#include <corelib/ncbistd.hpp>
#include <corelib/ncbistr.hpp>
#include <objects/seq/Seq_descr.hpp>
#include <objects/seq/Seqdesc.hpp>
#include <objects/seqfeat/BioSource.hpp>
#include <objects/seqtable/Seq_table.hpp>
#include <objects/seqtable/SeqTable_column.hpp>
#include <objmgr/bioseq_handle.hpp>

Include dependency graph for biosample_util.hpp:

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

Go to the source code of this file.

Go to the SVN repository for this file.

Classes
class	CBiosampleFieldDiff

Typedefs
typedef map< string, CRef< CSeq_descr > >	TBioSamples

typedef map< string, CRef< CSeq_descr > >::iterator	TBioSamplesIterator

typedef map< string, EStatus >	TStatuses

typedef map< string, EStatus >::iterator	TStatusesIterator

typedef pair< string, biosample_util::EStatus >	TStatus

typedef vector< CRef< CBiosampleFieldDiff > >	TBiosampleFieldDiffList

Enumerations
enum	EStatus { eStatus_Unknown = 0 , eStatus_Live , eStatus_Hup , eStatus_Withdrawn , eStatus_Suppressed , eStatus_ToBeCurated , eStatus_Replaced }

Functions
CRef< CSeq_descr >	GetBiosampleData (const string &accession, bool use_dev_server=false, TBioSamples *cache=NULL)

EStatus	GetBiosampleStatus (const string &accession, bool use_dev_server=false, TStatuses *cache=NULL)

void	GetBiosampleStatus (TStatuses &status, bool use_dev_server=false)

string	GetBiosampleStatusName (EStatus status)

vector< string >	GetBiosampleIDs (CBioseq_Handle bh)

vector< string >	GetBioProjectIDs (CBioseq_Handle bh)

TBiosampleFieldDiffList	GetBioseqDiffs (CBioseq_Handle bh, const string &biosample_accession, size_t &num_processed, vector< string > &unprocessed_ids, bool use_dev_server=false, bool compare_structured_comments=false, const string &expected_prefix="", TBioSamples *cache=NULL)

TBiosampleFieldDiffList	GetFieldDiffs (const string &sequence_id, const string &biosample_id, const CBioSource &src, const CBioSource &sample)

TBiosampleFieldDiffList	GetFieldDiffs (const string &sequence_id, const string &biosample_id, CConstRef< CUser_object > src, CConstRef< CUser_object > sample)

bool	ResolveSuppliedBioSampleAccession (const string &biosample_accession, vector< string > &biosample_ids)

bool	DoDiffsContainConflicts (const TBiosampleFieldDiffList &diffs, CNcbiOstream *log)

void	AddBioseqToTable (CBioseq_Handle bh, CSeq_table &table, bool with_id, bool include_comments=false, const string &expected_prefix="")

string	GetBestBioseqLabel (CBioseq_Handle bsh)

bool	AttributeNamesAreEquivalent (string name1, string name2)

void	PrintBioseqXML (CBioseq_Handle bh, const string &id_prefix, CNcbiOstream *report_stream, const string &bioproject_accession, const string &default_owner, const string &hup_date, const string &comment, bool first_seq_only, bool report_structured_comments, const string &expected_prefix)

string	OwnerFromAffil (const CAffil &affil)

void	GenerateDiffListFromBioSource (const CSeq_descr &bioSample, const CBioSource &bioSource, TBiosampleFieldDiffList &diffs)

bool	GenerateDiffListFromBioSource (const string &bioSampleAcc, const CBioSource &bioSource, CBioSource &bioSampleSource, TBiosampleFieldDiffList &diffs)

bool	UpdateBiosourceFromBiosample (const CBioSource &existingBioSource, CBioSource &newBioource)

bool	UpdateBiosourceFromBiosample (const TBiosampleFieldDiffList &diffs, const CBioSource &existingBioSource, CBioSource &newBioource)

void	PrettyPrint (const TBiosampleFieldDiffList &diffList, CNcbiOstream &ostr, size_t keyWidth=20, size_t valueWidth=40)

Typedef Documentation

◆ TBiosampleFieldDiffList

typedef vector< CRef<CBiosampleFieldDiff> > TBiosampleFieldDiffList

Definition at line 119 of file biosample_util.hpp.

◆ TBioSamples

typedef map< string, CRef< CSeq_descr > > TBioSamples

Definition at line 52 of file biosample_util.hpp.

◆ TBioSamplesIterator

typedef map< string, CRef< CSeq_descr > >::iterator TBioSamplesIterator

Definition at line 53 of file biosample_util.hpp.

◆ TStatus

typedef pair<string, biosample_util::EStatus> TStatus

Definition at line 69 of file biosample_util.hpp.

◆ TStatuses

typedef map<string, EStatus> TStatuses

Definition at line 67 of file biosample_util.hpp.

◆ TStatusesIterator

typedef map<string, EStatus>::iterator TStatusesIterator

Definition at line 68 of file biosample_util.hpp.

Enumeration Type Documentation

◆ EStatus

enum EStatus

Enumerator
eStatus_Unknown
eStatus_Live
eStatus_Hup
eStatus_Withdrawn
eStatus_Suppressed
eStatus_ToBeCurated
eStatus_Replaced

Definition at line 57 of file biosample_util.hpp.

Function Documentation

◆ AddBioseqToTable()

void AddBioseqToTable	(	CBioseq_Handle	bh,
		CSeq_table &	table,
		bool	with_id,
		bool	include_comments = `false`,
		const string &	expected_prefix = `""`
	)

Definition at line 751 of file biosample_util.cpp.

References AddValueToTable(), CSeqdesc_Base::e_Pub, CSeqdesc_Base::e_Source, CSeqdesc_Base::e_User, eUnknown, CPub_equiv_Base::Get(), CAffil_Base::C_Std::GetAffil(), GetBestBioseqLabel(), GetBioProjectIDs(), GetBiosampleIDs(), CAffil_Base::C_Std::GetDiv(), CBioSource::GetNameValPairs(), CPubdesc_Base::GetPub(), CSeqdesc_Base::GetPub(), CSeqdesc_Base::GetSource(), CAffil_Base::GetStd(), CAffil_Base::GetStr(), GetStructuredCommentFields(), CSeqdesc_Base::GetUser(), i, int, CAffil_Base::C_Std::IsSetAffil(), CAffil_Base::C_Std::IsSetDiv(), CAffil_Base::IsStd(), CAffil_Base::IsStr(), ITERATE, kAffilDept, kAffilInst, kBioProject, kSequenceID, label, msg(), NCBI_THROW, row, s_IsCitSub(), s_IsReportableStructuredComment(), table, and val.

Referenced by CBiosampleChkApp::ProcessBioseqHandle().

◆ AttributeNamesAreEquivalent()

bool AttributeNamesAreEquivalent	(	string	name1,
		string	name2
	)

Definition at line 847 of file biosample_util.cpp.

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

Referenced by AddStructuredCommentToAttributes().

◆ DoDiffsContainConflicts()

bool DoDiffsContainConflicts	(	const TBiosampleFieldDiffList &	diffs,
		CNcbiOstream *	log
	)

Definition at line 606 of file biosample_util.cpp.

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

Referenced by CBiosampleChkApp::ProcessBioseqHandle().

◆ GenerateDiffListFromBioSource() [1/2]

void GenerateDiffListFromBioSource	(	const CSeq_descr &	bioSample,
		const CBioSource &	bioSource,
		TBiosampleFieldDiffList &	diffs
	)

Definition at line 1394 of file biosample_util.cpp.

References CSeq_descr_Base::Get(), GetFieldDiffs(), CSeqdesc_Base::GetSource(), and CSeqdesc_Base::IsSource().

Referenced by CompareBioSampleAccessionToBioSource(), GenerateDiffListFromDescriptors(), and CBsDiffApp::xCompareSeqEntry().

◆ GenerateDiffListFromBioSource() [2/2]

bool GenerateDiffListFromBioSource	(	const string &	bioSampleAcc,
		const CBioSource &	bioSource,
		CBioSource &	bioSampleSource,
		TBiosampleFieldDiffList &	diffs
	)

Definition at line 1431 of file biosample_util.cpp.

References CSerialObject::Assign(), CSeq_descr_Base::Get(), GetBiosampleData(), GetFieldDiffs(), CSeqdesc_Base::GetSource(), CSeqdesc_Base::IsSource(), and SaveSerialObject().

◆ GetBestBioseqLabel()

string GetBestBioseqLabel ( CBioseq_Handle bsh )

Definition at line 881 of file biosample_util.cpp.

References CSeq_id::BestRank(), FindBestChoice(), CBioseq_Handle::GetId(), CConstRef< C, Locker >::GetPointer(), CSeq_id_Base::IsGenbank(), ITERATE, label, and NULL.

Referenced by AddBioseqToTable(), GetBioseqDiffs(), PrintBioseqXML(), CBiosampleChkApp::ProcessBioseqForUpdate(), and CBiosampleChkApp::ProcessBioseqHandle().

◆ GetBioProjectIDs()

vector<string> GetBioProjectIDs ( CBioseq_Handle bh )

Definition at line 343 of file biosample_util.cpp.

References CSeqdesc_Base::e_User, GetDBLinkIDs(), and ITERATE.

Referenced by AddBioseqToTable(), and PrintBioseqXML().

◆ GetBiosampleData()

CRef< CSeq_descr > GetBiosampleData	(	const string &	accession,
		bool	use_dev_server = `false`,
		TBioSamples *	cache = `NULL`
	)

Definition at line 93 of file biosample_util.cpp.

References map_checker< Container >::end(), map_checker< Container >::find(), NStr::IntToString(), NULL, PrepareUrl(), and CRef< C, Locker >::Reset().

Referenced by BOOST_AUTO_TEST_CASE(), GenerateDiffListFromBioSource(), GetBioseqDiffs(), LoadBioSampleFromAcc(), and CBiosampleChkApp::ProcessBioseqForUpdate().

◆ GetBiosampleIDs()

vector<string> GetBiosampleIDs ( CBioseq_Handle bh )

Definition at line 327 of file biosample_util.cpp.

References ITERATE.

Referenced by AddBioseqToTable(), GetBioseqDiffs(), PrintBioseqXML(), CBiosampleStatusReport::ProcessBioseq(), and CBiosampleChkApp::ProcessBioseqForUpdate().

◆ GetBiosampleStatus() [1/2]

EStatus GetBiosampleStatus	(	const string &	accession,
		bool	use_dev_server = `false`,
		TStatuses *	cache = `NULL`
	)

Definition at line 184 of file biosample_util.cpp.

References xml::node::begin(), map_checker< Container >::end(), xml::node::end(), NStr::Equal(), eStatus_Unknown, map_checker< Container >::find(), xml::node::get_name(), xml::document::get_root_node(), GetBioSampleStatusFromNode(), PrepareUrl(), and ProcessBiosampleStatusNode().

Referenced by CBiosampleStatusReport::AddSummary(), and CBiosampleChkApp::Run().

◆ GetBiosampleStatus() [2/2]

void GetBiosampleStatus	(	TStatuses &	status,
		bool	use_dev_server = `false`
	)

Definition at line 245 of file biosample_util.cpp.

References map_checker< Container >::begin(), count, map_checker< Container >::end(), NStr::IsBlank(), and ProcessBulkBioSample().

◆ GetBiosampleStatusName()

string GetBiosampleStatusName ( EStatus status )

Definition at line 264 of file biosample_util.cpp.

References eStatus_Hup, eStatus_Live, eStatus_Replaced, eStatus_Suppressed, eStatus_ToBeCurated, eStatus_Unknown, eStatus_Withdrawn, and kEmptyStr.

Referenced by CBiosampleStatusReport::AddSummary(), and CBiosampleChkApp::Run().

◆ GetBioseqDiffs()

TBiosampleFieldDiffList GetBioseqDiffs	(	CBioseq_Handle	bh,
		const string &	biosample_accession,
		size_t &	num_processed,
		vector< string > &	unprocessed_ids,
		bool	use_dev_server = `false`,
		bool	compare_structured_comments = `false`,
		const string &	expected_prefix = `""`,
		TBioSamples *	cache = `NULL`
	)

◆ GetFieldDiffs() [1/2]

TBiosampleFieldDiffList GetFieldDiffs	(	const string &	sequence_id,
		const string &	biosample_id,
		CConstRef< CUser_object >	src,
		CConstRef< CUser_object >	sample
	)

Definition at line 573 of file biosample_util.cpp.

References GetAvailableFields(), ITERATE, and s_ShouldIgnoreStructuredCommentFieldDiff().

◆ GetFieldDiffs() [2/2]

TBiosampleFieldDiffList GetFieldDiffs	(	const string &	sequence_id,
		const string &	biosample_id,
		const CBioSource &	src,
		const CBioSource &	sample
	)

Definition at line 515 of file biosample_util.cpp.

References CBioSource::GetBiosampleDiffs(), and ITERATE.

Referenced by GenerateDiffListFromBioSource(), and GetBioseqDiffs().

◆ OwnerFromAffil()

string OwnerFromAffil ( const CAffil & affil )

Definition at line 1168 of file biosample_util.cpp.

References NStr::EqualNocase(), CAffil_Base::C_Std::GetAffil(), CAffil_Base::C_Std::GetDiv(), CAffil_Base::GetStd(), CAffil_Base::GetStr(), CAffil_Base::C_Std::IsSetAffil(), CAffil_Base::C_Std::IsSetDiv(), CAffil_Base::IsStd(), CAffil_Base::IsStr(), and NStr::Join().

Referenced by PrintBioseqXML(), and CBiosampleChkApp::ProcessSeqSubmit().

◆ PrettyPrint()

void PrettyPrint	(	const TBiosampleFieldDiffList &	diffList,
		CNcbiOstream &	ostr,
		size_t	keyWidth = `20`,
		size_t	valueWidth = `40`
	)

Definition at line 1604 of file biosample_util.cpp.

References sPrintField(), and string.

Referenced by PrintDiffList().

◆ PrintBioseqXML()

void PrintBioseqXML	(	CBioseq_Handle	bh,
		const string &	id_prefix,
		CNcbiOstream *	report_stream,
		const string &	bioproject_accession,
		const string &	default_owner,
		const string &	hup_date,
		const string &	comment,
		bool	first_seq_only,
		bool	report_structured_comments,
		const string &	expected_prefix
	)

Definition at line 1190 of file biosample_util.cpp.

References AddBioSourceToAttributes(), AddContact(), AddStructuredCommentToAttributes(), data, CSeqdesc_Base::e_Pub, CSeqdesc_Base::e_Source, CSeqdesc_Base::e_User, CTime::eCurrent, NStr::EqualNocase(), eUnknown, CPub_equiv_Base::Get(), xml::node::get_attributes(), xml::document::get_root_node(), CAuth_list_Base::GetAffil(), GetBestBioseqLabel(), GetBioProjectIDs(), GetBiosampleIDs(), CPubdesc_Base::GetPub(), CSeqdesc_Base::GetPub(), CSeqdesc_Base::GetSource(), CSeqdesc_Base::GetUser(), xml::attributes::insert(), xml::node::insert(), NStr::IsBlank(), CAuth_list_Base::IsSetAffil(), CPubdesc_Base::IsSetPub(), ITERATE, msg(), NCBI_THROW, NULL, OwnerFromAffil(), NStr::ReplaceInPlace(), s_IsReportableStructuredComment(), xml::document::set_encoding(), and xml::document::set_is_standalone().

Referenced by CBiosampleChkApp::ProcessBioseqHandle().

◆ ResolveSuppliedBioSampleAccession()

bool ResolveSuppliedBioSampleAccession	(	const string &	biosample_accession,
		vector< string > &	biosample_ids
	)

Definition at line 855 of file biosample_util.cpp.

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

Referenced by GetBioseqDiffs(), and CBiosampleChkApp::ProcessBioseqForUpdate().

◆ UpdateBiosourceFromBiosample() [1/2]

bool UpdateBiosourceFromBiosample	(	const CBioSource &	existingBioSource,
		CBioSource &	newBioource
	)

Definition at line 1467 of file biosample_util.cpp.

References CSerialObject::Assign(), COrgName_Base::GetMod(), CBioSource_Base::GetOrg(), COrg_ref_Base::GetOrgname(), CBioSource_Base::GetSubtype(), COrg_ref::GetTaxId(), CBioSource::HasSubtype(), CBioSource_Base::IsSetOrg(), COrg_ref_Base::IsSetOrgname(), CBioSource_Base::IsSetSubtype(), CBioSource::IsSetTaxname(), CBioSource_Base::SetOrg(), CBioSource_Base::SetSubtype(), and ZERO_TAX_ID.

◆ UpdateBiosourceFromBiosample() [2/2]

bool UpdateBiosourceFromBiosample	(	const TBiosampleFieldDiffList &	diffs,
		const CBioSource &	existingBioSource,
		CBioSource &	newBioource
	)

Definition at line 1546 of file biosample_util.cpp.

References COrgName_Base::GetMod(), CBioSource_Base::GetOrg(), COrg_ref_Base::GetOrgname(), CBioSource_Base::GetSubtype(), CSubSource::GetSubtypeName(), COrgMod::GetSubtypeName(), CBioSource_Base::IsSetOrg(), COrg_ref_Base::IsSetOrgname(), CBioSource_Base::IsSetSubtype(), CBioSource_Base::SetOrg(), CBioSource_Base::SetSubtype(), sGetDiffByFieldName(), and ZERO_TAX_ID.

Classes

Typedefs

Enumerations

Functions

Typedef Documentation

◆ TBiosampleFieldDiffList

◆ TBioSamples

◆ TBioSamplesIterator

◆ TStatus

◆ TStatuses

◆ TStatusesIterator

Enumeration Type Documentation

◆ EStatus

Function Documentation

◆ AddBioseqToTable()

◆ AttributeNamesAreEquivalent()

◆ DoDiffsContainConflicts()

◆ GenerateDiffListFromBioSource() [1/2]

◆ GenerateDiffListFromBioSource() [2/2]

◆ GetBestBioseqLabel()

◆ GetBioProjectIDs()

◆ GetBiosampleData()

◆ GetBiosampleIDs()

◆ GetBiosampleStatus() [1/2]

◆ GetBiosampleStatus() [2/2]

◆ GetBiosampleStatusName()

◆ GetBioseqDiffs()

◆ GetFieldDiffs() [1/2]

◆ GetFieldDiffs() [2/2]

◆ OwnerFromAffil()

◆ PrettyPrint()

◆ PrintBioseqXML()

◆ ResolveSuppliedBioSampleAccession()

◆ UpdateBiosourceFromBiosample() [1/2]

◆ UpdateBiosourceFromBiosample() [2/2]