#include <objtools/cleanup/cleanup.hpp>

Inheritance diagram for CCleanup:

Collaboration diagram for CCleanup:

Public Types
enum	EValidOptions { eClean_NoReporting = 0x1 , eClean_GpipeMode = 0x2 , eClean_NoNcbiUserObjects = 0x4 , eClean_SyncGenCodes = 0x8 , eClean_NoProteinTitles = 0x10 , eClean_KeepTopSet = 0x20 , eClean_KeepSingleSeqSet = 0x40 , eClean_InHugeSeqSet = 0x80 }

enum	EScopeOptions { eScope_Copy , eScope_UseInPlace }

using	TChanges = CConstRef< CCleanupChange >

typedef pair< CSeq_feat_Handle, CSeq_feat_Handle >	TFeatGenePair

Public Types inherited from CObject
enum	EAllocFillMode { eAllocFillNone = 1 , eAllocFillZero , eAllocFillPattern }
	Control filling of newly allocated memory. More...

typedef CObjectCounterLocker	TLockerType
	Default locker type for CRef. More...

typedef atomic< Uint8 >	TCounter
	Counter type is CAtomiCounter. More...

typedef Uint8	TCount
	Alias for value type of counter. More...

Public Member Functions
	CCleanup (CScope *scope=nullptr, EScopeOptions scope_handling=eScope_Copy)

	CCleanup (const CCleanup &)=delete

CCleanup &	operator= (const CCleanup &)=delete

	~CCleanup ()

void	SetScope (CScope *scope)

TChanges	BasicCleanup (CSeq_entry &se, Uint4 options=0)

TChanges	BasicCleanup (CSeq_submit &ss, Uint4 options=0)
	Cleanup a Seq-submit. More...

TChanges	BasicCleanup (CBioseq_set &bss, Uint4 options=0)
	Cleanup a Bioseq_set. More...

TChanges	BasicCleanup (CSeq_annot &sa, Uint4 options=0)
	Cleanup a Seq-Annot. More...

TChanges	BasicCleanup (CSeq_feat &sf, Uint4 options=0)
	Cleanup a Seq-feat. More...

TChanges	BasicCleanup (CBioSource &src, Uint4 options=0)
	Cleanup a BioSource. More...

TChanges	BasicCleanup (CSubmit_block &block, Uint4 options=0)

TChanges	BasicCleanup (CSeqdesc &desc, Uint4 options=0)

TChanges	BasicCleanup (CSeq_descr &desc, Uint4 options=0)

TChanges	BasicCleanup (CSeq_entry_Handle &seh, Uint4 options=0)

TChanges	BasicCleanup (CBioseq_Handle &bsh, Uint4 options=0)

TChanges	BasicCleanup (CBioseq_set_Handle &bssh, Uint4 options=0)

TChanges	BasicCleanup (CSeq_annot_Handle &sak, Uint4 options=0)

TChanges	BasicCleanup (CSeq_feat_Handle &sfh, Uint4 options=0)

TChanges	ExtendedCleanup (CSeq_entry &se, Uint4 options=0)
	Cleanup a Seq-entry. More...

TChanges	ExtendedCleanup (CSeq_submit &ss, Uint4 options=0)
	Cleanup a Seq-submit. More...

TChanges	ExtendedCleanup (CSeq_annot &sa, Uint4 options=0)
	Cleanup a Seq-Annot. More...

Public Member Functions inherited from CObject
	CObject (void)
	Constructor. More...

	CObject (const CObject &src)
	Copy constructor. More...

virtual	~CObject (void)
	Destructor. More...

CObject &	operator= (const CObject &src) THROWS_NONE
	Assignment operator. More...

bool	CanBeDeleted (void) const THROWS_NONE
	Check if object can be deleted. More...

bool	IsAllocatedInPool (void) const THROWS_NONE
	Check if object is allocated in memory pool (not system heap) More...

bool	Referenced (void) const THROWS_NONE
	Check if object is referenced. More...

bool	ReferencedOnlyOnce (void) const THROWS_NONE
	Check if object is referenced only once. More...

void	AddReference (void) const
	Add reference to object. More...

void	RemoveReference (void) const
	Remove reference to object. More...

void	ReleaseReference (void) const
	Remove reference without deleting object. More...

virtual void	DoNotDeleteThisObject (void)
	Mark this object as not allocated in heap – do not delete this object. More...

virtual void	DoDeleteThisObject (void)
	Mark this object as allocated in heap – object can be deleted. More...

void *	operator new (size_t size)
	Define new operator for memory allocation. More...

void *	operator new[] (size_t size)
	Define new[] operator for 'array' memory allocation. More...

void	operator delete (void *ptr)
	Define delete operator for memory deallocation. More...

void	operator delete[] (void *ptr)
	Define delete[] operator for memory deallocation. More...

void *	operator new (size_t size, void *place)
	Define new operator. More...

void	operator delete (void ptr, void place)
	Define delete operator. More...

void *	operator new (size_t size, CObjectMemoryPool *place)
	Define new operator using memory pool. More...

void	operator delete (void ptr, CObjectMemoryPool place)
	Define delete operator. More...

virtual void	DebugDump (CDebugDumpContext ddc, unsigned int depth) const
	Define method for dumping debug information. More...

Public Member Functions inherited from CDebugDumpable
	CDebugDumpable (void)

virtual	~CDebugDumpable (void)

void	DebugDumpText (ostream &out, const string &bundle, unsigned int depth) const

void	DebugDumpFormat (CDebugDumpFormatter &ddf, const string &bundle, unsigned int depth) const

void	DumpToConsole (void) const

Static Public Member Functions
static TChanges	ExtendedCleanup (CSeq_entry_Handle &seh, Uint4 options=0)

static bool	ShouldStripPubSerial (const CBioseq &bs)

static bool	MoveProteinSpecificFeats (CSeq_entry_Handle seh)
	Moves protein-specific features from nucleotide sequences in the Seq-entry to the appropriate protein sequence. More...

static bool	MoveFeatToProtein (CSeq_feat_Handle fh)
	Moves one feature from nucleotide bioseq to the appropriate protein sequence. More...

static bool	IsGeneXrefUnnecessary (const CSeq_feat &sf, CScope &scope, const CGene_ref &gene_xref)
	Calculates whether a Gene-xref is unnecessary (because it refers to the same gene as would be calculated using overlap) More...

static bool	RemoveUnnecessaryGeneXrefs (CSeq_feat &f, CScope &scope)
	Removes unnecessary Gene-xrefs. More...

static bool	RemoveUnnecessaryGeneXrefs (CSeq_entry_Handle seh)
	Removes unnecessary Gene-xrefs on features in Seq-entry. More...

static bool	RemoveNonsuppressingGeneXrefs (CSeq_feat &f)
	Removes non-suppressing Gene-xrefs. More...

static bool	RepairXrefs (const CSeq_feat &f, const CTSE_Handle &tse)
	Repairs non-reciprocal xref pairs for specified feature if xrefs between subtypes are permitted and feature with missing xref does not have an xref to a different feature of the same subtype. More...

static bool	RepairXrefs (const CSeq_feat &src, CSeq_feat_Handle &dst, const CTSE_Handle &tse)
	Repairs non-reciprocal xref pairs for specified feature pair if xrefs between subtypes are permitted and feature with missing xref does not have an xref to a different feature of the same subtype. More...

static bool	RepairXrefs (CSeq_entry_Handle seh)
	Repairs non-reciprocal xref pairs in specified seq-entry. More...

static bool	FindMatchingLocusGene (CSeq_feat &f, const CGene_ref &gene_xref, CBioseq_Handle bsh)
	Detects gene features with matching locus. More...

static bool	RemoveOrphanLocusGeneXrefs (CSeq_feat &f, CBioseq_Handle bsh)
	Removes orphaned locus Gene-xrefs. More...

static bool	FindMatchingLocus_tagGene (CSeq_feat &f, const CGene_ref &gene_xref, CBioseq_Handle bsh)
	Detects gene features with matching locus_tag. More...

static bool	RemoveOrphanLocus_tagGeneXrefs (CSeq_feat &f, CBioseq_Handle bsh)
	Removes orphaned locus_tag Gene-xrefs. More...

static bool	SeqLocExtend (CSeq_loc &loc, size_t pos, CScope &scope)
	Extends a location to the specificed position. More...

static bool	ExtendToStopIfShortAndNotPartial (CSeq_feat &f, CBioseq_Handle bsh, bool check_for_stop=true)
	Extends a coding region up to 50 nt. More...

static bool	LocationMayBeExtendedToMatch (const CSeq_loc &orig, const CSeq_loc &improved)
	Checks whether it is possible to extend the original location up to improved one. More...

static bool	ExtendToStopCodon (CSeq_feat &f, CBioseq_Handle bsh, size_t limit)
	Extends a feature up to limit nt to a stop codon, or to the end of the sequence if limit == 0 (partial will be set if location extends to end of sequence but no stop codon is found) More...

static bool	ExtendStopPosition (CSeq_feat &f, const CSeq_feat *cdregion, size_t extension=0)

static bool	SetBestFrame (CSeq_feat &cds, CScope &scope)
	Translates coding region and selects best frame (without stops, or longest) More...

static bool	SetFrameFromLoc (CCdregion &cdregion, const CSeq_loc &loc, CScope &scope)
	Chooses best frame based on location 1. More...

static bool	SetFrameFromLoc (CCdregion::EFrame &frame, const CSeq_loc &loc, CScope &scope)

static bool	SetCDSPartialsByFrameAndTranslation (CSeq_feat &cds, CScope &scope)
	1. More...

static bool	ClearInternalPartials (CSeq_loc &loc, bool is_first=true, bool is_last=true)
	Clear internal partials. More...

static bool	ClearInternalPartials (CSeq_loc_mix &mix, bool is_first=true, bool is_last=true)

static bool	ClearInternalPartials (CPacked_seqint &pint, bool is_first=true, bool is_last=true)

static bool	ClearInternalPartials (CSeq_entry_Handle seh)

static bool	SetFeaturePartial (CSeq_feat &f)
	Set feature partial based on feature location. More...

static bool	UpdateECNumbers (CProt_ref::TEc &ec_num_list)
	Update EC numbers. More...

static bool	RemoveBadECNumbers (CProt_ref::TEc &ec_num_list)
	Delete EC numbers. More...

static bool	FixECNumbers (CSeq_entry_Handle entry)
	Fix EC numbers. More...

static bool	SetGenePartialByLongestContainedFeature (CSeq_feat &gene, CScope &scope)
	Set partialness of gene to match longest feature contained in gene. More...

static void	SetProteinName (CProt_ref &prot, const string &protein_name, bool append)

static void	SetProteinName (CSeq_feat &cds, const string &protein_name, bool append, CScope &scope)

static void	SetMrnaName (CSeq_feat &mrna, const string &protein_name)

static const string &	GetProteinName (const CProt_ref &prot)

static const string &	GetProteinName (const CSeq_feat &cds, CSeq_entry_Handle seh)

static bool	SetMolinfoTech (CBioseq_Handle seq, CMolInfo::ETech tech)
	Sets MolInfo::tech for a sequence. More...

static bool	SetMolinfoBiomol (CBioseq_Handle seq, CMolInfo::EBiomol biomol)
	Sets MolInfo::biomol for a sequence. More...

static bool	AddMissingMolInfo (CBioseq &seq, bool is_product)
	Adds missing MolInfo descriptor to sequence. More...

static bool	AddProteinTitle (CBioseq_Handle bsh)
	Creates missing protein title descriptor. More...

static bool	RemoveNcbiCleanupObject (CSeq_entry &seq_entry)
	Removes NcbiCleanup User Objects in the Seq-entry. More...

static void	AddNcbiCleanupObject (int ncbi_cleanup_version, CSeq_descr &descr)
	Adds NcbiCleanup User Object to Seq-descr. More...

static bool	TaxonomyLookup (CSeq_entry_Handle seh)
	Looks up Org-refs in the Seq-entry. More...

static bool	SetGeneticCodes (CBioseq_Handle bsh)
	Sets genetic codes for coding regions on Bioseq-Handle. More...

static bool	AddPartialToProteinTitle (CBioseq &bioseq)
	Adjusts protein title to reflect partialness. More...

static bool	RemovePseudoProduct (CSeq_feat &cds, CScope &scope)
	Removes protein product from pseudo coding region. More...

static CRef< CSeq_entry >	AddProtein (const CSeq_feat &cds, CScope &scope)

static bool	ExpandGeneToIncludeChildren (CSeq_feat &gene, CTSE_Handle &tse)
	Expands gene to include features it cross-references. More...

static bool	WGSCleanup (CSeq_entry_Handle entry, bool instantiate_missing_proteins=true, Uint4 options=0, bool run_extended_cleanup=true)
	Performs WGS specific cleanup. More...

static bool	AddLowQualityException (CSeq_entry_Handle entry)
	For table2asn -c s Adds an exception of "low-quality sequence region" to coding regions and mRNAs that are not pseudo and have an intron <11bp in length. More...

static bool	NormalizeDescriptorOrder (CSeq_descr &descr)
	Normalize Descriptor Order on a specific Seq-entry. More...

static bool	NormalizeDescriptorOrder (CSeq_entry_Handle seh)
	Normalize Descriptor Order on a specific Seq-entry. More...

static bool	RemoveUnseenTitles (CSeq_entry_EditHandle::TSeq seq)
	Remove all titles in Seqdescr except the last, because it is the only one that would be displayed in the flatfile. More...

static bool	RemoveUnseenTitles (CSeq_entry_EditHandle::TSet set)
	Remove all titles in Seqdescr except the last, because it is the only one that would be displayed in the flatfile. More...

static bool	AddGenBankWrapper (CSeq_entry_Handle seh)
	Add GenBank Wrapper Set. More...

static void	GetPubdescLabels (const CPubdesc &pd, vector< TEntrezId > &pmids, vector< TEntrezId > &muids, vector< int > &serials, vector< string > &published_labels, vector< string > &unpublished_labels)
	For Publication Citations Get labels for a pubdesc. More...

static vector< CConstRef< CPub > >	GetCitationList (CBioseq_Handle bsh)
	Get list of pubs that can be used for citations for Seq-feat on a Bioseq-handle. More...

static bool	RemoveDuplicatePubs (CSeq_descr &descr)
	Remove duplicate publications. More...

static bool	OkToPromoteNpPub (const CPubdesc &pd)
	Some pubs should not be promoted to nuc-prot set from sequence. More...

static bool	OkToPromoteNpPub (const CBioseq &b)
	For some sequences, pubs should not be promoted to nuc-prot set from sequence. More...

static bool	PubAlreadyInSet (const CPubdesc &pd, const CSeq_descr &descr)

static bool	ConvertPubFeatsToPubDescs (CSeq_entry_Handle seh)
	Convert full-length publication features to publication descriptors. More...

static bool	RescueSiteRefPubs (CSeq_entry_Handle seh)
	Rescue pubs from Site-ref features. More...

static bool	IsMinPub (const CPubdesc &pd, bool is_refseq_prot)
	Is this a "minimal" pub? (If yes, do not rescue from a Seq-feat.cit) More...

static void	MoveOneFeatToPubdesc (CSeq_feat_Handle feat, CRef< CSeqdesc > d, CBioseq_Handle b, bool remove_feat=true)

static bool	RemoveDupBioSource (CSeq_descr &descr)
	Remove duplicate biosource descriptors. More...

static CRef< CBioSource >	BioSrcFromFeat (const CSeq_feat &f)
	Get BioSource from feature to use for source descriptor. More...

static bool	AreBioSourcesMergeable (const CBioSource &src1, const CBioSource &src2)

static bool	MergeDupBioSources (CSeq_descr &descr)

static bool	MergeDupBioSources (CBioSource &src1, const CBioSource &add)

static bool	ConvertSrcFeatsToSrcDescs (CSeq_entry_Handle seh)
	Convert full-length source features to source descriptors. More...

static bool	FixGeneXrefSkew (CSeq_entry_Handle seh)
	Examine all genes and gene xrefs in the Seq-entry. More...

static bool	RenormalizeNucProtSets (CSeq_entry_Handle seh)
	Convert nuc-prot sets with just one sequence to just the sequence can't be done during the explore phase because it changes a seq to a set. More...

static bool	DecodeXMLMarkChanged (std::string &str)
	decodes various tags, including carriage-return-line-feed constructs More...

static CRef< CSeq_loc >	GetProteinLocationFromNucleotideLocation (const CSeq_loc &nuc_loc, CScope &scope)

static CRef< CSeq_loc >	GetProteinLocationFromNucleotideLocation (const CSeq_loc &nuc_loc, const CSeq_feat &cds, CScope &scope, bool require_inframe=false)

static bool	RepackageProteins (CSeq_entry_Handle seh)
	Find proteins that are not packaged in the same nuc-prot set as the coding region for which they are a product, and move them to that nuc-prot set. More...

static bool	RepackageProteins (const CSeq_feat &cds, CBioseq_set_Handle np)

static bool	ConvertDeltaSeqToRaw (CSeq_entry_Handle seh, CSeq_inst::EMol filter=CSeq_inst::eMol_not_set)

static bool	ParseCodeBreak (const CSeq_feat &feat, CCdregion &cds, const CTempString &str, CScope &scope, IObjtoolsListener *pMessageListener=nullptr)
	Parse string into code break and add to coding region. More...

static bool	ParseCodeBreaks (CSeq_feat &feat, CScope &scope)
	Parses all valid transl_except Gb-quals into code-breaks for cdregion, then removes the transl_except Gb-quals that were successfully parsed. More...

static size_t	MakeSmallGenomeSet (CSeq_entry_Handle entry)

static bool	MakeIRDFeatsFromSourceXrefs (CSeq_entry_Handle entry)
	From SQD-4329 For each sequence with a source that has an IRD db_xref, create a misc_feature across the entire span and move the IRD db_xref from the source to the misc_feature. More...

static bool	FixRNAEditingCodingRegion (CSeq_feat &cds)
	From GB-7563 An action has been requested that will do the following: 1. More...

static void	SetCodeBreakLocation (CCode_break &cb, size_t pos, const CSeq_feat &cds)
	utility function for setting code break location given offset pos is the position of the amino acid where the translation exception occurs (starts with 1) More...

static bool	IsMethionine (const CCode_break &cb)

static CConstRef< CCode_break >	GetCodeBreakForLocation (size_t pos, const CSeq_feat &cds)
	utility function for finding the code break for a given amino acid position pos is the position of the amino acid where the translation exception occurs (starts with 1) More...

static bool	NormalizeGeneQuals (CSeq_feat &cds, CSeq_feat &gene)

static bool	NormalizeGeneQuals (CBioseq_Handle bsh)

static bool	NormalizeGeneQuals (CSeq_entry_Handle seh)

static vector< TFeatGenePair >	GetNormalizableGeneQualPairs (CBioseq_Handle bsh)

static bool	CleanupUserObject (CUser_object &object)

static bool	CleanupAuthor (CAuthor &author, bool fix_initials=true)

static bool	CleanupAuthList (CAuth_list &al, bool fix_initials=true)

static void	ResetAuthorNames (CAuth_list::TNames &names)

static bool	CleanupAffil (CAffil &af)

static bool	IsEmpty (const CAuth_list::TAffil &affil)

static bool	CleanupCollectionDates (CSeq_entry_Handle seh, bool month_first)

static void	AutodefId (CSeq_entry_Handle seh)

static char	ValidAminoAcid (string_view abbrev)

Static Public Member Functions inherited from CObject
static NCBI_XNCBI_EXPORT void	ThrowNullPointerException (void)
	Define method to throw null pointer exception. More...

static NCBI_XNCBI_EXPORT void	ThrowNullPointerException (const type_info &type)

static EAllocFillMode	GetAllocFillMode (void)

static void	SetAllocFillMode (EAllocFillMode mode)

static void	SetAllocFillMode (const string &value)
	Set mode from configuration parameter value. More...

Static Public Member Functions inherited from CDebugDumpable
static void	EnableDebugDump (bool on)

Static Private Member Functions
static bool	x_CleanupUserField (CUser_field &field)

static bool	x_MergeDupOrgNames (COrgName &on1, const COrgName &add)

static bool	x_MergeDupOrgRefs (COrg_ref &org1, const COrg_ref &add)

static bool	x_HasShortIntron (const CSeq_loc &loc, size_t min_len=11)

static bool	x_AddLowQualityException (CSeq_feat &feat)

static bool	x_AddLowQualityException (CSeq_entry_Handle entry, CSeqFeatData::ESubtype subtype)

static bool	s_IsProductOnFeat (const CSeq_feat &cds)

static void	s_SetProductOnFeat (CSeq_feat &feat, const string &protein_name, bool append)

static bool	s_CleanupGeneOntology (CUser_object &obj)

static bool	s_CleanupStructuredComment (CUser_object &obj)

static bool	s_RemoveEmptyFields (CUser_object &obj)

static bool	s_CleanupGenomeAssembly (CUser_object &obj)

static bool	s_CleanupDBLink (CUser_object &obj)

static bool	s_AddNumToUserField (CUser_field &field)

static bool	s_CleanupNameStdBC (CName_std &name, bool fix_initials)

static void	s_ExtractSuffixFromInitials (CName_std &name)

static void	s_FixEtAl (CName_std &name)

static bool	s_Flatten (CPub_equiv &pub_equiv)

Private Attributes
CRef< CScope >	m_Scope

Additional Inherited Members
Static Public Attributes inherited from CObject
static const TCount	eCounterBitsCanBeDeleted = 1 << 0
	Define possible object states. More...

static const TCount	eCounterBitsInPlainHeap = 1 << 1
	Heap signature was found. More...

static const TCount	eCounterBitsPlaceMask
	Mask for 'in heap' state flags. More...

static const int	eCounterStep = 1 << 2
	Skip over the "in heap" bits. More...

static const TCount	eCounterValid = TCount(1) << (sizeof(TCount) * 8 - 2)
	Minimal value for valid objects (reference counter is zero) Must be a single bit value. More...

static const TCount	eCounterStateMask
	Valid object, and object in heap. More...

Protected Member Functions inherited from CObject
virtual void	DeleteThis (void)
	Virtual method "deleting" this object. More...

Detailed Description

Definition at line 68 of file cleanup.hpp.

Member Typedef Documentation

◆ TChanges

using CCleanup::TChanges = CConstRef<CCleanupChange>

Definition at line 97 of file cleanup.hpp.

◆ TFeatGenePair

typedef pair<CSeq_feat_Handle, CSeq_feat_Handle> CCleanup::TFeatGenePair

Definition at line 590 of file cleanup.hpp.

Member Enumeration Documentation

◆ EScopeOptions

enum CCleanup::EScopeOptions

Enumerator
eScope_Copy
eScope_UseInPlace

Definition at line 83 of file cleanup.hpp.

◆ EValidOptions

enum CCleanup::EValidOptions

Enumerator
eClean_NoReporting
eClean_GpipeMode
eClean_NoNcbiUserObjects
eClean_SyncGenCodes
eClean_NoProteinTitles
eClean_KeepTopSet
eClean_KeepSingleSeqSet
eClean_InHugeSeqSet

Definition at line 72 of file cleanup.hpp.

Constructor & Destructor Documentation

◆ CCleanup() [1/2]

CCleanup::CCleanup	(	CScope *	scope = `nullptr`,
		EScopeOptions	scope_handling = `eScope_Copy`
	)

Definition at line 89 of file cleanup.cpp.

References CScope::AddScope(), eScope_UseInPlace, CObjectManager::GetInstance(), and m_Scope.

◆ CCleanup() [2/2]

CCleanup::CCleanup ( const CCleanup & )

delete

◆ ~CCleanup()

CCleanup::~CCleanup ( void )

Definition at line 103 of file cleanup.cpp.

Member Function Documentation

◆ AddGenBankWrapper()

bool CCleanup::AddGenBankWrapper ( CSeq_entry_Handle seh )

static

Add GenBank Wrapper Set.

Parameters

entry Seq-entry to edit

Returns: Boolean return value indicates whether object changed

Definition at line 3065 of file cleanup.cpp.

References CSeq_entry_EditHandle::ConvertSeqToSet(), CBioseq_set_Base::eClass_genbank, CBioseq_set_Handle::GetClass(), CSeq_entry_Handle::GetSet(), CSeq_entry_Handle::IsSet(), and CBioseq_set_Handle::IsSetClass().

◆ AddLowQualityException()

bool CCleanup::AddLowQualityException ( CSeq_entry_Handle entry )

static

For table2asn -c s Adds an exception of "low-quality sequence region" to coding regions and mRNAs that are not pseudo and have an intron <11bp in length.

Parameters

entry Seq-entry to edit

Returns: Boolean return value indicates whether object was updated

Definition at line 2931 of file cleanup.cpp.

References CSeqFeatData::eSubtype_cdregion, CSeqFeatData::eSubtype_mRNA, and x_AddLowQualityException().

Referenced by CTable2AsnValidator::Cleanup().

◆ AddMissingMolInfo()

bool CCleanup::AddMissingMolInfo	(	CBioseq &	seq,
		bool	is_product
	)

static

Adds missing MolInfo descriptor to sequence.

Parameters

seq	Bioseq to edit

Returns: Boolean return value indicates whether descriptor was added

Definition at line 1824 of file cleanup.cpp.

References CMolInfo_Base::eBiomol_mRNA, CMolInfo_Base::eBiomol_peptide, CMolInfo_Base::eBiomol_unknown, CSeq_inst_Base::eMol_rna, CMolInfo_Base::eTech_concept_trans, CMolInfo_Base::eTech_standard, CBioseq_Base::GetInst(), CSeq_inst_Base::GetMol(), CBioseq::IsAa(), CBioseq_Base::IsSetDescr(), CBioseq_Base::IsSetInst(), CSeq_inst_Base::IsSetMol(), NON_CONST_ITERATE, CMolInfo_Base::SetBiomol(), CBioseq_Base::SetDescr(), CSeqdesc_Base::SetMolinfo(), and CMolInfo_Base::SetTech().

Referenced by CNewCleanup_imp::CreateMissingMolInfo().

◆ AddNcbiCleanupObject()

void CCleanup::AddNcbiCleanupObject	(	int	ncbi_cleanup_version,
		CSeq_descr &	descr
	)

static

Adds NcbiCleanup User Object to Seq-descr.

Definition at line 1929 of file cleanup.cpp.

References CUser_object::eObjectType_Cleanup, CSeq_descr_Base::IsSet(), Ref(), and CSeq_descr_Base::Set().

Referenced by CNewCleanup_imp::AddNcbiCleanupObject(), and CCleanupHugeAsnReader::x_CleanupTopLevelDescriptors().

◆ AddPartialToProteinTitle()

bool CCleanup::AddPartialToProteinTitle ( CBioseq & bioseq )

static

Adjusts protein title to reflect partialness.

Parameters

Bioseq to adjust

Returns: Boolean indicates whether title was updated

Definition at line 2371 of file cleanup.cpp.

Referenced by CNewCleanup_imp::x_AddPartialToProteinTitle().

◆ AddProtein()

CRef< CSeq_entry > CCleanup::AddProtein	(	const CSeq_feat &	cds,
		CScope &	scope
	)

static

Definition at line 2051 of file cleanup.cpp.

References CSeq_entry_EditHandle::AddSeqdesc(), CSeq_id::Assign(), CSeq_entry_EditHandle::AttachEntry(), CSeq_entry_EditHandle::ConvertSeqToSet(), copy(), CMolInfo_Base::eBiomol_peptide, CBioseq_set_Base::eClass_nuc_prot, CRef< C, Locker >::Empty(), CMolInfo_Base::eTech_concept_trans, CScope::GetBioseqHandle(), CBioseq_set_Handle::GetClass(), CBioseq_set_Handle::GetCompleteBioseq_set(), CSeq_entry_Handle::GetEditHandle(), CSeq_loc::GetId(), CSeq_feat_Base::GetLocation(), CSeq_entry_EditHandle::GetParentBioseq_set(), CBioseq_set_Handle::GetParentEntry(), CSeq_feat_Base::GetProduct(), CSeq_entry_Handle::GetScope(), CBioseq_Handle::GetSeq_entry_Handle(), CScope::GetSeq_entryEditHandle(), CSeq_entry_Handle::GetSet(), CSeq_entry_Handle::IsSet(), CBioseq_set_Handle::IsSetClass(), CSeq_feat_Base::IsSetProduct(), nuc, CMolInfo_Base::SetBiomol(), CBioseq_Base::SetDescr(), CBioseq_Base::SetId(), CSeqdesc_Base::SetMolinfo(), CSeq_entry_Base::SetSeq(), CSeq_entry_EditHandle::SetSet(), CMolInfo_Base::SetTech(), and CSeqTranslator::TranslateToProtein().

Referenced by g_InstantiateMissingProteins(), and WGSCleanup().

◆ AddProteinTitle()

bool CCleanup::AddProteinTitle ( CBioseq_Handle bsh )

static

Creates missing protein title descriptor.

Parameters

seq	Bioseq to edit

Returns: Boolean return value indicates whether title was added

Definition at line 1865 of file cleanup.cpp.

References CSeq_id_Base::e_Patent, CSeq_id_Base::e_Pdb, CSeq_id_Base::e_Pir, CSeq_id_Base::e_Prf, CSeq_id_Base::e_Swissprot, CSeqdesc_Base::e_Title, CBioseq_Handle::GetEditHandle(), CBioseq_Handle::GetId(), CBioseq_Handle::GetInst(), CBioseq_Handle::IsAa(), CBioseq_Handle::IsSetId(), CBioseq_Handle::IsSetInst(), CSeq_inst_Base::IsSetMol(), ITERATE, CAutoAddDesc::Set(), CBioseq_EditHandle::SetDescr(), and CSeqdesc_Base::SetTitle().

Referenced by CNewCleanup_imp::AddProteinTitles(), and CTable2AsnValidator::Cleanup().

◆ AreBioSourcesMergeable()

bool CCleanup::AreBioSourcesMergeable	(	const CBioSource &	src1,
		const CBioSource &	src2
	)

static

Definition at line 3536 of file cleanup.cpp.

References NStr::Equal(), CBioSource_Base::GetOrg(), COrg_ref_Base::GetTaxname(), CBioSource_Base::IsSetOrg(), and COrg_ref_Base::IsSetTaxname().

Referenced by CMergeBiosources::apply(), MergeDupBioSources(), and CNewCleanup_imp::x_MoveNpSrc().

◆ AutodefId()

void CCleanup::AutodefId ( CSeq_entry_Handle seh )

static

Definition at line 4942 of file cleanup.cpp.

References CSeq_entry_EditHandle::AddSeqdesc(), CSerialObject::Assign(), b, CAutoDef::CreateIDOptions(), CSeqdesc_Base::e_User, CSeq_entry_Handle::GetEditHandle(), CAutoDef::RegenerateSequenceDefLines(), CSeq_entry_EditHandle::RemoveSeqdesc(), CSeqdesc_Base::SetUser(), and ud().

Referenced by CCleanupApp::x_ProcessXOptions().

◆ BasicCleanup() [1/14]

CCleanup::TChanges CCleanup::BasicCleanup	(	CBioseq_Handle &	bsh,
		Uint4	options = `0`
	)

Definition at line 198 of file cleanup.cpp.

References CNewCleanup_imp::BasicCleanupBioseqHandle(), CBioseq_Handle::GetScope(), makeCleanupChange(), and CNewCleanup_imp::SetScope().

◆ BasicCleanup() [2/14]

CCleanup::TChanges CCleanup::BasicCleanup	(	CBioseq_set &	bss,
		Uint4	options = `0`
	)

Cleanup a Bioseq_set.

Definition at line 156 of file cleanup.cpp.

References CLEANUP_SETUP.

◆ BasicCleanup() [3/14]

CCleanup::TChanges CCleanup::BasicCleanup	(	CBioseq_set_Handle &	bssh,
		Uint4	options = `0`
	)

Definition at line 208 of file cleanup.cpp.

References CNewCleanup_imp::BasicCleanupBioseqSetHandle(), CBioseq_set_Handle::GetScope(), makeCleanupChange(), and CNewCleanup_imp::SetScope().

◆ BasicCleanup() [4/14]

CCleanup::TChanges CCleanup::BasicCleanup	(	CBioSource &	src,
		Uint4	options = `0`
	)

Cleanup a BioSource.

Definition at line 180 of file cleanup.cpp.

References CLEANUP_SETUP.

◆ BasicCleanup() [5/14]

CCleanup::TChanges CCleanup::BasicCleanup	(	CSeq_annot &	sa,
		Uint4	options = `0`
	)

Cleanup a Seq-Annot.

Definition at line 164 of file cleanup.cpp.

References CLEANUP_SETUP.

◆ BasicCleanup() [6/14]

CCleanup::TChanges CCleanup::BasicCleanup	(	CSeq_annot_Handle &	sak,
		Uint4	options = `0`
	)

Definition at line 218 of file cleanup.cpp.

References CNewCleanup_imp::BasicCleanupSeqAnnotHandle(), CSeq_annot_Handle::GetScope(), makeCleanupChange(), and CNewCleanup_imp::SetScope().

◆ BasicCleanup() [7/14]

CCleanup::TChanges CCleanup::BasicCleanup	(	CSeq_descr &	desc,
		Uint4	options = `0`
	)

Definition at line 247 of file cleanup.cpp.

References CLEANUP_SETUP, and CSeq_descr_Base::Set().

◆ BasicCleanup() [8/14]

CCleanup::TChanges CCleanup::BasicCleanup	(	CSeq_entry &	se,
		Uint4	options = `0`
	)

Definition at line 132 of file cleanup.cpp.

References CLEANUP_SETUP.

Referenced by CMytestApplication::DoProcess(), CAsnvalThreadState::ProcessSeqAnnot(), CAsnvalThreadState::ProcessSeqEntry(), CAsnvalThreadState::ProcessSeqFeat(), CAsnvalThreadState::ProcessSeqSubmit(), and CAsnvalThreadState::ReadClassMember().

◆ BasicCleanup() [9/14]

CCleanup::TChanges CCleanup::BasicCleanup	(	CSeq_entry_Handle &	seh,
		Uint4	options = `0`
	)

Definition at line 188 of file cleanup.cpp.

References CNewCleanup_imp::BasicCleanupSeqEntryHandle(), CSeq_entry_Handle::GetScope(), makeCleanupChange(), and CNewCleanup_imp::SetScope().

◆ BasicCleanup() [10/14]

CCleanup::TChanges CCleanup::BasicCleanup	(	CSeq_feat &	sf,
		Uint4	options = `0`
	)

Cleanup a Seq-feat.

Definition at line 172 of file cleanup.cpp.

References CLEANUP_SETUP.

◆ BasicCleanup() [11/14]

CCleanup::TChanges CCleanup::BasicCleanup	(	CSeq_feat_Handle &	sfh,
		Uint4	options = `0`
	)

Definition at line 228 of file cleanup.cpp.

References CNewCleanup_imp::BasicCleanupSeqFeatHandle(), CSeq_feat_Handle::GetScope(), makeCleanupChange(), and CNewCleanup_imp::SetScope().

◆ BasicCleanup() [12/14]

CCleanup::TChanges CCleanup::BasicCleanup	(	CSeq_submit &	ss,
		Uint4	options = `0`
	)

Cleanup a Seq-submit.

Definition at line 140 of file cleanup.cpp.

References CLEANUP_SETUP.

◆ BasicCleanup() [13/14]

CCleanup::TChanges CCleanup::BasicCleanup	(	CSeqdesc &	desc,
		Uint4	options = `0`
	)

Definition at line 238 of file cleanup.cpp.

References CLEANUP_SETUP.

◆ BasicCleanup() [14/14]

CCleanup::TChanges CCleanup::BasicCleanup	(	CSubmit_block &	block,
		Uint4	options = `0`
	)

Definition at line 148 of file cleanup.cpp.

References CLEANUP_SETUP.

◆ BioSrcFromFeat()

CRef< CBioSource > CCleanup::BioSrcFromFeat ( const CSeq_feat & f )

static

Get BioSource from feature to use for source descriptor.

Definition at line 3824 of file cleanup.cpp.

References a, CSerialObject::Assign(), CSubSource_Base::eSubtype_other, CNewCleanup_imp::ExtendedCleanup(), f, ITERATE, makeCleanupChange(), CSubSource_Base::SetName(), CBioSource_Base::SetOrg(), CSubSource_Base::SetSubtype(), and CBioSource_Base::SetSubtype().

Referenced by CMergeBiosources::apply(), and ConvertSrcFeatsToSrcDescs().

◆ CleanupAffil()

bool CCleanup::CleanupAffil ( CAffil & af )

static

◆ CleanupAuthList()

bool CCleanup::CleanupAuthList	(	CAuth_list &	al,
		bool	fix_initials = `true`
	)

static

◆ CleanupAuthor()

bool CCleanup::CleanupAuthor	(	CAuthor &	author,
		bool	fix_initials = `true`
	)

static

Definition at line 51 of file cleanup_author.cpp.

References CPerson_id_Base::e_Consortium, CPerson_id_Base::e_Ml, CPerson_id_Base::e_Name, CPerson_id_Base::e_Str, NStr::IsBlank(), n, and NStr::TruncateSpacesInPlace().

Referenced by CleanupAuthList().

◆ CleanupCollectionDates()

bool CCleanup::CleanupCollectionDates	(	CSeq_entry_Handle	seh,
		bool	month_first
	)

static

Definition at line 4883 of file cleanup.cpp.

References CSerialObject::Assign(), CSeqFeatData_Base::e_Biosrc, NStr::Equal(), CSubSource_Base::eSubtype_collection_date, CSubSource::FixDateFormat(), CSeqFeatData_Base::GetBiosrc(), CSeq_entry_Handle::GetCompleteSeq_entry(), CSeq_feat_Handle::GetData(), CSeq_feat_Handle::GetOriginalSeq_feat(), GetSourceDescriptors(), CBioSource_Base::IsSetSubtype(), CSeq_feat_EditHandle::Replace(), CSeq_feat_Base::SetData(), CSeqdesc_Base::SetSource(), and CBioSource_Base::SetSubtype().

Referenced by CTable2AsnValidator::Cleanup().

◆ CleanupUserObject()

bool CCleanup::CleanupUserObject ( CUser_object & object )

static

Definition at line 50 of file cleanup_user_object.cpp.

References CleanVisString().

Referenced by CLoadStructComments::LoadWithAccessions(), CStructuredCommentPanel::OnExport(), CMacroFunction_ParseToStructComm::TheFunction(), CStructuredCommentPanel::UpdateSeqdesc(), CNewCleanup_imp::UserObjectBC(), and x_CleanupUserField().

◆ ClearInternalPartials() [1/4]

bool CCleanup::ClearInternalPartials	(	CPacked_seqint &	pint,
		bool	is_first = `true`,
		bool	is_last = `true`
	)

static

Definition at line 1595 of file cleanup.cpp.

References eExtreme_Biological, NON_CONST_ITERATE, and CPacked_seqint_Base::Set().

◆ ClearInternalPartials() [2/4]

bool CCleanup::ClearInternalPartials ( CSeq_entry_Handle seh )

static

Definition at line 1615 of file cleanup.cpp.

References CSerialObject::Assign(), ClearInternalPartials(), f, CSeq_feat_EditHandle::Replace(), and CSeq_feat_Base::SetLocation().

◆ ClearInternalPartials() [3/4]

bool CCleanup::ClearInternalPartials	(	CSeq_loc &	loc,
		bool	is_first = `true`,
		bool	is_last = `true`
	)

static

Clear internal partials.

Definition at line 1553 of file cleanup.cpp.

References CSeq_loc_Base::e_Mix, CSeq_loc_Base::e_Packed_int, CSeq_loc::SetMix(), CSeq_loc::SetPacked_int(), and CSeq_loc_Base::Which().

Referenced by ClearInternalPartials(), and CCleanupApp::x_ProcessFeatureOptions().

◆ ClearInternalPartials() [4/4]

bool CCleanup::ClearInternalPartials	(	CSeq_loc_mix &	mix,
		bool	is_first = `true`,
		bool	is_last = `true`
	)

static

Definition at line 1570 of file cleanup.cpp.

References ClearInternalPartials(), eExtreme_Biological, NON_CONST_ITERATE, and CSeq_loc_mix_Base::Set().

◆ ConvertDeltaSeqToRaw()

bool CCleanup::ConvertDeltaSeqToRaw	(	CSeq_entry_Handle	seh,
		CSeq_inst::EMol	filter = `CSeq_inst::eMol_not_set`
	)

static

Definition at line 4408 of file cleanup.cpp.

References CSerialObject::Assign(), CSeq_inst::ConvertDeltaToRaw(), CBioseq_Handle::GetInst(), and CBioseq_EditHandle::SetInst().

Referenced by CCleanupApp::x_ProcessXOptions().

◆ ConvertPubFeatsToPubDescs()

bool CCleanup::ConvertPubFeatsToPubDescs ( CSeq_entry_Handle seh )

static

Convert full-length publication features to publication descriptors.

Parameters

seh	Seq-entry to edit

Returns: bool indicates whether any changes were made

Definition at line 3388 of file cleanup.cpp.

References CSerialObject::Assign(), b, CSeqFeatData_Base::e_Pub, eExtreme_Biological, CPubdesc_Base::GetComment(), CSeqdesc_Base::GetPub(), NStr::IsBlank(), CPubdesc_Base::IsSetComment(), MoveOneFeatToPubdesc(), CPubdesc_Base::SetComment(), and CSeqdesc_Base::SetPub().

Referenced by CTbl2AsnApp::ProcessOneEntry(), CTbl2AsnApp::ProcessSingleEntry(), and CNewCleanup_imp::x_ExtendedCleanupExtra().

◆ ConvertSrcFeatsToSrcDescs()

bool CCleanup::ConvertSrcFeatsToSrcDescs ( CSeq_entry_Handle seh )

static

Convert full-length source features to source descriptors.

Parameters

seh	Seq-entry to edit

Returns: bool indicates whether any changes were made

Definition at line 3857 of file cleanup.cpp.

References CBioseq_EditHandle::AddSeqdesc(), CBioseq_set_EditHandle::AddSeqdesc(), CSerialObject::Assign(), b, BioSrcFromFeat(), CSeqFeatData_Base::e_Biosrc, CSeqdesc_Base::e_Source, CBioseq_set_Base::eClass_nuc_prot, eExtreme_Biological, CSubSource_Base::eSubtype_transgenic, CSeq_feat_EditHandle::GetAnnot(), CBioseq_set_Handle::GetClass(), CSeq_annot_Handle::GetCompleteSeq_annot(), CSeqdesc_Base::GetSource(), CBioSource::HasSubtype(), CBioseq_set_Handle::IsSetClass(), CBioSource_Base::IsSetIs_focus(), MergeDupBioSources(), NormalizeDescriptorOrder(), CSeq_annot_EditHandle::Remove(), CSeq_feat_EditHandle::Remove(), RemoveDupBioSource(), CBioseq_EditHandle::SetDescr(), CBioseq_set_EditHandle::SetDescr(), CSeqdesc_Base::SetSource(), and CNewCleanup_imp::ShouldRemoveAnnot().

Referenced by CNewCleanup_imp::x_ExtendedCleanupExtra().

◆ DecodeXMLMarkChanged()

bool CCleanup::DecodeXMLMarkChanged ( std::string & str )

static

decodes various tags, including carriage-return-line-feed constructs

Definition at line 4099 of file cleanup.cpp.

References _ASSERT, CTextFsm< MatchType >::AddWord(), ArraySize(), copy(), CTextFsm< MatchType >::GetInitialState(), CTextFsm< MatchType >::GetMatches(), CTextFsm< MatchType >::GetNextState(), CTextFsm< MatchType >::IsMatchFound(), CTextFsm< MatchType >::IsPrimed(), NPOS, CTextFsm< MatchType >::Prime(), result, and str().

Referenced by CNewCleanup_imp::x_DecodeXMLMarkChanged().

◆ ExpandGeneToIncludeChildren()

bool CCleanup::ExpandGeneToIncludeChildren	(	CSeq_feat &	gene,
		CTSE_Handle &	tse
	)

static

Expands gene to include features it cross-references.

Parameters

gene	Seq-feat to adjust
tse	Top-level Seq-entry in which to find other features

Returns: Boolean indicates whether anything changed

Definition at line 2571 of file cleanup.cpp.

References eExtreme_Positional, CSeqFeatData::eSubtype_any, f, CTSE_Handle::GetFeaturesWithId(), CObject_id_Base::GetId(), CSeq_feat_Base::GetLocation(), CSeq_loc::GetStart(), CSeq_loc::GetStop(), CSeq_feat_Base::GetXref(), CSeq_loc_Base::IsInt(), CSeq_feat_Base::IsSetLocation(), CSeq_feat_Base::IsSetXref(), ITERATE, and CSeq_feat_Base::SetLocation().

Referenced by WGSCleanup().

◆ ExtendedCleanup() [1/4]

CCleanup::TChanges CCleanup::ExtendedCleanup	(	CSeq_annot &	sa,
		Uint4	options = `0`
	)

Cleanup a Seq-Annot.

Definition at line 276 of file cleanup.cpp.

References CLEANUP_SETUP.

◆ ExtendedCleanup() [2/4]

CCleanup::TChanges CCleanup::ExtendedCleanup	(	CSeq_entry &	se,
		Uint4	options = `0`
	)

Cleanup a Seq-entry.

Definition at line 259 of file cleanup.cpp.

References CLEANUP_SETUP.

Referenced by CTable2AsnValidator::Cleanup(), and WGSCleanup().

◆ ExtendedCleanup() [3/4]

CCleanup::TChanges CCleanup::ExtendedCleanup	(	CSeq_entry_Handle &	seh,
		Uint4	options = `0`
	)

static

Definition at line 283 of file cleanup.cpp.

References CNewCleanup_imp::ExtendedCleanupSeqEntryHandle(), and makeCleanupChange().

◆ ExtendedCleanup() [4/4]

CCleanup::TChanges CCleanup::ExtendedCleanup	(	CSeq_submit &	ss,
		Uint4	options = `0`
	)

Cleanup a Seq-submit.

Definition at line 268 of file cleanup.cpp.

References CLEANUP_SETUP.

◆ ExtendStopPosition()

bool CCleanup::ExtendStopPosition	(	CSeq_feat &	f,
		const CSeq_feat *	cdregion,
		size_t	extension = `0`
	)

static

◆ ExtendToStopCodon()

bool CCleanup::ExtendToStopCodon	(	CSeq_feat &	f,
		CBioseq_Handle	bsh,
		size_t	limit
	)

static

Extends a feature up to limit nt to a stop codon, or to the end of the sequence if limit == 0 (partial will be set if location extends to end of sequence but no stop codon is found)

Parameters

f	Seq-feat to edit
bsh	CBioseq_Handle on which the feature is located
limit	maximum number of nt to extend, or 0 if unlimited

Returns: Boolean return value indicates whether the feature was extended

Definition at line 1113 of file cleanup.cpp.

References CSeqVector::begin(), CBioseq_Handle::eCoding_Iupac, eExtreme_Biological, CCdregion_Base::eFrame_not_set, CCdregion_Base::eFrame_three, CCdregion_Base::eFrame_two, eNa_strand_minus, ExtendStopPosition(), f, CBioseq_Handle::GetBioseqLength(), CTrans_table::GetCodonResidue(), CBioseq_Handle::GetId(), CBioseq_Handle::GetInst_Length(), GetLength(), CBioseq_Handle::GetScope(), CSeq_loc::GetStop(), CSeq_loc::GetStrand(), CGen_code_table::GetTransTable(), i, CSeq_loc::IsSetStrand(), len, mod(), CTrans_table::NextCodonState(), CSeq_loc::SetInt(), CSeq_loc::SetStrand(), and CSeqVector::size().

Referenced by ExtendToStopIfShortAndNotPartial(), and CCleanupApp::x_BatchExtendCDS().

◆ ExtendToStopIfShortAndNotPartial()

bool CCleanup::ExtendToStopIfShortAndNotPartial	(	CSeq_feat &	f,
		CBioseq_Handle	bsh,
		bool	check_for_stop = `true`
	)

static

Extends a coding region up to 50 nt.

if the coding region: 1. does not end with a stop codon 2. is adjacent to a stop codon 3. is not pseudo

Parameters

f	Seq-feat to edit
bsh	CBioseq_Handle on which the feature is located

Returns: Boolean return value indicates whether the feature was extended

Definition at line 1291 of file cleanup.cpp.

References eExtreme_Biological, NStr::EndsWith(), ExtendToStopCodon(), f, CSeq_feat_Base::GetLocation(), GetmRNAforCDS(), CBioseq_Handle::GetScope(), CSeq_loc::GetStop(), IsPseudo(), s_IsLocationEndAtOtherLocationInternalEndpoint(), and CSeqTranslator::Translate().

Referenced by CNewCleanup_imp::CdRegionEC(), CFeatTableEdit::xGenerate_mRNA_Product(), and CFeatureTableReader::xTranslateProtein().

◆ FindMatchingLocus_tagGene()

bool CCleanup::FindMatchingLocus_tagGene	(	CSeq_feat &	f,
		const CGene_ref &	gene_xref,
		CBioseq_Handle	bsh
	)

static

Detects gene features with matching locus_tag.

Parameters

f	Seq-feat parent feature of gene_xref [in]
gene_xref	Gene-ref of gene-xref [in]
bsh	CBioseq_Handle parent bioseq in which to search for genes [in]

Returns: Boolean return value indicates whether a gene feature with matching locus_tag has been found

Definition at line 990 of file cleanup.cpp.

References CSeqFeatData::eSubtype_gene, f, CGene_ref_Base::GetLocus_tag(), CGene_ref_Base::IsSetLocus_tag(), and match().

Referenced by RemoveOrphanLocus_tagGeneXrefs().

◆ FindMatchingLocusGene()

bool CCleanup::FindMatchingLocusGene	(	CSeq_feat &	f,
		const CGene_ref &	gene_xref,
		CBioseq_Handle	bsh
	)

static

Detects gene features with matching locus.

Parameters

f	Seq-feat parent feature of gene_xref [in]
gene_xref	Gene-ref of gene-xref [in]
bsh	CBioseq_Handle parent bioseq in which to search for genes [in]

Returns: Boolean return value indicates whether a gene feature with matching locus has been found

Definition at line 942 of file cleanup.cpp.

References CSeqFeatData::eSubtype_gene, f, CGene_ref_Base::GetLocus(), CGene_ref_Base::IsSetLocus(), and match().

Referenced by RemoveOrphanLocusGeneXrefs().

◆ FixECNumbers()

bool CCleanup::FixECNumbers ( CSeq_entry_Handle entry )

static

Fix EC numbers.

Parameters

entry Seq-entry-handle to clean

Returns: Boolean value indicates whether any changes were made

Definition at line 1713 of file cleanup.cpp.

References CSerialObject::Assign(), CSeqFeatData_Base::e_Prot, f, CSeq_feat_Base::GetData(), CProt_ref_Base::GetEc(), CSeqFeatData_Base::GetProt(), RemoveBadECNumbers(), CSeq_feat_EditHandle::Replace(), CSeq_feat_Base::SetData(), and UpdateECNumbers().

Referenced by CCleanupApp::x_ProcessFeatureOptions().

◆ FixGeneXrefSkew()

bool CCleanup::FixGeneXrefSkew ( CSeq_entry_Handle seh )

static

Examine all genes and gene xrefs in the Seq-entry.

If no genes have locus and some have locus tag AND no gene xrefs have locus-tag and some gene xrefs have locus, change all gene xrefs to use locus tag. If no genes have locus tag and some have locus AND no gene xrefs have locus and some gene xrefs have locus tag, change all gene xrefs to use locus.

Parameters

seh	Seq-entry to edit

Returns: bool indicates whether any changes were made

Definition at line 3915 of file cleanup.cpp.

References CSerialObject::Assign(), g(), CSeq_feat_Handle::GetData(), CSeqFeatData_Base::GetGene(), CSeq_feat_Handle::GetGeneXref(), CMappedFeat::GetSeq_feat(), CSeqFeatData_Base::IsGene(), CGene_ref_Base::IsSetLocus(), CGene_ref_Base::IsSetLocus_tag(), CSeq_feat_Handle::IsSetXref(), NON_CONST_ITERATE, CSeq_feat_EditHandle::Replace(), CFeat_CI::Rewind(), and CSeq_feat_Base::SetXref().

Referenced by CNewCleanup_imp::x_ExtendedCleanupExtra().

◆ FixRNAEditingCodingRegion()

bool CCleanup::FixRNAEditingCodingRegion ( CSeq_feat & cds )

static

From GB-7563 An action has been requested that will do the following: 1.

This action should be limited to protein sequences where the product is an exact match to a specified text (the usual string constraint is not needed). 2. Protein sequences for which the coding region is 5' partial should not be affected. 3. When the protein name matches, the following actions should be taken if and only if the first amino acid of the protein sequence is not M (methionine): a. The first amino acid of the protein sequence should be changed to methionine. b. The coding region should have the text "RNA editing" added to Seq-feat.except_text (separated from any existing text by a semicolon). If Seq-feat.except is not already true, it should be set to true. c. A code-break should be added to Cdregion.code-break where the Code-break.loc is the location of the first codon of the coding region and Code-break.aa is ncbieaa 'M' (Indexers will refer to "code-breaks" as "translation exceptions" because these appear in the flatfile as a /transl_except qualifier.

It will be the responsibility of the caller to only invoke this function for coding regions where the product name is a match, and the protein sequence does not already start with an M.

Definition at line 4849 of file cleanup.cpp.

References eExtreme_Biological, NStr::Find(), GetCodeBreakForLocation(), CSeq_feat_Base::GetData(), CSeq_feat_Base::GetExcept(), CSeq_feat_Base::GetExcept_text(), CSeq_feat_Base::GetLocation(), NStr::IsBlank(), CSeqFeatData_Base::IsCdregion(), IsMethionine(), CSeq_loc::IsPartialStart(), CSeq_feat_Base::IsSetData(), CSeq_feat_Base::IsSetExcept(), CSeq_feat_Base::IsSetExcept_text(), CSeq_feat_Base::IsSetLocation(), CSeq_feat_Base::SetExcept(), and CSeq_feat_Base::SetExcept_text().

Referenced by CRestoreRNAediting::OnApply().

◆ GetCitationList()

vector< CConstRef< CPub > > CCleanup::GetCitationList ( CBioseq_Handle bsh )

static

Get list of pubs that can be used for citations for Seq-feat on a Bioseq-handle.

Parameters

bsh	Bioseq-handle to search

Returns: vector<CConstRef<CPub> > ordered list of pubs Note that Seq-feat.cit appear in the flatfile using the position in the list

Definition at line 3207 of file cleanup.cpp.

References CSeqdesc_Base::e_Pub, CSeqFeatData_Base::e_Pub, CSeq_feat_Handle::GetData(), CSeqdesc_Base::GetPub(), CSeqFeatData_Base::GetPub(), GetPubdescLabels(), CAliasBase< TPrim >::Set(), CCit_gen_Base::SetCit(), CPub_Base::SetGen(), CPub_Base::SetMuid(), CPub_Base::SetPmid(), and CCit_gen_Base::SetSerial_number().

Referenced by CNewCleanup_imp::MoveCitationQuals().

◆ GetCodeBreakForLocation()

CConstRef< CCode_break > CCleanup::GetCodeBreakForLocation	(	size_t	pos,
		const CSeq_feat &	cds
	)

static

utility function for finding the code break for a given amino acid position pos is the position of the amino acid where the translation exception occurs (starts with 1)

Definition at line 4720 of file cleanup.cpp.

References CCdregion_Base::eFrame_not_set, CCdregion_Base::eFrame_one, CCdregion_Base::eFrame_three, CCdregion_Base::eFrame_two, CSeqFeatData_Base::GetCdregion(), CCdregion_Base::GetCode_break(), CSeq_feat_Base::GetData(), CCdregion_Base::GetFrame(), CSeq_feat_Base::GetLocation(), CSeqFeatData_Base::IsCdregion(), CCdregion_Base::IsSetCode_break(), CSeq_feat_Base::IsSetData(), CCdregion_Base::IsSetFrame(), CSeq_feat_Base::IsSetLocation(), LocationOffset(), and offset.

Referenced by FixRNAEditingCodingRegion().

◆ GetNormalizableGeneQualPairs()

vector< CCleanup::TFeatGenePair > CCleanup::GetNormalizableGeneQualPairs ( CBioseq_Handle bsh )

static

Definition at line 154 of file gene_qual_normalization.cpp.

References CSeqdesc_Base::e_Source, f, NStr::Find(), CBioSource::GetLineage(), CMappedFeat::GetOriginalFeature(), CSeqdesc_Base::GetSource(), CBioseq_Handle::IsAa(), IsMappablePair(), CBioSource::IsSetLineage(), CSeqdesc_Base::IsSource(), and NPOS.

Referenced by GetNormalizeGeneQualsCommand(), and NormalizeGeneQuals().

◆ GetProteinLocationFromNucleotideLocation() [1/2]

CRef< CSeq_loc > CCleanup::GetProteinLocationFromNucleotideLocation	(	const CSeq_loc &	nuc_loc,
		const CSeq_feat &	cds,
		CScope &	scope,
		bool	require_inframe = `false`
	)

static

◆ GetProteinLocationFromNucleotideLocation() [2/2]

CRef< CSeq_loc > CCleanup::GetProteinLocationFromNucleotideLocation	(	const CSeq_loc &	nuc_loc,
		CScope &	scope
	)

static

Definition at line 4348 of file cleanup.cpp.

References GetOverlappingCDS(), and CSeq_feat_Base::IsSetProduct().

Referenced by MoveFeatToProtein(), CFeatureTableReader::MoveRegionsToProteins(), and s_MoveProteinSpecificFeats().

◆ GetProteinName() [1/2]

const string & CCleanup::GetProteinName ( const CProt_ref & prot )

static

Definition at line 1467 of file cleanup.cpp.

References kEmptyStr, and prot.

Referenced by g_InstantiateMissingProteins(), GetProteinName(), s_GetProteinNameFromXrefOrQual(), WGSCleanup(), and CCleanupApp::x_FixCDS().

◆ GetProteinName() [2/2]

const string & CCleanup::GetProteinName	(	const CSeq_feat &	cds,
		CSeq_entry_Handle	seh
	)

static

Definition at line 1497 of file cleanup.cpp.

References CSeqFeatData::eSubtype_prot, f, CSeq_entry_Handle::GetBioseqHandle(), CSeq_loc::GetId(), CSeq_feat_Base::GetProduct(), GetProteinName(), CSeq_feat_Base::IsSetProduct(), prot, and s_GetProteinNameFromXrefOrQual().

◆ GetPubdescLabels()

void CCleanup::GetPubdescLabels	(	const CPubdesc &	pd,
		vector< TEntrezId > &	pmids,
		vector< TEntrezId > &	muids,
		vector< int > &	serials,
		vector< string > &	published_labels,
		vector< string > &	unpublished_labels
	)

static

For Publication Citations Get labels for a pubdesc.

To be used in citations.

Definition at line 3139 of file cleanup.cpp.

References CPub::eContent, NStr::eNocase, IAbstractCitation::fLabel_Unique, CPub_equiv_Base::Get(), CPubdesc_Base::GetPub(), NStr::IsBlank(), CPubdesc_Base::IsSetPub(), ITERATE, label, s_GetAuthorsString(), and NStr::StartsWith().

Referenced by GetCitationList(), CValidator::CCacheImpl::GetPubdescToInfo(), s_CollectPubDescriptorLabels(), and CValidError_imp::ValidateCitations().

◆ IsEmpty()

bool CCleanup::IsEmpty ( const CAuth_list::TAffil & affil )

static

Definition at line 424 of file cleanup_author.cpp.

References CAffil_Base::GetStd(), CAffil_Base::GetStr(), NStr::IsBlank(), CAffil_Base::C_Std::IsSetAffil(), CAffil_Base::C_Std::IsSetCity(), CAffil_Base::C_Std::IsSetCountry(), CAffil_Base::C_Std::IsSetDiv(), CAffil_Base::C_Std::IsSetEmail(), CAffil_Base::C_Std::IsSetFax(), CAffil_Base::C_Std::IsSetPhone(), CAffil_Base::C_Std::IsSetPostal_code(), CAffil_Base::C_Std::IsSetStreet(), CAffil_Base::C_Std::IsSetSub(), CAffil_Base::IsStd(), and CAffil_Base::IsStr().

Referenced by CCitSubCleaner::Clean(), and CleanupAuthList().

◆ IsGeneXrefUnnecessary()

bool CCleanup::IsGeneXrefUnnecessary	(	const CSeq_feat &	sf,
		CScope &	scope,
		const CGene_ref &	gene_xref
	)

static

Calculates whether a Gene-xref is unnecessary (because it refers to the same gene as would be calculated using overlap)

Parameters

sf	Seq-feat with the xref [in]
scope	Scope in which to search for location [in]
gene_xref	Gene-ref of gene-xref [in]

Returns: Boolean return value indicates whether gene-xref is unnecessary

Definition at line 744 of file cleanup.cpp.

References Compare(), CSeqFeatData_Base::e_Gene, eOverlap_Contained, eSame, CSeqFeatData::eSubtype_gene, fCompareOverlapping, g(), CSeq_feat_Base::GetData(), CSeqFeatData_Base::GetGene(), CSeq_feat_Base::GetLocation(), GetOverlappingFeatures(), GetOverlappingGene(), CConstRef< C, Locker >::GetPointer(), CSeqFeatData_Base::IsGene(), CSeq_feat_Base::IsSetData(), CGene_ref::IsSuppressed(), ITERATE, and CGene_ref::RefersToSameGene().

Referenced by RemoveUnnecessaryGeneXrefs(), and CMacroFunction_RemoveGeneXref::s_GeneXrefMatchesNecessary().

◆ IsMethionine()

bool CCleanup::IsMethionine ( const CCode_break & cb )

static

Definition at line 4688 of file cleanup.cpp.

References CCode_break_Base::C_Aa::e_Ncbi8aa, CCode_break_Base::C_Aa::e_Ncbieaa, CCode_break_Base::C_Aa::e_Ncbistdaa, CCode_break_Base::GetAa(), CCode_break_Base::C_Aa::GetNcbi8aa(), CCode_break_Base::C_Aa::GetNcbieaa(), CCode_break_Base::C_Aa::GetNcbistdaa(), CCode_break_Base::IsSetAa(), methionine_encoded, and CCode_break_Base::C_Aa::Which().

Referenced by FixRNAEditingCodingRegion().

◆ IsMinPub()

bool CCleanup::IsMinPub	(	const CPubdesc &	pd,
		bool	is_refseq_prot
	)

static

Is this a "minimal" pub? (If yes, do not rescue from a Seq-feat.cit)

Definition at line 3426 of file cleanup.cpp.

References gen, CPub_equiv_Base::Get(), CPubdesc_Base::GetPub(), CPubdesc_Base::IsSetPub(), and ITERATE.

Referenced by RescueSiteRefPubs().

◆ LocationMayBeExtendedToMatch()

bool CCleanup::LocationMayBeExtendedToMatch	(	const CSeq_loc &	orig,
		const CSeq_loc &	improved
	)

static

Checks whether it is possible to extend the original location up to improved one.

It is possible only if the original location is less than improved

Parameters

orig	Seq-loc to check
improved	Seq-loc original location may be extended to

Returns: Boolean return value indicates whether the extention is possible

Definition at line 1333 of file cleanup.cpp.

References eExtreme_Biological, eNa_strand_minus, CSeq_loc::GetStop(), and orig.

Referenced by CNewCleanup_imp::CdRegionEC(), CCleanupApp::x_FixCDS(), CFeatTableEdit::xGenerate_mRNA_Product(), and CFeatureTableReader::xTranslateProtein().

◆ MakeIRDFeatsFromSourceXrefs()

bool CCleanup::MakeIRDFeatsFromSourceXrefs ( CSeq_entry_Handle entry )

static

From SQD-4329 For each sequence with a source that has an IRD db_xref, create a misc_feature across the entire span and move the IRD db_xref from the source to the misc_feature.

Create a suppressing gene xref for the misc_feature.

Parameters

entry Seq-entry on which to search for sources and create features

Returns: bool indicates changes were made

Definition at line 4654 of file cleanup.cpp.

References AddIRDMiscFeature(), CSeqdesc_Base::e_Source, CSeq_inst_Base::eMol_na, NStr::Equal(), COrg_ref_Base::GetDb(), CBioSource_Base::GetOrg(), CSeqdesc_Base::GetSource(), COrg_ref_Base::IsSetDb(), CBioSource_Base::IsSetOrg(), COrg_ref_Base::ResetDb(), and COrg_ref_Base::SetDb().

Referenced by CCleanupApp::x_ProcessXOptions().

◆ MakeSmallGenomeSet()

size_t CCleanup::MakeSmallGenomeSet ( CSeq_entry_Handle entry )

static

Definition at line 4583 of file cleanup.cpp.

References CSeqdesc_Base::e_Source, CSeq_inst_Base::eMol_na, map_checker< Container >::end(), map_checker< Container >::find(), CInfluenzaSet::GetKey(), CBioSource_Base::GetOrg(), CSeqdesc_Base::GetSource(), NStr::IsBlank(), CBioSource_Base::IsSetOrg(), and ncbi::grid::netcache::search::fields::key.

Referenced by CCleanupApp::x_ProcessXOptions().

◆ MergeDupBioSources() [1/2]

bool CCleanup::MergeDupBioSources	(	CBioSource &	src1,
		const CBioSource &	add
	)

static

◆ MergeDupBioSources() [2/2]

bool CCleanup::MergeDupBioSources ( CSeq_descr & descr )

static

Definition at line 3765 of file cleanup.cpp.

References AreBioSourcesMergeable(), CNewCleanup_imp::ExtendedCleanup(), makeCleanupChange(), and CSeq_descr_Base::Set().

Referenced by CMergeBiosources::apply(), ConvertSrcFeatsToSrcDescs(), CNewCleanup_imp::x_MergeDupBioSources(), and CNewCleanup_imp::x_MoveNpSrc().

◆ MoveFeatToProtein()

bool CCleanup::MoveFeatToProtein ( CSeq_feat_Handle fh )

static

Moves one feature from nucleotide bioseq to the appropriate protein sequence.

Parameters

fh	Feature to edit

Returns: Boolean return value indicates whether any changes were made

Definition at line 589 of file cleanup.cpp.

References CSerialObject::Assign(), CBioseq_EditHandle::AttachAnnot(), CNewCleanup_imp::BasicCleanupSeqFeat(), ConvertProteinToImp(), CSeqFeatData_Base::e_Imp, CProt_ref_Base::eProcessed_mature, CProt_ref_Base::eProcessed_not_set, CProt_ref_Base::eProcessed_preprotein, ftable, CBioseq_Base::GetAnnot(), CSeq_feat_Handle::GetAnnot(), CBioseq_Handle::GetBioseqCore(), CScope::GetBioseqHandle(), CSeq_feat_Base::GetComment(), CBioseq_Handle::GetCompleteBioseq(), CSeq_annot_Handle::GetCompleteSeq_annot(), CSeq_feat_Base::GetData(), CSeq_feat_Handle::GetData(), CBioseq_Handle::GetEditHandle(), CBioseq_Base::GetId(), CSeqFeatData_Base::GetImp(), CImp_feat_Base::GetKey(), CSeq_feat_Base::GetLocation(), CSeq_feat_Handle::GetLocation(), CProt_ref_Base::GetProcessed(), CSeq_feat_Handle::GetProduct(), CSeqFeatData_Base::GetProt(), GetProteinLocationFromNucleotideLocation(), CSeq_feat_Handle::GetScope(), CScope::GetSeq_annotHandle(), CSeq_feat_Handle::GetSeq_feat(), CBioseq_Handle::IsAa(), NStr::IsBlank(), CSeqFeatData_Base::IsImp(), CSeqFeatData_Base::IsProt(), CBioseq_Base::IsSetAnnot(), CSeq_feat_Base::IsSetComment(), CImp_feat_Base::IsSetKey(), CProt_ref_Base::IsSetName(), CProt_ref_Base::IsSetProcessed(), CSeq_feat_Base::IsSetProduct(), CSeq_feat_Handle::IsSetProduct(), ITERATE, makeCleanupChange(), orig, CSeq_feat_EditHandle::Replace(), RescueProtProductQual(), CSeq_feat_Base::ResetComment(), CSeq_feat_Base::ResetLocation(), CSeq_feat_Base::ResetProduct(), s_GetCdsByLocation(), s_GetCdsByProduct(), s_IsPreprotein(), s_ProcessedFromKey(), CSeq_feat_Base::SetData(), SetFeaturePartial(), CSeq_feat_Base::SetLocation(), CNewCleanup_imp::SetScope(), CNewCleanup_imp::ShouldRemoveAnnot(), CSeq_annot_EditHandle::TakeFeat(), and CSeqFeatData_Base::Which().

Referenced by MoveProteinSpecificFeats().

◆ MoveOneFeatToPubdesc()

void CCleanup::MoveOneFeatToPubdesc	(	CSeq_feat_Handle	feat,
		CRef< CSeqdesc >	d,
		CBioseq_Handle	b,
		bool	remove_feat = `true`
	)

static

◆ MoveProteinSpecificFeats()

bool CCleanup::MoveProteinSpecificFeats ( CSeq_entry_Handle seh )

static

Moves protein-specific features from nucleotide sequences in the Seq-entry to the appropriate protein sequence.

Parameters

seh	Seq-entry Handle to edit [in]

Returns: Boolean return value indicates whether any changes were made

Definition at line 724 of file cleanup.cpp.

References CSeqFeatData_Base::e_Bond, CSeqFeatData_Base::e_Imp, CSeqFeatData_Base::e_Prot, CSeqFeatData_Base::e_Psec_str, CSeq_inst_Base::eMol_na, SAnnotSelector::IncludeFeatType(), and MoveFeatToProtein().

Referenced by s_MoveProteinSpecificFeats(), and CNewCleanup_imp::x_ExtendedCleanupExtra().

◆ NormalizeDescriptorOrder() [1/2]

bool CCleanup::NormalizeDescriptorOrder ( CSeq_descr & descr )

static

Normalize Descriptor Order on a specific Seq-entry.

Parameters

entry Seq-entry to edit

Returns: Boolean return value indicates whether object was updated

Definition at line 3000 of file cleanup.cpp.

References s_SeqDescLessThan(), seq_mac_is_sorted(), and CSeq_descr_Base::Set().

Referenced by ConvertSrcFeatsToSrcDescs(), CCleanupApp::HandleSeqEntry(), MoveOneFeatToPubdesc(), NormalizeDescriptorOrder(), RenormalizeNucProtSets(), WGSCleanup(), CCleanupHugeAsnReader::x_CleanupTopLevelDescriptors(), CCleanupApp::x_ProcessXOptions(), and CNewCleanup_imp::x_SortSeqDescs().

◆ NormalizeDescriptorOrder() [2/2]

bool CCleanup::NormalizeDescriptorOrder ( CSeq_entry_Handle seh )

static

Normalize Descriptor Order on a specific Seq-entry.

Parameters

seh	Seq-entry-Handle to edit

Returns: Boolean return value indicates whether object was updated

Definition at line 3010 of file cleanup.cpp.

References CSeq_entry_CI::fIncludeGivenEntry, CSeq_entry_CI::fRecursive, and NormalizeDescriptorOrder().

◆ NormalizeGeneQuals() [1/3]

bool CCleanup::NormalizeGeneQuals ( CBioseq_Handle bsh )

static

Definition at line 203 of file gene_qual_normalization.cpp.

References CSerialObject::Assign(), GetNormalizableGeneQualPairs(), NormalizeGeneQuals(), and CSeq_feat_EditHandle::Replace().

◆ NormalizeGeneQuals() [2/3]

bool CCleanup::NormalizeGeneQuals ( CSeq_entry_Handle seh )

static

Definition at line 224 of file gene_qual_normalization.cpp.

References CSeq_inst_Base::eMol_na, and NormalizeGeneQuals().

◆ NormalizeGeneQuals() [3/3]

bool CCleanup::NormalizeGeneQuals	(	CSeq_feat &	cds,
		CSeq_feat &	gene
	)

static

Definition at line 98 of file gene_qual_normalization.cpp.

References NStr::Equal(), CSeq_feat_Base::GetData(), CSeqFeatData_Base::GetGene(), CGene_ref_Base::GetLocus_tag(), NStr::IsBlank(), CSeqFeatData_Base::IsGene(), CGene_ref_Base::IsSetLocus(), CGene_ref_Base::IsSetLocus_tag(), CSeq_feat_Base::IsSetXref(), CSeq_feat_Base::SetData(), and CSeq_feat_Base::SetXref().

Referenced by GetNormalizeGeneQualsCommand(), NormalizeGeneQuals(), and CNewCleanup_imp::x_ExtendedCleanupExtra().

◆ OkToPromoteNpPub() [1/2]

bool CCleanup::OkToPromoteNpPub ( const CBioseq & b )

static

For some sequences, pubs should not be promoted to nuc-prot set from sequence.

Definition at line 3330 of file cleanup.cpp.

References b, and ITERATE.

◆ OkToPromoteNpPub() [2/2]

bool CCleanup::OkToPromoteNpPub ( const CPubdesc & pd )

static

Some pubs should not be promoted to nuc-prot set from sequence.

Definition at line 3343 of file cleanup.cpp.

References CPubdesc_Base::IsSetComment(), CPubdesc_Base::IsSetFig(), CPubdesc_Base::IsSetName(), and CPubdesc_Base::IsSetNum().

Referenced by MoveOneFeatToPubdesc(), and CNewCleanup_imp::x_MoveNpPub().

◆ operator=()

CCleanup& CCleanup::operator= ( const CCleanup & )

delete

◆ ParseCodeBreak()

bool CCleanup::ParseCodeBreak	(	const CSeq_feat &	feat,
		CCdregion &	cds,
		const CTempString &	str,
		CScope &	scope,
		IObjtoolsListener *	pMessageListener = `nullptr`
	)

static

Parse string into code break and add to coding region.

Parameters

feat	feature that contains coding region - necessary to determine codon boundaries
cds	coding region to which code breaks will be added
str	string from which to parse code break
scope	scope in which to find sequences referenced (used for location comparisons)

Returns: bool indicates string was successfully parsed and code break was added

Definition at line 4425 of file cleanup.cpp.

References Compare(), CCleanupMessage::eCodeBreak, eContained, eDiag_Error, eNa_strand_minus, eNa_strand_plus, fCompareOverlapping, FIELD_IS_SET, NStr::Find(), GET_FIELD, CSeq_loc::GetId(), GetLength(), CSeq_feat_Base::GetLocation(), isalpha(), CSeq_loc_Base::IsInt(), CSeq_loc_Base::IsPnt(), CSeq_feat_Base::IsSetLocation(), isspace(), len, msg(), NPOS, IObjtoolsListener::PutMessage(), ReadLocFromText(), RESET_FIELD, CCode_break_Base::SetAa(), CCdregion_Base::SetCode_break(), CCode_break_Base::SetLoc(), CCode_break_Base::C_Aa::SetNcbieaa(), CSeq_loc::SetStrand(), str(), NStr::TruncateSpaces_Unsafe(), and x_ValidAminoAcid().

Referenced by ParseCodeBreaks().

◆ ParseCodeBreaks()

bool CCleanup::ParseCodeBreaks	(	CSeq_feat &	feat,
		CScope &	scope
	)

static

Parses all valid transl_except Gb-quals into code-breaks for cdregion, then removes the transl_except Gb-quals that were successfully parsed.

Parameters

feat	feature that contains coding region
scope	scope in which to find sequences referenced (used for location comparisons)

Returns: bool indicates changes were made

Definition at line 4556 of file cleanup.cpp.

References NStr::EqualNocase(), CSeq_feat_Base::GetData(), CSeq_feat_Base::GetQual(), CSeqFeatData_Base::IsCdregion(), CSeq_feat_Base::IsSetData(), CSeq_feat_Base::IsSetLocation(), CSeq_feat_Base::IsSetQual(), ParseCodeBreak(), CSeq_feat_Base::ResetQual(), CSeq_feat_Base::SetData(), and CSeq_feat_Base::SetQual().

Referenced by CNewCleanup_imp::x_CleanSeqFeatQuals(), CImportFeatTable::x_DoImportCDS(), and CFeatureTableReader::xMoveCdRegions().

◆ PubAlreadyInSet()

bool CCleanup::PubAlreadyInSet	(	const CPubdesc &	pd,
		const CSeq_descr &	descr
	)

static

Definition at line 3319 of file cleanup.cpp.

References CSeq_descr_Base::Get(), CPubdesc_Base::GetPub(), ITERATE, and s_FirstPubMatchesSecond().

Referenced by MoveOneFeatToPubdesc(), and CNewCleanup_imp::x_MoveNpPub().

◆ RemoveBadECNumbers()

bool CCleanup::RemoveBadECNumbers ( CProt_ref::TEc & ec_num_list )

static

Delete EC numbers.

Parameters

ec_num_list Prot-ref ec number list to clean

Returns: Boolean value indicates whether any changes were made

Definition at line 1689 of file cleanup.cpp.

References CleanVisStringJunk(), CProt_ref::eEC_deleted, CProt_ref::eEC_unknown, CProt_ref::GetECNumberStatus(), and CProt_ref::IsECNumberSplit().

Referenced by FixECNumbers().

◆ RemoveDupBioSource()

bool CCleanup::RemoveDupBioSource ( CSeq_descr & descr )

static

Remove duplicate biosource descriptors.

Definition at line 3794 of file cleanup.cpp.

References ITERATE, and CSeq_descr_Base::Set().

Referenced by ConvertSrcFeatsToSrcDescs().

◆ RemoveDuplicatePubs()

bool CCleanup::RemoveDuplicatePubs ( CSeq_descr & descr )

static

Remove duplicate publications.

Definition at line 3281 of file cleanup.cpp.

References CSeq_descr_Base::Set().

Referenced by MoveOneFeatToPubdesc(), and CNewCleanup_imp::x_RemoveDupPubs().

◆ RemoveNcbiCleanupObject()

bool CCleanup::RemoveNcbiCleanupObject ( CSeq_entry & seq_entry )

static

Removes NcbiCleanup User Objects in the Seq-entry.

Parameters

seq_entry Seq-entry to edit

Returns: Boolean return value indicates whether object was removed

Definition at line 1898 of file cleanup.cpp.

References CUser_object::eObjectType_Cleanup, CSeq_entry_Base::GetSet(), CSeq_entry_Base::IsSeq(), CSeq_entry_Base::IsSet(), CSeq_entry::IsSetDescr(), CBioseq_set_Base::IsSetSeq_set(), NON_CONST_ITERATE, CBioseq_Base::ResetDescr(), CBioseq_set_Base::ResetDescr(), CSeq_entry::SetDescr(), CSeq_entry_Base::SetSeq(), CBioseq_set_Base::SetSeq_set(), and CSeq_entry_Base::SetSet().

Referenced by CCleanupApp::HandleSeqEntry(), and CNewCleanup_imp::x_AddNcbiCleanupObject().

◆ RemoveNonsuppressingGeneXrefs()

bool CCleanup::RemoveNonsuppressingGeneXrefs ( CSeq_feat & f )

static

Removes non-suppressing Gene-xrefs.

Parameters

f	Seq-feat to edit [in]

Returns: Boolean return value indicates whether gene-xrefs were removed

Definition at line 828 of file cleanup.cpp.

References f.

Referenced by CRemoveGeneXrefs::GetCommand(), and CRemoveGeneXrefs::RemoveNonsuppressing().

◆ RemoveOrphanLocus_tagGeneXrefs()

bool CCleanup::RemoveOrphanLocus_tagGeneXrefs	(	CSeq_feat &	f,
		CBioseq_Handle	bsh
	)

static

Removes orphaned locus_tag Gene-xrefs.

Parameters

f	Seq-feat to edit [in]
bsh	CBioseq_Handle in which to search for gene features [in]

Returns: Boolean return value indicates whether gene-xrefs were removed

Definition at line 1013 of file cleanup.cpp.

References f, and FindMatchingLocus_tagGene().

Referenced by CRemoveGeneXrefs::RemoveOrphanLocus_tag().

◆ RemoveOrphanLocusGeneXrefs()

bool CCleanup::RemoveOrphanLocusGeneXrefs	(	CSeq_feat &	f,
		CBioseq_Handle	bsh
	)

static

Removes orphaned locus Gene-xrefs.

Parameters

f	Seq-feat to edit [in]
bsh	CBioseq_Handle in which to search for gene features [in]

Returns: Boolean return value indicates whether gene-xrefs were removed

Definition at line 965 of file cleanup.cpp.

References f, and FindMatchingLocusGene().

Referenced by CRemoveGeneXrefs::RemoveOrphanLocus().

◆ RemovePseudoProduct()

bool CCleanup::RemovePseudoProduct	(	CSeq_feat &	cds,
		CScope &	scope
	)

static

Removes protein product from pseudo coding region.

Parameters

cds	Seq-feat to adjust
scope	Scope in which to find protein sequence and remove it

Returns: Boolean indicates whether anything changed

Definition at line 2537 of file cleanup.cpp.

References CSeqFeatData::eSubtype_prot, CScope::GetBioseqHandle(), CSeq_feat_Base::GetComment(), CSeq_feat_Base::GetData(), CSeq_feat_Base::GetProduct(), NStr::IsBlank(), CSeqFeatData_Base::IsCdregion(), IsPseudo(), CSeq_feat_Base::IsSetComment(), CSeq_feat_Base::IsSetData(), CSeq_feat_Base::IsSetProduct(), label, prot, CBioseq_EditHandle::Remove(), CSeq_feat_Base::ResetProduct(), and CSeq_feat_Base::SetComment().

Referenced by CNewCleanup_imp::CdRegionEC(), and WGSCleanup().

◆ RemoveUnnecessaryGeneXrefs() [1/2]

bool CCleanup::RemoveUnnecessaryGeneXrefs ( CSeq_entry_Handle seh )

static

Removes unnecessary Gene-xrefs on features in Seq-entry.

Parameters

seh	Seq-entry-Handle to edit [in]

Returns: Boolean return value indicates whether gene-xrefs were removed

Definition at line 804 of file cleanup.cpp.

References CSerialObject::Assign(), CSeq_entry_Handle::GetScope(), RemoveUnnecessaryGeneXrefs(), and CSeq_feat_EditHandle::Replace().

◆ RemoveUnnecessaryGeneXrefs() [2/2]

bool CCleanup::RemoveUnnecessaryGeneXrefs	(	CSeq_feat &	f,
		CScope &	scope
	)

static

Removes unnecessary Gene-xrefs.

Parameters

f	Seq-feat to edit [in]
scope	Scope in which to search for locations [in]

Returns: Boolean return value indicates whether gene-xrefs were removed

Definition at line 779 of file cleanup.cpp.

References f, and IsGeneXrefUnnecessary().

Referenced by CRemoveGeneXrefs::GetCommand(), CRemoveGeneXrefs::RemoveUnnecessary(), RemoveUnnecessaryGeneXrefs(), and CCleanupApp::x_ProcessFeatureOptions().

◆ RemoveUnseenTitles() [1/2]

bool CCleanup::RemoveUnseenTitles ( CSeq_entry_EditHandle::TSeq seq )

static

Remove all titles in Seqdescr except the last, because it is the only one that would be displayed in the flatfile.

Parameters

seq	Bioseq-Handle to edit

Returns: Boolean return value indicates whether any titles were removed

Definition at line 3027 of file cleanup.cpp.

References CSeq_descr_Base::Get(), CBioseq_Handle::GetDescr(), CBioseq_Handle::IsSetDescr(), ITERATE, CBioseq_EditHandle::RemoveSeqdesc(), and CConstRef< C, Locker >::Reset().

Referenced by RenormalizeNucProtSets().

◆ RemoveUnseenTitles() [2/2]

bool CCleanup::RemoveUnseenTitles ( CSeq_entry_EditHandle::TSet set )

static

Remove all titles in Seqdescr except the last, because it is the only one that would be displayed in the flatfile.

Parameters

set	Bioseq-set-Handle to edit

Returns: Boolean return value indicates whether any titles were removed

Definition at line 3046 of file cleanup.cpp.

References ITERATE, and CConstRef< C, Locker >::Reset().

◆ RenormalizeNucProtSets()

bool CCleanup::RenormalizeNucProtSets ( CSeq_entry_Handle seh )

static

Convert nuc-prot sets with just one sequence to just the sequence can't be done during the explore phase because it changes a seq to a set.

Parameters

seh	Seq-entry to edit

Returns: bool indicates whether any changes were made

Definition at line 4063 of file cleanup.cpp.

References CSeq_entry_EditHandle::ConvertSetToSeq(), CBioseq_set_Base::eClass_eco_set, CBioseq_set_Base::eClass_gen_prod_set, CBioseq_set_Base::eClass_genbank, CBioseq_set_Base::eClass_mut_set, CBioseq_set_Base::eClass_nuc_prot, CBioseq_set_Base::eClass_phy_set, CBioseq_set_Base::eClass_pop_set, CBioseq_set_Base::eClass_small_genome_set, CBioseq_set_Base::eClass_wgs_set, CBioseq_set_Handle::GetClass(), CSeq_entry_Handle::GetCompleteSeq_entry(), CSeq_entry_Handle::GetEditHandle(), CSeq_entry_Handle::GetScope(), CSeq_entry_Handle::GetSeq(), CScope::GetSeq_entryHandle(), CBioseq_set_Base::GetSeq_set(), CSeq_entry_Base::GetSet(), CSeq_entry_Handle::GetSet(), CSeq_entry_Handle::IsSet(), CBioseq_set_Handle::IsSetClass(), CBioseq_Handle::IsSetDescr(), CBioseq_set_Base::IsSetSeq_set(), ITERATE, NormalizeDescriptorOrder(), RemoveUnseenTitles(), and CSeq_entry_EditHandle::SetSeq().

Referenced by CNewCleanup_imp::x_ExtendedCleanupExtra(), and CCleanupApp::x_RemoveDuplicateFeatures().

◆ RepackageProteins() [1/2]

bool CCleanup::RepackageProteins	(	const CSeq_feat &	cds,
		CBioseq_set_Handle	np
	)

static

Definition at line 4362 of file cleanup.cpp.

References CTSE_Handle::GetBioseqHandle(), CBioseq_Handle::GetParentBioseq_set(), CSeq_feat_Base::GetProduct(), CBioseq_Handle::GetSeq_entry_Handle(), CBioseq_set_Handle::GetTSE_Handle(), CSeq_loc_Base::GetWhole(), CSeq_feat_Base::IsSetProduct(), CSeq_loc_Base::IsWhole(), and CBioseq_set_EditHandle::TakeEntry().

◆ RepackageProteins() [2/2]

bool CCleanup::RepackageProteins ( CSeq_entry_Handle seh )

static

Find proteins that are not packaged in the same nuc-prot set as the coding region for which they are a product, and move them to that nuc-prot set.

Ignore coding regions that are in gen-prod-sets.

Parameters

seh	Seq-entry to edit

Returns: bool indicates whether any changes were made

Definition at line 4385 of file cleanup.cpp.

References CSeq_entry_Base::e_Set, CBioseq_set_Base::eClass_nuc_prot, CSeq_entry_CI::fIncludeGivenEntry, CSeq_entry_CI::fRecursive, ITERATE, and si.

Referenced by CNewCleanup_imp::x_ExtendedCleanupExtra().

◆ RepairXrefs() [1/3]

bool CCleanup::RepairXrefs	(	const CSeq_feat &	f,
		const CTSE_Handle &	tse
	)

static

Repairs non-reciprocal xref pairs for specified feature if xrefs between subtypes are permitted and feature with missing xref does not have an xref to a different feature of the same subtype.

Parameters

f	Seq-feat to edit [in]
tse	top-level Seq-entry in which to search for the other half of the xref pair

Returns: Boolean return value indicates whether xrefs were created

Definition at line 905 of file cleanup.cpp.

References CSeqFeatData_Base::e_not_set, f, CTSE_Handle::GetFeaturesWithId(), CObject_id_Base::GetId(), and ITERATE.

Referenced by CSequenceEditingEventHandler::FixNonReciprocalLinks(), RepairXrefs(), and CNewCleanup_imp::x_ExtendedCleanupExtra().

◆ RepairXrefs() [2/3]

bool CCleanup::RepairXrefs	(	const CSeq_feat &	src,
		CSeq_feat_Handle &	dst,
		const CTSE_Handle &	tse
	)

static

Repairs non-reciprocal xref pairs for specified feature pair if xrefs between subtypes are permitted and feature with missing xref does not have an xref to a different feature of the same subtype.

Parameters

f	Seq-feat to edit [in]
tse	top-level Seq-entry in which to search for the other half of the xref pair

Returns: Boolean return value indicates whether xrefs were created

Definition at line 854 of file cleanup.cpp.

References CSeq_feat::AddSeqFeatXref(), CSeqFeatData::AllowXref(), CSerialObject::Assign(), CSeqFeatData_Base::e_not_set, CSeq_feat_Handle::GetAnnot(), CSeq_feat_Base::GetData(), CSeq_feat_Handle::GetData(), CSeq_annot_Handle::GetEditHandle(), CTSE_Handle::GetFeaturesWithId(), CObject_id_Base::GetId(), CSeq_feat_Base::GetId(), CSeq_feat_Handle::GetSeq_feat(), CSeqFeatData::GetSubtype(), CSeq_feat_Handle::GetXref(), CSeqFeatData_Base::IsGene(), CFeat_id_Base::IsLocal(), CSeq_feat_Base::IsSetId(), CSeq_feat_Handle::IsSetXref(), ITERATE, and CSeq_feat_EditHandle::Replace().

◆ RepairXrefs() [3/3]

bool CCleanup::RepairXrefs ( CSeq_entry_Handle seh )

static

Repairs non-reciprocal xref pairs in specified seq-entry.

Parameters

seh	Seq-entry to edit [in]

Returns: Boolean return value indicates whether xrefs were created

Definition at line 926 of file cleanup.cpp.

References CMappedFeat::GetSeq_feat(), CSeq_entry_Handle::GetTSE_Handle(), and RepairXrefs().

◆ RescueSiteRefPubs()

bool CCleanup::RescueSiteRefPubs ( CSeq_entry_Handle seh )

static

Rescue pubs from Site-ref features.

Parameters

seh	Seq-entry to edit

Returns: bool indicates whether any changes were made

Definition at line 3457 of file cleanup.cpp.

References CSerialObject::Assign(), b, CNewCleanup_imp::BasicCleanup(), CSeqFeatData_Base::e_Imp, CPub_set_Base::e_Pub, CPubdesc_Base::eReftype_feats, CPubdesc_Base::eReftype_sites, f, CSeq_feat_EditHandle::GetAnnot(), CSeq_annot_Handle::GetCompleteSeq_annot(), IsMinPub(), IsSiteRef(), ITERATE, makeCleanupChange(), MoveOneFeatToPubdesc(), CSeq_annot_EditHandle::Remove(), CSeq_feat_EditHandle::Remove(), CPubdesc_Base::SetPub(), CSeqdesc_Base::SetPub(), CPubdesc_Base::SetReftype(), CNewCleanup_imp::ShouldRemoveAnnot(), ShouldStripPubSerial(), and t.

Referenced by CNewCleanup_imp::x_ExtendedCleanupExtra().

◆ ResetAuthorNames()

void CCleanup::ResetAuthorNames ( CAuth_list::TNames & names )

static

Definition at line 567 of file cleanup_author.cpp.

References names.

Referenced by CleanupAuthList().

◆ s_AddNumToUserField()

bool CCleanup::s_AddNumToUserField ( CUser_field & field )

staticprivate

Definition at line 390 of file cleanup_user_object.cpp.

References CUser_field_Base::C_Data::e_Ints, CUser_field_Base::C_Data::e_Oss, CUser_field_Base::C_Data::e_Reals, CUser_field_Base::C_Data::e_Strs, CUser_field_Base::GetData(), CUser_field_Base::C_Data::GetInts(), CUser_field_Base::GetNum(), CUser_field_Base::C_Data::GetOss(), CUser_field_Base::C_Data::GetReals(), CUser_field_Base::C_Data::GetStrs(), CUser_field_Base::IsSetData(), CUser_field_Base::IsSetNum(), CUser_field_Base::SetNum(), and CUser_field_Base::C_Data::Which().

Referenced by x_CleanupUserField().

◆ s_CleanupDBLink()

bool CCleanup::s_CleanupDBLink ( CUser_object & obj )

staticprivate

Definition at line 370 of file cleanup_user_object.cpp.

References CUser_object::eObjectType_DBLink, CUser_object::GetObjectType(), CUser_object_Base::IsSetData(), CUser_object_Base::SetData(), and val.

◆ s_CleanupGeneOntology()

bool CCleanup::s_CleanupGeneOntology ( CUser_object & obj )

staticprivate

Definition at line 125 of file cleanup_user_object.cpp.

References DEFINE_STATIC_ARRAY_MAP(), NStr::eNocase, NStr::Equal(), CUser_field_Base::GetData(), CUser_field_Base::GetLabel(), CObject_id_Base::GetStr(), CUser_object_Base::GetType(), CUser_field_Base::C_Data::IsFields(), CUser_field_Base::IsSetData(), CUser_object_Base::IsSetData(), CUser_field_Base::IsSetLabel(), CUser_object_Base::IsSetType(), CObject_id_Base::IsStr(), CUser_field_Base::C_Data::IsStr(), CUser_field_Base::SetData(), CUser_object_Base::SetData(), and NStr::StartsWith().

◆ s_CleanupGenomeAssembly()

bool CCleanup::s_CleanupGenomeAssembly ( CUser_object & obj )

staticprivate

◆ s_CleanupNameStdBC()

bool CCleanup::s_CleanupNameStdBC	(	CName_std &	name,
		bool	fix_initials
	)

staticprivate

Definition at line 98 of file cleanup_author.cpp.

◆ s_CleanupStructuredComment()

bool CCleanup::s_CleanupStructuredComment ( CUser_object & obj )

staticprivate

Definition at line 177 of file cleanup_user_object.cpp.

◆ s_ExtractSuffixFromInitials()

void CCleanup::s_ExtractSuffixFromInitials ( CName_std & name )

staticprivate

Definition at line 375 of file cleanup_author.cpp.

References _ASSERT, EXTRACTSUFFIXFROMINITIALS, CName_std_Base::IsSetInitials(), CName_std_Base::IsSetSuffix(), NPOS, and CName_std_Base::SetInitials().

Referenced by s_CleanupNameStdBC().

◆ s_FixEtAl()

void CCleanup::s_FixEtAl ( CName_std & name )

staticprivate

Definition at line 403 of file cleanup_author.cpp.

References NStr::Equal(), CName_std_Base::GetFirst(), CName_std_Base::GetInitials(), CName_std_Base::GetLast(), init(), NStr::IsBlank(), CName_std_Base::IsSetFirst(), CName_std_Base::IsSetInitials(), CName_std_Base::IsSetLast(), CName_std_Base::ResetFirst(), CName_std_Base::ResetInitials(), and CName_std_Base::SetLast().

Referenced by s_CleanupNameStdBC().

◆ s_Flatten()

static bool CCleanup::s_Flatten ( CPub_equiv & pub_equiv )

staticprivate

◆ s_IsProductOnFeat()

bool CCleanup::s_IsProductOnFeat ( const CSeq_feat & cds )

staticprivate

Definition at line 1379 of file cleanup.cpp.

References NStr::EqualNocase(), CSeq_feat_Base::GetQual(), CSeq_feat_Base::GetXref(), CSeq_feat_Base::IsSetQual(), and CSeq_feat_Base::IsSetXref().

◆ s_RemoveEmptyFields()

bool CCleanup::s_RemoveEmptyFields ( CUser_object & obj )

staticprivate

Definition at line 333 of file cleanup_user_object.cpp.

References CUser_field_Base::C_Data::e_not_set, CUser_object::eObjectType_StructuredComment, CUser_object::GetObjectType(), NStr::IsBlank(), CUser_object_Base::IsSetData(), CUser_object_Base::SetData(), and val.

Referenced by s_CleanupStructuredComment().

◆ s_SetProductOnFeat()

void CCleanup::s_SetProductOnFeat	(	CSeq_feat &	feat,
		const string &	protein_name,
		bool	append
	)

staticprivate

Definition at line 1400 of file cleanup.cpp.

References append(), NStr::EqualNocase(), NStr::IsBlank(), CSeq_feat_Base::IsSetQual(), CSeq_feat_Base::IsSetXref(), SetProteinName(), CSeq_feat_Base::SetQual(), and CSeq_feat_Base::SetXref().

Referenced by SetProteinName().

◆ SeqLocExtend()

bool CCleanup::SeqLocExtend	(	CSeq_loc &	loc,
		size_t	pos,
		CScope &	scope
	)

static

Extends a location to the specificed position.

Parameters

loc	Seq-loc to extend
pos	position of new end of location
scope	Scope in which to look for sequences

Returns: Boolean return value indicates whether the location was extended

Definition at line 1038 of file cleanup.cpp.

References CSeq_loc::Assign(), eExtreme_Positional, CSeq_loc::fMerge_AbuttingOnly, CSeq_loc::fSort, CSeq_loc::GetId(), CSeq_loc::GetStart(), CSeq_loc::GetStop(), CSeq_loc::GetStrand(), CSeq_loc::IsPartialStart(), CSeq_loc::IsPartialStop(), Seq_loc_Add(), CSeq_loc::SetPartialStart(), and CSeq_loc::SetPartialStop().

Referenced by ExtendToGapsOrEnds().

◆ SetBestFrame()

bool CCleanup::SetBestFrame	(	CSeq_feat &	cds,
		CScope &	scope
	)

static

Translates coding region and selects best frame (without stops, or longest)

Parameters

cds	Coding region Seq-feat to edit
scope	Scope in which to find coding region

Returns: Boolean return value indicates whether the coding region was changed

Definition at line 1194 of file cleanup.cpp.

References CCdregion_Base::eFrame_not_set, CSeqTranslator::FindBestFrame(), CSeqFeatData_Base::GetCdregion(), CSeq_feat_Base::GetData(), CCdregion_Base::GetFrame(), CCdregion_Base::IsSetFrame(), and CSeq_feat_Base::SetData().

Referenced by WGSCleanup().

◆ SetCDSPartialsByFrameAndTranslation()

bool CCleanup::SetCDSPartialsByFrameAndTranslation	(	CSeq_feat &	cds,
		CScope &	scope
	)

static

1.

Set the partial flags when the CDS is partial and codon_start is 2 or 3 2. Make the CDS partial at the 5' end if there is no start codon 3. Make the CDS partial at the 3' end if there is no stop codon

Parameters

cds	Coding region Seq-feat to edit
scope	Scope in which to find coding region and coding region's protein product sequence

Returns: Boolean return value indicates whether the coding region changed

Definition at line 1513 of file cleanup.cpp.

References AdjustFeaturePartialFlagForLocation(), eExtreme_Biological, CCdregion_Base::eFrame_not_set, CCdregion_Base::eFrame_one, NStr::EndsWith(), CSeqFeatData_Base::GetCdregion(), CSeq_feat_Base::GetData(), CCdregion_Base::GetFrame(), CSeq_feat_Base::GetLocation(), NStr::IsBlank(), CSeq_loc::IsPartialStart(), CSeq_loc::IsPartialStop(), CCdregion_Base::IsSetFrame(), CSeq_feat_Base::SetLocation(), NStr::StartsWith(), and CSeqTranslator::Translate().

Referenced by WGSCleanup().

◆ SetCodeBreakLocation()

void CCleanup::SetCodeBreakLocation	(	CCode_break &	cb,
		size_t	pos,
		const CSeq_feat &	cds
	)

static

utility function for setting code break location given offset pos is the position of the amino acid where the translation exception occurs (starts with 1)

Definition at line 4757 of file cleanup.cpp.

References CSeq_loc_Base::e_Packed_int, CCdregion_Base::eFrame_not_set, CCdregion_Base::eFrame_one, CCdregion_Base::eFrame_three, CCdregion_Base::eFrame_two, eNa_strand_minus, CPacked_seqint_Base::Get(), CSeqFeatData_Base::GetCdregion(), CSeq_feat_Base::GetData(), CCdregion_Base::GetFrame(), CSeq_feat_Base::GetLocation(), CSeq_loc_Base::GetPacked_int(), CSeqFeatData_Base::IsCdregion(), CSeq_feat_Base::IsSetData(), CCdregion_Base::IsSetFrame(), len, offset, CCode_break_Base::ResetLoc(), CCode_break_Base::SetLoc(), CSeq_loc::SetPacked_int(), tmp, and CSeq_loc_Base::Which().

◆ SetFeaturePartial()

bool CCleanup::SetFeaturePartial ( CSeq_feat & f )

static

Set feature partial based on feature location.

Definition at line 1633 of file cleanup.cpp.

References f, CSeq_loc_CI::GetFuzzFrom(), and CSeq_loc_CI::GetFuzzTo().

Referenced by MoveFeatToProtein().

◆ SetFrameFromLoc() [1/2]

bool CCleanup::SetFrameFromLoc	(	CCdregion &	cdregion,
		const CSeq_loc &	loc,
		CScope &	scope
	)

static

Chooses best frame based on location 1.

If the location is 5' complete, then the frame must be one. 2. If the location is 5' partial and 3' complete, select a frame using the value of the location length modulo 3.

Parameters

cdregion	Coding Region in which to set frame
loc	Location to use for setting frame
scope	Scope in which to find location sequence(s)

Returns: Boolean return value indicates whether the frame was changed

Definition at line 1253 of file cleanup.cpp.

References CCdregion_Base::eFrame_not_set, CCdregion_Base::GetFrame(), CCdregion_Base::IsSetFrame(), and CCdregion_Base::SetFrame().

Referenced by CkCdRegion(), CNewCleanup_imp::ImpFeatBC(), and CCleanupApp::x_FixCDS().

◆ SetFrameFromLoc() [2/2]

bool CCleanup::SetFrameFromLoc	(	CCdregion::EFrame &	frame,
		const CSeq_loc &	loc,
		CScope &	scope
	)

static

Definition at line 1211 of file cleanup.cpp.

References _ASSERT, eExtreme_Biological, CCdregion_Base::eFrame_not_set, CCdregion_Base::eFrame_one, CCdregion_Base::eFrame_three, CCdregion_Base::eFrame_two, GetLength(), CSeq_loc::IsPartialStart(), and CSeq_loc::IsPartialStop().

◆ SetGenePartialByLongestContainedFeature()

bool CCleanup::SetGenePartialByLongestContainedFeature	(	CSeq_feat &	gene,
		CScope &	scope
	)

static

Set partialness of gene to match longest feature contained in gene.

Parameters

gene	Seq-feat to edit
scope	Scope in which to find gene

Returns: Boolean return value indicates whether the gene changed

Definition at line 1739 of file cleanup.cpp.

References Compare(), CopyFeaturePartials(), eContains, eSame, fCompareOverlapping, CScope::GetBioseqHandle(), CSeq_feat_Handle::GetData(), GetLength(), CSeq_feat_Base::GetLocation(), CMappedFeat::GetLocation(), CMappedFeat::GetSeq_feat(), CSeqFeatData_Base::IsGene(), len, and CConstRef< C, Locker >::Reset().

Referenced by WGSCleanup().

◆ SetGeneticCodes()

bool CCleanup::SetGeneticCodes ( CBioseq_Handle bsh )

static

Sets genetic codes for coding regions on Bioseq-Handle.

Parameters

Bioseq-Handle to examine

Returns: Boolean indicates whether any coding regions were updated

Definition at line 2122 of file cleanup.cpp.

References CSerialObject::Assign(), CSeqFeatData_Base::e_Cdregion, CSeqdesc_Base::e_Source, CSeqFeatData_Base::GetCdregion(), CCdregion_Base::GetCode(), CSeq_feat_Base::GetData(), CBioSource::GetGenCode(), CGenetic_code::GetId(), CBioSource_Base::GetOrg(), COrg_ref_Base::GetOrgname(), CMappedFeat::GetOriginalFeature(), CSeqdesc_Base::GetSource(), CSeq_feat::HasExceptionText(), CBioseq_Handle::IsNa(), CCdregion_Base::IsSetCode(), CSeq_feat_EditHandle::Replace(), CCdregion_Base::ResetCode(), CCdregion_Base::SetCode(), and CSeq_feat_Base::SetData().

Referenced by CNewCleanup_imp::SetGeneticCode(), WGSCleanup(), and CCleanupApp::x_FixCDS().

◆ SetMolinfoBiomol()

bool CCleanup::SetMolinfoBiomol	(	CBioseq_Handle	seq,
		CMolInfo::EBiomol	biomol
	)

static

Sets MolInfo::biomol for a sequence.

Parameters

seq	Bioseq to edit
biomol	biomol value to set

Returns: Boolean biomol was changed

Definition at line 1802 of file cleanup.cpp.

References CBioseq_EditHandle::AddSeqdesc(), CSeqdesc_Base::e_Molinfo, CMolInfo_Base::GetBiomol(), CBioseq_Handle::GetEditHandle(), CSeqdesc_Base::GetMolinfo(), CMolInfo_Base::IsSetTech(), CMolInfo_Base::SetBiomol(), and CSeqdesc_Base::SetMolinfo().

◆ SetMolinfoTech()

bool CCleanup::SetMolinfoTech	(	CBioseq_Handle	seq,
		CMolInfo::ETech	tech
	)

static

Sets MolInfo::tech for a sequence.

Parameters

seq	Bioseq to edit
tech	tech value to set

Returns: Boolean tech was changed

Definition at line 1776 of file cleanup.cpp.

References CBioseq_EditHandle::AddSeqdesc(), CSeqdesc_Base::e_Molinfo, CMolInfo_Base::eBiomol_peptide, CBioseq_Handle::GetEditHandle(), CBioseq_Handle::GetInst(), CSeqdesc_Base::GetMolinfo(), CMolInfo_Base::GetTech(), CBioseq_Handle::IsAa(), CBioseq_Handle::IsSetInst(), CSeq_inst_Base::IsSetMol(), CMolInfo_Base::IsSetTech(), CMolInfo_Base::SetBiomol(), CSeqdesc_Base::SetMolinfo(), and CMolInfo_Base::SetTech().

Referenced by CNewCleanup_imp::CdRegionEC().

◆ SetMrnaName()

void CCleanup::SetMrnaName	(	CSeq_feat &	mrna,
		const string &	protein_name
	)

static

Definition at line 1358 of file cleanup.cpp.

References NStr::EqualNocase(), CSeq_feat_Base::GetData(), CSeqFeatData_Base::GetRna(), CSeqFeatData_Base::IsRna(), CSeq_feat_Base::IsSetData(), CRNA_ref_Base::IsSetExt(), CSeq_feat_Base::IsSetQual(), CSeq_feat_Base::SetData(), and CSeq_feat_Base::SetQual().

Referenced by WGSCleanup().

◆ SetProteinName() [1/2]

void CCleanup::SetProteinName	(	CProt_ref &	prot,
		const string &	protein_name,
		bool	append
	)

static

Definition at line 1345 of file cleanup.cpp.

References append(), CProt_ref_Base::GetName(), NStr::IsBlank(), CProt_ref_Base::IsSetName(), and CProt_ref_Base::SetName().

Referenced by s_SetProductOnFeat(), SetProteinName(), WGSCleanup(), CCleanupApp::x_FixCDS(), and CNewCleanup_imp::x_ProtGBQualBC().

◆ SetProteinName() [2/2]

void CCleanup::SetProteinName	(	CSeq_feat &	cds,
		const string &	protein_name,
		bool	append,
		CScope &	scope
	)

static

Definition at line 1425 of file cleanup.cpp.

References AddProteinFeature(), append(), CSerialObject::Assign(), CSeqFeatData::eSubtype_prot, CScope::GetBioseqHandle(), CMappedFeat::GetOriginalFeature(), CSeq_feat_Base::GetProduct(), CMappedFeat::GetSeq_feat_Handle(), CSeq_feat_Base::IsSetProduct(), CSeq_feat_Base::IsSetXref(), NON_CONST_ITERATE, prot, CSeq_feat_EditHandle::Replace(), s_SetProductOnFeat(), CSeq_feat_Base::SetData(), SetProteinName(), and CSeq_feat_Base::SetXref().

◆ SetScope()

void CCleanup::SetScope ( CScope * scope )

Definition at line 108 of file cleanup.cpp.

References CScope::AddScope(), CObjectManager::GetInstance(), m_Scope, and CRef< C, Locker >::Reset().

Referenced by CAsnvalThreadState::ProcessSeqAnnot(), CAsnvalThreadState::ProcessSeqEntry(), CAsnvalThreadState::ProcessSeqFeat(), CAsnvalThreadState::ProcessSeqSubmit(), and CAsnvalThreadState::ReadClassMember().

◆ ShouldStripPubSerial()

bool CCleanup::ShouldStripPubSerial ( const CBioseq & bs )

static

Definition at line 4016 of file cleanup.cpp.

References FIELD_IS_SET, GET_FIELD, CBioseq_Base::GetId(), ITERATE, NCBI_SEQID, and CSeq_id_Base::Which().

Referenced by RescueSiteRefPubs(), and CNewCleanup_imp::SetGlobalFlags().

◆ TaxonomyLookup()

bool CCleanup::TaxonomyLookup ( CSeq_entry_Handle seh )

static

Looks up Org-refs in the Seq-entry.

Parameters

seh	Seq-entry to edit

Returns: Boolean return value indicates whether object was updated

Definition at line 1972 of file cleanup.cpp.

References CSerialObject::Assign(), CSeqFeatData_Base::e_Biosrc, CSeqFeatData_Base::GetBiosrc(), CSeq_entry_Handle::GetCompleteSeq_entry(), CSeq_feat_Handle::GetData(), CBioSource_Base::GetOrg(), CSeq_feat_Handle::GetOriginalSeq_feat(), CSeq_entry_Handle::GetScope(), CScope::GetSeq_featHandle(), GetSourceDescriptors(), CBioSource_Base::IsSetOrg(), CSeq_feat_EditHandle::Replace(), CTaxon3::SendOrgRefList(), CSeq_feat_Base::SetData(), CBioSource_Base::SetOrg(), CSeqdesc_Base::SetSource(), and CTaxon3::yes.

◆ UpdateECNumbers()

bool CCleanup::UpdateECNumbers ( CProt_ref::TEc & ec_num_list )

static

Update EC numbers.

Parameters

ec_num_list Prot-ref ec number list to clean

Returns: Boolean value indicates whether any changes were made

Definition at line 1663 of file cleanup.cpp.

References CleanVisStringJunk(), CProt_ref::eEC_replaced, CProt_ref::GetECNumberReplacement(), CProt_ref::GetECNumberStatus(), NStr::IsBlank(), CProt_ref::IsECNumberSplit(), and NON_CONST_ITERATE.

Referenced by FixECNumbers(), and CNewCleanup_imp::x_CleanupECNumberListEC().

◆ ValidAminoAcid()

char CCleanup::ValidAminoAcid ( string_view abbrev )

static

Definition at line 4974 of file cleanup.cpp.

References x_ValidAminoAcid().

Referenced by CkQualPosSeqaa(), and GetQualValueAa().

◆ WGSCleanup()

bool CCleanup::WGSCleanup	(	CSeq_entry_Handle	entry,
		bool	instantiate_missing_proteins = `true`,
		Uint4	options = `0`,
		bool	run_extended_cleanup = `true`
	)

static

Performs WGS specific cleanup.

Parameters

entry Seq-entry to edit

Returns: Boolean return value indicates whether object was updated

Definition at line 2653 of file cleanup.cpp.

References AddProtein(), AddProteinFeature(), AdjustForCDSPartials(), CSerialObject::Assign(), CopyFeaturePartials(), CSeqFeatData_Base::e_Cdregion, CSeqFeatData_Base::e_Gene, CSeqFeatData_Base::e_Rna, eExtreme_Biological, CSeq_inst_Base::eMol_na, eNa_strand_minus, NStr::Equal(), NStr::EqualNocase(), CSeqFeatData::eSubtype_prot, CSeqFeatData::eSubtype_rRNA, ExpandGeneToIncludeChildren(), ExtendedCleanup(), CAliasBase< TPrim >::Get(), CScope::GetBioseqHandle(), CScope::GetBioseqHandleFromTSE(), CBioseq_Handle::GetCompleteBioseq(), CSeq_feat_Base::GetData(), CSeq_loc::GetId(), CBioseq_Handle::GetInst(), CSeq_data_Base::GetIupacaa(), GetLength(), CSeq_feat_Base::GetLocation(), GetmRNAforCDS(), GetNewProtId(), CSeq_feat_Base::GetProduct(), GetProteinName(), CSeq_feat_Base::GetQual(), CSeqFeatData_Base::GetRna(), CRNA_ref::GetRnaProductName(), CSeq_entry_Handle::GetScope(), CSeq_inst_Base::GetSeq_data(), CScope::GetSeq_featHandle(), CSeq_loc::GetStart(), CSeq_loc::GetStop(), CSeq_loc::GetStrand(), CSeqFeatData::GetSubtype(), CSeq_entry_Handle::GetTopLevelEntry(), CSeq_entry_Handle::GetTSE_Handle(), NStr::IsBlank(), IsGeneralIdProtPresent(), CSeq_data_Base::IsIupacaa(), IsPseudo(), CSeq_feat_Base::IsSetData(), CBioseq_Handle::IsSetInst(), CSeq_feat_Base::IsSetProduct(), CSeq_feat_Base::IsSetQual(), CSeq_inst_Base::IsSetSeq_data(), CSeq_loc::IsSetStrand(), NormalizeDescriptorOrder(), prot, RemovePseudoProduct(), CSeq_feat_EditHandle::Replace(), RetranslateCDS(), s_CleanupIsShortrRNA(), SetBestFrame(), SetCDSPartialsByFrameAndTranslation(), SetGenePartialByLongestContainedFeature(), SetGeneticCodes(), CBioseq_EditHandle::SetInst(), CSeq_feat_Base::SetLocation(), SetMrnaName(), CSeq_feat_Base::SetProduct(), SetProteinName(), and CSeq_inst_Base::SetSeq_data().

Referenced by CTable2AsnValidator::Cleanup(), and CCleanupApp::x_ProcessXOptions().

◆ x_AddLowQualityException() [1/2]

bool CCleanup::x_AddLowQualityException	(	CSeq_entry_Handle	entry,
		CSeqFeatData::ESubtype	subtype
	)

staticprivate

Definition at line 2906 of file cleanup.cpp.

References CSerialObject::Assign(), CSeq_entry_Handle::GetScope(), IsPseudo(), CSeq_feat_EditHandle::Replace(), x_AddLowQualityException(), and x_HasShortIntron().

◆ x_AddLowQualityException() [2/2]

bool CCleanup::x_AddLowQualityException ( CSeq_feat & feat )

staticprivate

Definition at line 2888 of file cleanup.cpp.

References NStr::Find(), CSeq_feat_Base::GetExcept_text(), NStr::IsBlank(), CSeq_feat_Base::IsSetExcept(), CSeq_feat_Base::IsSetExcept_text(), kLowQualitySequence, CSeq_feat_Base::SetExcept(), and CSeq_feat_Base::SetExcept_text().

Referenced by AddLowQualityException(), and x_AddLowQualityException().

◆ x_CleanupUserField()

bool CCleanup::x_CleanupUserField ( CUser_field & field )

staticprivate

◆ x_HasShortIntron()

bool CCleanup::x_HasShortIntron	(	const CSeq_loc &	loc,
		size_t	min_len = `11`
	)

staticprivate

Definition at line 2841 of file cleanup.cpp.

References abs, eNa_strand_minus, eNa_strand_plus, CRange_Base::GetFrom(), CSeq_loc_CI::GetRange(), CSeq_loc_CI::GetStrand(), CRange_Base::GetTo(), CSeq_loc_CI::IsEmpty(), and CSeq_loc_CI::IsSetStrand().

Referenced by x_AddLowQualityException().

◆ x_MergeDupOrgNames()

bool CCleanup::x_MergeDupOrgNames	(	COrgName &	on1,
		const COrgName &	add
	)

staticprivate

◆ x_MergeDupOrgRefs()

bool CCleanup::x_MergeDupOrgRefs	(	COrg_ref &	org1,
		const COrg_ref &	add
	)

staticprivate

Definition at line 3726 of file cleanup.cpp.

References a, COrg_ref_Base::GetDb(), COrg_ref_Base::GetMod(), COrg_ref_Base::GetOrgname(), COrg_ref_Base::GetSyn(), HasMod(), COrg_ref_Base::IsSetDb(), COrg_ref_Base::IsSetMod(), COrg_ref_Base::IsSetOrgname(), COrg_ref_Base::IsSetSyn(), ITERATE, COrg_ref_Base::SetDb(), COrg_ref_Base::SetMod(), COrg_ref_Base::SetOrgname(), COrg_ref_Base::SetSyn(), and x_MergeDupOrgNames().

Referenced by MergeDupBioSources().

Member Data Documentation

◆ m_Scope

CRef<CScope> CCleanup::m_Scope

private

Definition at line 613 of file cleanup.hpp.

Referenced by CCleanup(), and SetScope().

The documentation for this class was generated from the following files:

include/objtools/cleanup/cleanup.hpp
src/objtools/cleanup/cleanup.cpp
src/objtools/cleanup/cleanup_author.cpp
src/objtools/cleanup/cleanup_user_object.cpp
src/objtools/cleanup/gene_qual_normalization.cpp

Public Types

Public Member Functions

Static Public Member Functions

Static Private Member Functions

Private Attributes

Additional Inherited Members

Detailed Description

Member Typedef Documentation

◆ TChanges

◆ TFeatGenePair

Member Enumeration Documentation

◆ EScopeOptions

◆ EValidOptions

Constructor & Destructor Documentation

◆ CCleanup() [1/2]

◆ CCleanup() [2/2]

◆ ~CCleanup()

Member Function Documentation

◆ AddGenBankWrapper()

◆ AddLowQualityException()

◆ AddMissingMolInfo()

◆ AddNcbiCleanupObject()

◆ AddPartialToProteinTitle()

◆ AddProtein()

◆ AddProteinTitle()

◆ AreBioSourcesMergeable()

◆ AutodefId()

◆ BasicCleanup() [1/14]

◆ BasicCleanup() [2/14]

◆ BasicCleanup() [3/14]

◆ BasicCleanup() [4/14]

◆ BasicCleanup() [5/14]

◆ BasicCleanup() [6/14]

◆ BasicCleanup() [7/14]

◆ BasicCleanup() [8/14]

◆ BasicCleanup() [9/14]

◆ BasicCleanup() [10/14]

◆ BasicCleanup() [11/14]

◆ BasicCleanup() [12/14]

◆ BasicCleanup() [13/14]

◆ BasicCleanup() [14/14]

◆ BioSrcFromFeat()

◆ CleanupAffil()

◆ CleanupAuthList()

◆ CleanupAuthor()

◆ CleanupCollectionDates()

◆ CleanupUserObject()

◆ ClearInternalPartials() [1/4]

◆ ClearInternalPartials() [2/4]

◆ ClearInternalPartials() [3/4]

◆ ClearInternalPartials() [4/4]

◆ ConvertDeltaSeqToRaw()

◆ ConvertPubFeatsToPubDescs()

◆ ConvertSrcFeatsToSrcDescs()

◆ DecodeXMLMarkChanged()

◆ ExpandGeneToIncludeChildren()

◆ ExtendedCleanup() [1/4]

◆ ExtendedCleanup() [2/4]

◆ ExtendedCleanup() [3/4]

◆ ExtendedCleanup() [4/4]

◆ ExtendStopPosition()

◆ ExtendToStopCodon()

◆ ExtendToStopIfShortAndNotPartial()

◆ FindMatchingLocus_tagGene()

◆ FindMatchingLocusGene()

◆ FixECNumbers()

◆ FixGeneXrefSkew()

◆ FixRNAEditingCodingRegion()

◆ GetCitationList()

◆ GetCodeBreakForLocation()

◆ GetNormalizableGeneQualPairs()

◆ GetProteinLocationFromNucleotideLocation() [1/2]

◆ GetProteinLocationFromNucleotideLocation() [2/2]

◆ GetProteinName() [1/2]

◆ GetProteinName() [2/2]

◆ GetPubdescLabels()

◆ IsEmpty()

◆ IsGeneXrefUnnecessary()

◆ IsMethionine()

◆ IsMinPub()