().

Referenced by CConvertBadCdsAndRnaToMiscFeat::apply(), HasBadProteinStart(), and CValidError_bioseq::ValidateRawConst().

HasBadStartCodon() [1/2]

bool HasBadStartCodon	(	const CSeq_feat &	feat,
		CScope &	scope,
		bool	ignore_exceptions
	)

Definition at line 2362 of file utilities.cpp.

References CSeq_feat_Base::CanGetExcept(), CSeq_feat_Base::CanGetExcept_text(), NStr::EqualNocase(), FOR_EACH_GBQUAL_ON_FEATURE, CSeq_feat_Base::GetData(), CSeq_feat_Base::GetExcept(), CSeq_feat_Base::GetExcept_text(), CSeq_feat_Base::GetLocation(), HasBadStartCodon(), CSeqFeatData_Base::IsCdregion(), CSeq_feat_Base::IsSetData(), ReportTranslationErrors(), and TranslateCodingRegionForValidation().

Referenced by CConvertBadCdsAndRnaToMiscFeat::apply(), BOOST_AUTO_TEST_CASE(), CCDSTranslationProblems::CalculateTranslationProblems(), and GetInternalStopErrorMessage().

HasBadStartCodon() [2/2]

bool HasBadStartCodon	(	const CSeq_loc &	loc,
		const string &	transl_prot
	)

Definition at line 2307 of file utilities.cpp.

References eExtreme_Biological, and CSeq_loc::IsPartialStart().

Referenced by HasBadStartCodon().

HasECnumberPattern()

bool HasECnumberPattern

(

const string &

str

)

Definition at line 1088 of file utilities.cpp.

References NStr::IsBlank(), isdigit(), and str().

Referenced by CProtValidator::Validate(), and CCdregionValidator::x_ValidateFeatComment().

HasInternalStop()

bool HasInternalStop	(	const CSeq_feat &	feat,
		CScope &	scope,
		bool	ignore_exceptions
	)

Definition at line 2416 of file utilities.cpp.

References CSeq_feat_Base::CanGetExcept(), CSeq_feat_Base::CanGetExcept_text(), CountInternalStopCodons(), NStr::EqualNocase(), NStr::Find(), FOR_EACH_GBQUAL_ON_FEATURE, CSeq_feat_Base::GetData(), CSeq_feat_Base::GetExcept(), CSeq_feat_Base::GetExcept_text(), CSeqFeatData_Base::IsCdregion(), CSeq_feat_Base::IsSetData(), kUnclassifiedTranslationDiscrepancy, ReportTranslationErrors(), and TranslateCodingRegionForValidation().

Referenced by CConvertBadCdsAndRnaToMiscFeat::apply(), and BOOST_AUTO_TEST_CASE().

HasMisSpellFlag()

bool HasMisSpellFlag

(

const CT3Data &

data

)

Definition at line 1966 of file utilities.cpp.

References data, NStr::EqualNocase(), and ITERATE.

Referenced by CSpecificHostRequest::AddReply(), and InterpretSpecificHostResult().

HasNoStop()

bool HasNoStop	(	const CSeq_feat &	feat,
		CScope *	scope
	)

Definition at line 2673 of file utilities.cpp.

References NStr::EndsWith(), FeatureHasEnds(), CScope::GetBioseqHandle(), GetCDSProductSequence(), CSeq_feat_Base::GetLocation(), GetMismatches(), CSeq_feat_Base::GetPartial(), CBioseq_Handle::GetTSE_Handle(), CSeq_feat_Base::IsSetPartial(), and TranslateCodingRegionForValidation().

Referenced by CConvertBadCdsAndRnaToMiscFeat::apply(), and BOOST_AUTO_TEST_CASE().

HasStopInProtein()

bool HasStopInProtein	(	const CSeq_feat &	feat,
		CScope &	scope
	)

Definition at line 2519 of file utilities.cpp.

References CountProteinStops(), CScope::GetBioseqHandle(), CSeq_feat_Base::GetData(), CSeq_feat_Base::GetProduct(), CBioseq_Handle::IsAa(), CSeqFeatData_Base::IsCdregion(), CSeq_feat_Base::IsSetData(), CSeq_feat_Base::IsSetProduct(), and MakeSeqVectorForResidueCounting().

Referenced by CConvertBadCdsAndRnaToMiscFeat::apply(), and BOOST_AUTO_TEST_CASE().

InterpretSpecificHostResult()

string InterpretSpecificHostResult	(	const string &	host,
		const CT3Reply &	reply,
		const string &	orig_host = kEmptyStr
	)

Definition at line 2085 of file utilities.cpp.

References FindMatchInOrgRef(), NStr::FindNoCase(), CT3Reply_Base::GetData(), CT3Reply_Base::GetError(), CT3Error_Base::GetMessage(), CT3Data_Base::GetOrg(), COrg_ref_Base::GetTaxname(), HasMisSpellFlag(), NStr::IsBlank(), IsCommonName(), CT3Reply_Base::IsData(), CT3Reply_Base::IsError(), CT3Error_Base::IsSetMessage(), and NStr::StartsWith().

Referenced by CSpecificHostRequest::AddReply().

IsBadSubmissionFirstName()

bool IsBadSubmissionFirstName

(

const string &

first

)

Definition at line 3102 of file utilities.cpp.

References NStr::EqualNocase(), and first().

Referenced by CValidError_imp::ValidateAuthorsInPubequiv(), and CValidError_imp::ValidateSubmitBlock().

IsBadSubmissionLastName()

bool IsBadSubmissionLastName

(

const string &

last

)

Definition at line 3116 of file utilities.cpp.

References NStr::EqualNocase(), and last().

Referenced by CValidError_imp::ValidateAuthorsInPubequiv(), and CValidError_imp::ValidateSubmitBlock().

IsBioseqTSA()

bool IsBioseqTSA	(	const CBioseq &	seq,
		CScope *	scope
	)

Definition at line 884 of file utilities.cpp.

References CSeqdesc_Base::e_Molinfo, CMolInfo_Base::eTech_tsa, CScope::GetBioseqHandle(), CSeqdesc_Base::GetMolinfo(), CMolInfo_Base::GetTech(), and CMolInfo_Base::IsSetTech().

Referenced by CValidError_bioseq::ValidateNsAndGaps().

IsBlankStringList()

bool IsBlankStringList

(

const list< string > &

str_list

)

Definition at line 114 of file utilities.cpp.

References NStr::IsBlank(), ITERATE, and str().

Referenced by CValidError_imp::HasName().

IsClassInEntry()

bool IsClassInEntry	(	const CSeq_entry &	se,
		CBioseq_set::EClass	clss
	)

Definition at line 79 of file utilities.cpp.

References si.

Referenced by IsDeltaOrFarSeg().

IsCommonName()

bool IsCommonName

(

const CT3Data &

data

)

Definition at line 1944 of file utilities.cpp.

References data, NStr::eCase, NStr::eNocase, NStr::Equal(), and ITERATE.

Referenced by CSpecificHostRequest::AddReply(), and InterpretSpecificHostResult().

IsDateInPast()

bool IsDateInPast

(

const CDate &

date

)

Definition at line 826 of file utilities.cpp.

References CDate_Base::GetStd(), CDate_Base::IsStd(), and t.

Referenced by BOOST_AUTO_TEST_CASE(), and CValidError_imp::ValidateSubmitBlock().

IsDeltaOrFarSeg()

bool IsDeltaOrFarSeg	(	const CSeq_loc &	loc,
		CScope *	scope
	)

Definition at line 90 of file utilities.cpp.

References BioseqHandleFromLocation(), CBioseq_set_Base::eClass_nuc_prot, CBioseq_set_Base::eClass_parts, CSeq_inst_Base::eRepr_delta, CSeq_inst_Base::eRepr_seg, CSeq_entry_Handle::GetCompleteSeq_entry(), CBioseq_Handle::GetInst_Repr(), CBioseq_Handle::GetTopLevelEntry(), IsClassInEntry(), and CBioseq_Handle::IsSetInst_Repr().

Referenced by CCdregionValidator::x_ReportTranslationProblems().

IsDicistronic()

bool IsDicistronic

(

const CSeq_feat_Handle &

f

)

Definition at line 1825 of file utilities.cpp.

References f(), NStr::FindNoCase(), and NPOS.

Referenced by IsDicistronicGene().

IsDicistronicGene()

bool IsDicistronicGene

(

const CSeq_feat_Handle &

f

)

Indicates whether feature is a dicistronic gene.

Parameters

f	Seq-feat-Handle [in]

Returns

Boolean

Definition at line 1818 of file utilities.cpp.

References CSeqFeatData::eSubtype_gene, f(), and IsDicistronic().

Referenced by CValidError_bioseq::x_SuppressDicistronic().

IsDuplicate()

EDuplicateFeatureType IsDuplicate	(	const CSeq_feat_Handle &	f1,
		const CSeq_feat_Handle &	f2,
		bool	check_partials = false,
		bool	case_sensitive = false
	)

Reports how two features duplicate each other.

Parameters

f1	Seq-feat-Handle [in]
f2	Seq-feat-Handle [in]

Returns

EDuplicateFeatureType return value indicates how features are duplicates

Definition at line 1838 of file utilities.cpp.

References Compare(), eDuplicate_Duplicate, eDuplicate_DuplicateDifferentTable, eDuplicate_Not, eDuplicate_SameIntervalDifferentLabel, eDuplicate_SameIntervalDifferentLabelDifferentTable, eSame, CSeqFeatData::eSubtype_pub, CSeqFeatData::eSubtype_region, CSeqFeatData::eSubtype_variation, fCompareOverlapping, CSeq_feat_Handle::GetData(), CSeq_feat_Handle::GetDbxref(), CSeq_feat_Handle::GetLocation(), CSeq_feat_Handle::GetScope(), CSeqFeatData::GetSubtype(), CSeq_feat_Handle::IsSetDbxref(), PartialsSame(), s_AreCodingRegionsLinkedToDifferentmRNAs(), s_AreDifferentVariations(), s_AreFeatureLabelsSame(), s_AreFullLengthCodingRegionsWithDifferentFrames(), s_AremRNAsLinkedToDifferentCodingRegions(), s_IsDifferentDbxrefs(), s_IsSameSeqAnnot(), and s_IsSameStrand().

Referenced by CRmDupFeaturesDlg::GetCommand(), GetDuplicateFeaturesForRemoval(), CRemoveFeaturesDlg::RemoveDuplicateFeatures(), and CValidError_bioseq::x_ReportDupOverlapFeaturePair().

IsFarLocation()

bool IsFarLocation	(	const CSeq_loc &	loc,
		const CSeq_entry_Handle &	seh
	)

Definition at line 159 of file utilities.cpp.

References CScope::GetBioseqHandleFromTSE(), and CSeq_entry_Handle::GetScope().

Referenced by CRemoveUnindexedFeatures::apply(), CRemoveFeaturesDlg::GetProductId(), CValidError_bioseq::NumOfIntervals(), CCdsToMatPeptide::RemoveProteins(), CDesktopAlignItem::s_ContainsFarPointers(), s_IsFarLocation(), CValidError_bioseq::ValidateSeqFeatContext(), and CValidError_imp::x_IsFarFetchFailure().

IsInOrganelleSmallGenomeSet()

bool IsInOrganelleSmallGenomeSet	(	const CSeq_id &	id,
		CScope &	scope
	)

Definition at line 3050 of file utilities.cpp.

References CSeqdesc_Base::e_Source, CBioseq_set_Base::eClass_nuc_prot, CBioseq_set_Base::eClass_small_genome_set, CScope::GetBioseqHandle(), CBioSource_Base::GetGenome(), CBioseq_Handle::GetParentBioseq_set(), CSeqdesc_Base::GetSource(), IsOrganelle(), and CBioSource_Base::IsSetGenome().

Referenced by BadMultipleSequenceLocation(), and CValidError_imp::IsInOrganelleSmallGenomeSet().

IsLikelyTaxname()

bool IsLikelyTaxname

(

const string &

val

)

Definition at line 2150 of file utilities.cpp.

References NStr::Find(), CTaxon1::GetOrgRef(), CTaxon1::GetTaxIdByName(), CTaxon1::Init(), INVALID_TAX_ID, isalpha(), IsCommon(), NPOS, val, and ZERO_TAX_ID.

Referenced by CSpecificHostRequest::AddReply(), and BOOST_AUTO_TEST_CASE().

IsLocFullLength()

bool IsLocFullLength	(	const CSeq_loc &	loc,
		const CBioseq_Handle &	bsh
	)

Definition at line 1451 of file utilities.cpp.

References CSeq_interval_Base::GetFrom(), CBioseq_Handle::GetInst_Length(), CSeq_loc_Base::GetInt(), CSeq_interval_Base::GetTo(), and CSeq_loc_Base::IsInt().

Referenced by s_AreFullLengthCodingRegionsWithDifferentFrames(), CValidError_bioseq::x_ValidatePubFeatures(), and CValidError_bioseq::x_ValidateSourceFeatures().

IsNG() [1/2]

bool IsNG

(

const CBioseq &

seq

)

Definition at line 2801 of file utilities.cpp.

References FOR_EACH_SEQID_ON_BIOSEQ, and IsNG().

IsNG() [2/2]

bool IsNG

(

const CSeq_id &

id

)

Definition at line 2790 of file utilities.cpp.

References GetAccession(), and NStr::StartsWith().

Referenced by IsNG(), and CValidError_bioseq::ValidateSeqIds().

IsNTNCNWACAccession() [1/3]

bool IsNTNCNWACAccession

(

const CBioseq &

seq

)

Definition at line 2777 of file utilities.cpp.

References FOR_EACH_SEQID_ON_BIOSEQ, and IsNTNCNWACAccession().

IsNTNCNWACAccession() [2/3]

bool IsNTNCNWACAccession

(

const CSeq_id &

id

)

Definition at line 2766 of file utilities.cpp.

References GetAccession(), and IsNTNCNWACAccession().

IsNTNCNWACAccession() [3/3]

bool IsNTNCNWACAccession

(

const string &

acc

)

Definition at line 2755 of file utilities.cpp.

References NStr::StartsWith().

Referenced by GetBioseqEndWarning(), IsNTNCNWACAccession(), CValidError_bioseq::ValidateDelta(), CValidError_bioseq::ValidateSeqIds(), and CSingleFeatValidator::x_ValidateExceptText().

IsOrganelle() [1/2]

bool IsOrganelle

(

const CBioseq_Handle &

seq

)

Definition at line 2857 of file utilities.cpp.

References CSeqdesc_Base::e_Source, CBioSource_Base::GetGenome(), CSeqdesc_Base::GetSource(), IsOrganelle(), and CBioSource_Base::IsSetGenome().

IsOrganelle() [2/2]

bool IsOrganelle

(

int

genome

)

Definition at line 2831 of file utilities.cpp.

References CBioSource_Base::eGenome_apicoplast, CBioSource_Base::eGenome_chloroplast, CBioSource_Base::eGenome_chromatophore, CBioSource_Base::eGenome_chromoplast, CBioSource_Base::eGenome_cyanelle, CBioSource_Base::eGenome_hydrogenosome, CBioSource_Base::eGenome_kinetoplast, CBioSource_Base::eGenome_leucoplast, CBioSource_Base::eGenome_mitochondrion, CBioSource_Base::eGenome_nucleomorph, CBioSource_Base::eGenome_plastid, and CBioSource_Base::eGenome_proplastid.

Referenced by CSpliceProblems::CalculateSpliceProblems(), IsInOrganelleSmallGenomeSet(), CValidError_feat::IsIntronShort(), IsOrganelle(), CAdjustForConsensusSpliceSite::s_IsBioseqGood_AdjustEnds(), CIntronValidator::Validate(), CValidError_bioseq::ValidateFeatPartialInContext(), CMRnaForCDS::x_ChkAndMakeMRna(), CIntronValidator::x_IsIntronShort(), CValidError_bioseq::x_ReportStartStopPartialProblem(), CValidError_bioseq::x_SetupCommonFlags(), and CValidError_bioseq::x_ValidateAbuttingRNA().

IsResidue()

bool IsResidue ( unsigned char  residue )

inline

Definition at line 88 of file utilities.hpp.

Referenced by s_PartialAtGapOrNs(), CIntronValidator::Validate(), CValidError_bioseq::ValidateRawConst(), CCDSTranslationProblems::x_Is5AtEndSpliceSiteOrGap(), CValidError_bioseq::x_IsPartialAtSpliceSiteOrGap(), and CValidError_graph::x_ValidateGraphValues().

IsSequenceFetchable() [1/2]

bool IsSequenceFetchable	(	const CSeq_id &	id,
		CScope *	scope = nullptr
	)

Definition at line 2714 of file utilities.cpp.

References CScope::AddDefaults(), CScope::GetBioseqHandle(), CSeq_id_Handle::GetHandle(), CScope::GetIds(), and CObjectManager::GetInstance().

Referenced by IsSequenceFetchable(), and CValidError_feat::ValidateInferenceAccession().

IsSequenceFetchable() [2/2]

bool IsSequenceFetchable	(	const string &	seq_id,
		CScope *	scope = nullptr
	)

Definition at line 2740 of file utilities.cpp.

References IsSequenceFetchable().

IsSpecificHostValid()

bool IsSpecificHostValid	(	const string &	host,
		string &	error_msg
	)

returns true and error_msg will be empty, if specific host is valid returns true and error_msg will be "Host is empty", if specific host is empty returns false if specific host is invalid

Definition at line 2182 of file utilities.cpp.

References CTaxValidationAndCleanup::IsOneSpecificHostValid(), and val.

Referenced by BOOST_AUTO_TEST_CASE(), and CMakeBadSpecificHostTable::x_CheckSpecificHosts().

IsTemporary()

bool IsTemporary

(

const CSeq_id &

id

)

Definition at line 2817 of file utilities.cpp.

References NStr::EqualNocase().

Referenced by CValidError_imp::ValidateSeqLocIds(), and CValidError_bioseq::x_CheckGeneralIDs().

MakeSeqVectorForResidueCounting()

CRef<CSeqVector> MakeSeqVectorForResidueCounting

(

const CBioseq_Handle &

bsh

)

Definition at line 2455 of file utilities.cpp.

References CSeq_data_Base::e_Ncbieaa, CSeq_data_Base::e_Ncbistdaa, CSeq_data_Base::e_not_set, CBioseq_Handle::eCoding_Iupac, CBioseq_Handle::GetInst(), CSeq_inst_Base::GetSeq_data(), CSeq_inst_Base::IsSetSeq_data(), CSeqVector::SetCoding(), and CSeq_data_Base::Which().

Referenced by HasBadProteinStart(), HasStopInProtein(), and CValidError_bioseq::ValidateRawConst().

PartialsSame()

bool PartialsSame	(	const CSeq_loc &	loc1,
		const CSeq_loc &	loc2
	)

Definition at line 1463 of file utilities.cpp.

References eExtreme_Biological, CSeq_loc::IsPartialStart(), and CSeq_loc::IsPartialStop().

Referenced by IsDuplicate(), CValidError_bioseq::x_ReportDupOverlapFeaturePair(), and CCdregionValidator::x_ValidateProductPartials().

ReportTranslationErrors()

bool ReportTranslationErrors

(

const string &

except_text

)

Definition at line 2341 of file utilities.cpp.

References NStr::FindNoCase(), ITERATE, and NPOS.

Referenced by CCDSTranslationProblems::CalculateTranslationProblems(), HasBadStartCodon(), and HasInternalStop().

s_FeatureIdsMatch()

bool s_FeatureIdsMatch	(	const CFeat_id &	f1,
		const CFeat_id &	f2
	)

Definition at line 717 of file utilities.cpp.

References CObject_id::Compare(), CFeat_id_Base::GetLocal(), and CFeat_id_Base::IsLocal().

Referenced by s_IdXrefsAreReciprocal(), s_IdXrefsNotReciprocal(), and CValidError_bioseq::x_IdXrefsNotReciprocal().

s_IdXrefsAreReciprocal()

bool s_IdXrefsAreReciprocal	(	const CSeq_feat &	cds,
		const CSeq_feat &	mrna
	)

Definition at line 6595 of file validerror_bioseq.cpp.

References FOR_EACH_SEQFEATXREF_ON_SEQFEAT, CSeq_feat_Base::GetId(), CFeat_id_Base::IsLocal(), CSeq_feat_Base::IsSetId(), match(), and s_FeatureIdsMatch().

Referenced by s_SetUpXrefPairs(), and CValidError_bioseq::x_IdXrefsAreReciprocal().

s_PartialAtGapOrNs()

bool s_PartialAtGapOrNs	(	CScope *	scope,
		const CSeq_loc &	loc,
		unsigned int	tag,
		bool	only_gap = false
	)

Definition at line 1176 of file utilities.cpp.

References CBioseq_Handle::eCoding_Iupac, eNa_strand_minus, eSeqlocPartial_Nostart, eSeqlocPartial_Nostop, first(), CScope::GetBioseqHandle(), CRange_Base::GetFrom(), CSeq_loc::GetId(), CSeq_loc_CI::GetRange(), CSeq_loc_CI::GetRangeAsSeq_loc(), CBioseq_Handle::GetSeqVector(), CSeq_loc_CI::GetStrand(), CRange_Base::GetTo(), isalpha(), CSeqVector::IsInGap(), IsResidue(), last(), len, result, CSeqVector::size(), swap(), and tag.

Referenced by CValidError_bioseq::ValidateFeatPartialInContext(), and CCDSTranslationProblems::x_CheckCDSFrame().

s_StringHasPMID()

bool s_StringHasPMID

(

const string &

str

)

Definition at line 727 of file utilities.cpp.

References NStr::Find(), NStr::IsBlank(), isdigit(), and str().

Referenced by CValidError_desc::ValidateTitle(), and CProtValidator::x_ValidateProteinName().

SeqIsPatent() [1/2]

bool SeqIsPatent

(

const CBioseq &

seq

)

Definition at line 1155 of file utilities.cpp.

References FOR_EACH_SEQID_ON_BIOSEQ.

Referenced by CValidError_imp::AddBioseqWithNoBiosource(), CValidError_imp::AddProtWithoutFullRef(), CValidError_bioseq::CheckForMissingChromosome(), SeqIsPatent(), CValidError_bioseq::ValidateBioseqContext(), CValidError_bioseq::ValidateNsAndGaps(), CValidError_bioseq::ValidateSeqDescContext(), CValidError_bioseq::ValidateSeqFeatContext(), and CValidError_bioseq::ValidateSeqIds().

SeqIsPatent() [2/2]

bool SeqIsPatent

(

const CBioseq_Handle &

seq

)

Definition at line 1170 of file utilities.cpp.

References CBioseq_Handle::GetCompleteBioseq(), and SeqIsPatent().

ShouldCheckForNsAndGap()

bool ShouldCheckForNsAndGap

(

const CBioseq_Handle &

bsh

)

Definition at line 1307 of file utilities.cpp.

References CSeq_inst_Base::eTopology_circular, CBioseq_Handle::GetInst_Length(), CBioseq_Handle::GetInst_Topology(), and CBioseq_Handle::IsSetInst_Topology().

Referenced by CheckBioseqEndsForNAndGap(), s_WillReportTerminalGap(), and CValidError_bioseq::ValidateNsAndGaps().

SpecificHostValueToCheck()

string SpecificHostValueToCheck

(

const string &

val

)

Definition at line 2042 of file utilities.cpp.

References AdjustSpecificHostForTaxServer(), NStr::Find(), NStr::IsBlank(), isupper(), kEmptyStr, NStr::StartsWith(), NStr::TruncateSpacesInPlace(), and val.

Referenced by CSpecificHostRequest::CSpecificHostRequest().

TranslateCodingRegionForValidation()

string TranslateCodingRegionForValidation	(	const CSeq_feat &	feat,
		CScope &	scope,
		bool &	alt_start
	)

Definition at line 2266 of file utilities.cpp.

References CSerialObject::Assign(), FixGeneticCode(), CScope::GetBioseqHandle(), CSeqFeatData_Base::GetCdregion(), CCdregion_Base::GetCode(), CSeq_feat_Base::GetData(), CCdregion_Base::GetFrame(), CBioseq_Handle::GetInst_Length(), CSeq_feat_Base::GetLocation(), CSeq_loc_Base::GetWhole(), CCdregion_Base::IsSetCode(), CCdregion_Base::IsSetFrame(), CSeq_loc_Base::IsWhole(), kEmptyStr, CSeq_feat_Base::SetData(), tmp, and CSeqTranslator::Translate().

Referenced by CCDSTranslationProblems::CalculateTranslationProblems(), HasBadStartCodon(), HasInternalStop(), HasNoStop(), and CCDSTranslationProblems::x_IsThreeBaseNonsense().

ValidateAccessionString()

EAccessionFormatError ValidateAccessionString	(	const string &	accession,
		bool	require_version
	)

Definition at line 624 of file utilities.cpp.

References eAccessionFormat_bad_version, eAccessionFormat_missing_version, eAccessionFormat_no_start_letters, eAccessionFormat_null, eAccessionFormat_too_long, eAccessionFormat_valid, eAccessionFormat_wrong_number_of_digits, isalpha(), NStr::IsBlank(), isdigit(), isupper(), NStr::StartsWith(), and str().

Referenced by CValidError_bioseq::ValidateSeqId(), CSingleFeatValidator::x_ValidateCompareVal(), and CSingleFeatValidator::x_ValidateSeqFeatProduct().