uk.ac.roslin.ensembl.dao.database

Class DBRegistry

java.lang.Object

uk.ac.roslin.ensembl.dao.database.DBRegistry

All Implemented Interfaces:

Registry

public class DBRegistry
extends Object
implements Registry

Field Summary

Fields

Modifier and Type	Field and Description
protected TreeSet<DBDatabase>	allDatabases
protected boolean	alreadyInitialized
protected boolean	alreadyParsed
protected TreeMap<String,String>	assemblyAccessionStemForAliasHash
protected TreeSet<DBCollectionDatabase>	collectionDatabases
protected TreeMap<String,TreeMap<String,DBCollection>>	collectionNameDBVersionHash
protected List<DBCollection>	collections
protected TreeMap<String,DBCollectionSpecies>	collectionspeciesHashByGCAccession
protected HashMap<EnsemblComparaDivision,HashMap<String,DBComparisonDatabase>>	comparaDBByDivision
protected TreeSet<DBComparisonDatabase>	comparisonDatabases
protected Properties	configProperties
protected DBConnection	dbConnection
protected TreeMap<String,String>	ensemblNameForAliasHash
protected TreeSet<Integer>	knownSchemaVersions
protected String	newline
protected TreeMap<String,DBCollectionSpecies>	oldstyle_collectionspeciesHashByDBName
protected Reader	reader
protected RegistryConfiguration	regConfig
protected HashMap<String,String>	renamedDBs
protected SchemaVersion	schemaVersion
protected TreeSet<DBSingleSpeciesDatabase>	singleSpeciesDatabases
protected TreeMap<String,DBSpecies>	speciesHash
protected org.apache.ibatis.session.SqlSessionFactory	sqlMapper
protected StringBuilder	tooNewDB
protected StringBuilder	unknownDB

Method Summary

Methods

Modifier and Type	Method and Description
boolean	addSessionAlias(Species sp, String alias)
boolean	addSessionAlias(String ensembl_genus_species_name, String alias) Method for an adhoc alias to be added by the user in this session.
static DBRegistry	createEmptyRegistry()
static DBRegistry	createRegistryForConfiguration(RegistryConfiguration conf)
static DBRegistry	createRegistryForDataSource(DBConnection.DataSource type)
static DBRegistry	createRegistryForDataSourceAtReleaseVersion(DBConnection.DataSource type, Integer release)
static DBRegistry	createRegistryForDataSourceCurrentRelease(DBConnection.DataSource type)
static DBRegistry	createUninitializedRegistryForConfiguration(RegistryConfiguration conf)
static DBRegistry	createUninitializedRegistryForDataSource(DBConnection.DataSource type)
String	findMybatisSchemaForSchemaVersion(DatabaseType type, String schema_version)
Set<? extends CollectionSpecies>	getBacterialSpeciesForAliasBeginning(String alias)
Set<? extends CollectionSpecies>	getBacterialSpeciesForAliasContaining(String alias)
StringBuilder	getBriefRegistryReport() Reports the connection and available database versions etc. for an initialized Registry.
DBCollection	getCollection(String name, String dbVersion)
Collection<DBCollection>	getCollectionRegistriesByName(String name)
List<DBCollection>	getCollections()
Collection<DBCollection>	getCollectionsByDBVersion(String dbVersion)
List<DBCollectionSpecies>	getCollectionSpecies() Returns the combined list of DBCollectionSpecies that are indexed by dbstylename (old_style_ and those indexed by GCAssemblyAccession (new style, post v16)
ComparisonDatabase	getComparaDatabase(EnsemblComparaDivision comparaDivision, String db_version)
Properties	getConfigProperties()
DBCollectionSpecies	getCSpeciesByGCAccessionStem(String key)
DBDatabase	getDatabase(String species_name_or_alias)
DBDatabase	getDatabase(String speciesNameOrAlias_collectionName_comparaDivision, DatabaseType database_type, String db_version)
DBDatabase	getDatabase(String species_name_or_alias, String db_version)
DBDatabase	getDatabaseForFullName(String fullname)
TreeSet<DBDatabase>	getDatabasesByType(DatabaseType type)
DBConnection.DataSource	getDatasourceType()
TranscriptionEngine	getDefaultTranscriptionEngine() note by default translation by this default engine is configured to convert non-Methionine initiations to Methionine
String	getEnsemblNameForAlias(String alias) Looks up the ensembl_genus_species_name for the given alias from the ensemblNameForAliasHash hash of aliases in this Registry.
Set<String>	getEnsemblNamesForAliasBeginning(String alias)
Set<String>	getEnsemblNamesForAliasContaining(String alias)
List<DBCollectionSpecies>	getGCAccessionedCollectionSpecies() Utility method to return only those DBCOllectionSpecies that possess and are indexed by a GCAssemblyAccession(Stem or Chain)
String	getGCAssemblyAccessionForAlias(String alias) Looks up the GC_assembly_accession_stem for the given alias of a collectionspecies.
Set<String>	getGCAssemblyAccessionsForAliasBeginning(String alias)
Set<String>	getGCAssemblyAccessionsForAliasContaining(String alias)
int	getHighestEnsemblSchemaVersion()
int	getHighestReleaseVersion()
TreeSet<Integer>	getKnownSchemaVersions()
DBCollection	getMostRecentCollection(String name)
String	getMostRecentDBVersion(String speciesOrCollection)
String	getMostRecentEnsemblVersion()
DBSpecies	getMostRecentSpeciesByAlias(String alias_or_name) Returns the most recent Species matching the alias.
List<DBCollectionSpecies>	getNamedCollectionSpecies() Utility method to return only those DBCOllectionSpecies that do not possess and are therefore not indexed by a GCAssemblyAccession but bey dbstylename
Set<? extends CollectionSpecies>	getOldStyleBacterialSpeciesForAliasBeginning(String alias)
Set<? extends CollectionSpecies>	getOldStyleBacterialSpeciesForAliasContaining(String alias)
Set<DBSpecies>	getPanComparaSpecies(String dbVersion)
File	getRegistryReport() Reports the connection, available database versions, species and actual database names for an initialized Registry.
HashMap<String,String>	getRenamedDBs()
Collection<? extends Species>	getSetOfEverySpeciesByAlias(String alias_or_name) Returns a HashSet of all DBSpecies and old and newstyle CollectionSpecies matching the given alias.
Collection<? extends Species>	getSetOfSpeciesByAlias(String alias_or_name) Returns a HashSet matching the given alias.
List<DBSpecies>	getSpecies() Fetches the list of all DBSpecies objects for this Registry.
DBSpecies	getSpeciesByAlias(String alias_or_name) Wraps getSetOfSpeciesByAlias(String alias_or_name) to return a single species if the returned Collection has only one member.
DBSpecies	getSpeciesByAlias(String alias_or_name, String version) Wraps getSetOfEverySpeciesByAlias(String alias_or_name) to return a single species if the returned Collection has only one member with the desired alias and version.
DBSpecies	getSpeciesByEnsemblName(String ensemblName)
Set<? extends Species>	getSpeciesForAliasBeginning(String alias)
Set<? extends Species>	getSpeciesForAliasContaining(String alias)
TranscriptionEngine	getTranscriptionEngine(Integer codonTable)
StringBuilder	getVersionReport() Reports the available database versions for this Registry.
void	initialize(Integer release) private initialisation method for the registry which connects to ensembl via a DBDatasourceDAO and retrieves details on all the current tables and creates database objects of the appropriate type for each.
boolean	isSpeciesInPanHomology(Species querySpecies, String dbVersion)
DBDatabase	makeDatabase(String db_name)
void	setConfiguration(DBConnection.DataSource type)
void	setConfiguration(RegistryConfiguration conf)
void	setSpeciesMetadata(CoreDatabase db) should only be called for non current singlespecies core databases

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

sqlMapper

protected org.apache.ibatis.session.SqlSessionFactory sqlMapper

configProperties

protected Properties configProperties

schemaVersion

protected SchemaVersion schemaVersion

dbConnection

protected DBConnection dbConnection

knownSchemaVersions

protected TreeSet<Integer> knownSchemaVersions

reader

protected Reader reader

speciesHash

protected TreeMap<String,DBSpecies> speciesHash

oldstyle_collectionspeciesHashByDBName

protected TreeMap<String,DBCollectionSpecies> oldstyle_collectionspeciesHashByDBName

collectionspeciesHashByGCAccession

protected TreeMap<String,DBCollectionSpecies> collectionspeciesHashByGCAccession

collections

protected List<DBCollection> collections

collectionNameDBVersionHash

protected TreeMap<String,TreeMap<String,DBCollection>> collectionNameDBVersionHash

ensemblNameForAliasHash

protected TreeMap<String,String> ensemblNameForAliasHash

assemblyAccessionStemForAliasHash

protected TreeMap<String,String> assemblyAccessionStemForAliasHash

allDatabases

protected TreeSet<DBDatabase> allDatabases

singleSpeciesDatabases

protected TreeSet<DBSingleSpeciesDatabase> singleSpeciesDatabases

comparisonDatabases

protected TreeSet<DBComparisonDatabase> comparisonDatabases

comparaDBByDivision

protected HashMap<EnsemblComparaDivision,HashMap<String,DBComparisonDatabase>> comparaDBByDivision

collectionDatabases

protected TreeSet<DBCollectionDatabase> collectionDatabases

newline

protected String newline

tooNewDB

protected StringBuilder tooNewDB

unknownDB

protected StringBuilder unknownDB

regConfig

protected RegistryConfiguration regConfig

alreadyInitialized

protected boolean alreadyInitialized

alreadyParsed

protected boolean alreadyParsed

renamedDBs

protected HashMap<String,String> renamedDBs

Method Detail

createEmptyRegistry

public static DBRegistry createEmptyRegistry()

createRegistryForDataSource

public static DBRegistry createRegistryForDataSource(DBConnection.DataSource type)
                                              throws ConfigurationException,
                                                     DAOException

Throws:

ConfigurationException

DAOException

createUninitializedRegistryForDataSource

public static DBRegistry createUninitializedRegistryForDataSource(DBConnection.DataSource type)
                                                           throws ConfigurationException,
                                                                  DAOException

Throws:

ConfigurationException

DAOException

createRegistryForConfiguration

public static DBRegistry createRegistryForConfiguration(RegistryConfiguration conf)
                                                 throws ConfigurationException,
                                                        DAOException

Throws:

ConfigurationException

DAOException

createUninitializedRegistryForConfiguration

public static DBRegistry createUninitializedRegistryForConfiguration(RegistryConfiguration conf)
                                                              throws ConfigurationException,
                                                                     DAOException

Throws:

ConfigurationException

DAOException

createRegistryForDataSourceCurrentRelease

public static DBRegistry createRegistryForDataSourceCurrentRelease(DBConnection.DataSource type)
                                                            throws ConfigurationException,
                                                                   DAOException

Throws:

ConfigurationException

DAOException

createRegistryForDataSourceAtReleaseVersion

public static DBRegistry createRegistryForDataSourceAtReleaseVersion(DBConnection.DataSource type,
                                                     Integer release)
                                                              throws ConfigurationException,
                                                                     DAOException

Throws:

ConfigurationException

DAOException

setConfiguration

public void setConfiguration(RegistryConfiguration conf)
                      throws ConfigurationException,
                             DAOException

Throws:

ConfigurationException

DAOException

setConfiguration

public void setConfiguration(DBConnection.DataSource type)
                      throws ConfigurationException,
                             DAOException

Throws:

ConfigurationException

DAOException

initialize

public void initialize(Integer release)
                throws DAOException,
                       ConfigurationException

private initialisation method for the registry which connects to ensembl via a DBDatasourceDAO and retrieves details on all the current tables and creates database objects of the appropriate type for each. Also retrieves dbSpeciesName aliases for all the current release dbSpeciesName. And finds out what dbSpeciesName are present in multispecies databases.

Parameters:

release -

Throws:

DAOException, - ConfigurationException

ConfigurationException

DAOException

getVersionReport

public StringBuilder getVersionReport()

Reports the available database versions for this Registry. Will successfully report for either initialized or uninitialized Registries. since v74

Specified by:

getVersionReport in interface Registry

Returns:

StringBuilder

getBriefRegistryReport

public StringBuilder getBriefRegistryReport()

Reports the connection and available database versions etc. for an initialized Registry. Lists databases for which this API is not configured (including those too old or too recent).

Specified by:

getBriefRegistryReport in interface Registry

Returns:

StringBuilder

getRegistryReport

public File getRegistryReport()
                       throws IOException

Reports the connection, available database versions, species and actual database names for an initialized Registry. Lists databases for which this API is not configured (including those too old or too recent).

Specified by:

getRegistryReport in interface Registry

Returns:

File

Throws:

IOException

getConfigProperties

public Properties getConfigProperties()

Specified by:

getConfigProperties in interface Registry

getMostRecentEnsemblVersion

public String getMostRecentEnsemblVersion()

Specified by:

getMostRecentEnsemblVersion in interface Registry

getMostRecentDBVersion

public String getMostRecentDBVersion(String speciesOrCollection)
                              throws NonUniqueException

Throws:

NonUniqueException

getKnownSchemaVersions

public TreeSet<Integer> getKnownSchemaVersions()

getHighestEnsemblSchemaVersion

public int getHighestEnsemblSchemaVersion()
                                   throws DAOException,
                                          ConfigurationException

Specified by:

getHighestEnsemblSchemaVersion in interface Registry

Throws:

DAOException

ConfigurationException

getHighestReleaseVersion

public int getHighestReleaseVersion()
                             throws DAOException,
                                    ConfigurationException

Specified by:

getHighestReleaseVersion in interface Registry

Throws:

DAOException

ConfigurationException

getDatabase

public DBDatabase getDatabase(String species_name_or_alias)
                       throws NonUniqueException

Specified by:

getDatabase in interface Registry

Throws:

NonUniqueException

getDatabase

public DBDatabase getDatabase(String species_name_or_alias,
                     String db_version)
                       throws NonUniqueException

Specified by:

getDatabase in interface Registry

Throws:

NonUniqueException

getDatabase

public DBDatabase getDatabase(String speciesNameOrAlias_collectionName_comparaDivision,
                     DatabaseType database_type,
                     String db_version)
                       throws NonUniqueException

Specified by:

getDatabase in interface Registry

Throws:

NonUniqueException

getComparaDatabase

public ComparisonDatabase getComparaDatabase(EnsemblComparaDivision comparaDivision,
                                    String db_version)

Specified by:

getComparaDatabase in interface Registry

getDatabaseForFullName

public DBDatabase getDatabaseForFullName(String fullname)

Specified by:

getDatabaseForFullName in interface Registry

getDatabasesByType

public TreeSet<DBDatabase> getDatabasesByType(DatabaseType type)

Specified by:

getDatabasesByType in interface Registry

findMybatisSchemaForSchemaVersion

public String findMybatisSchemaForSchemaVersion(DatabaseType type,
                                       String schema_version)

Specified by:

findMybatisSchemaForSchemaVersion in interface Registry

getEnsemblNameForAlias

public String getEnsemblNameForAlias(String alias)
                              throws NonUniqueException

Looks up the ensembl_genus_species_name for the given alias from the ensemblNameForAliasHash hash of aliases in this Registry.

Specified by:

getEnsemblNameForAlias in interface Registry

Parameters:

alias - String

Returns:

String the ensembl_genus_species_name

Throws:

NonUniqueException

getGCAssemblyAccessionForAlias

public String getGCAssemblyAccessionForAlias(String alias)
                                      throws NonUniqueException

Looks up the GC_assembly_accession_stem for the given alias of a collectionspecies. Post v16 this is the preferred identifier for a CollectionSpecies.

Parameters:

alias - String

Returns:

String the GC_assembly_accession_stem

Throws:

NonUniqueException

getEnsemblNamesForAliasBeginning

public Set<String> getEnsemblNamesForAliasBeginning(String alias)

Specified by:

getEnsemblNamesForAliasBeginning in interface Registry

getEnsemblNamesForAliasContaining

public Set<String> getEnsemblNamesForAliasContaining(String alias)

Specified by:

getEnsemblNamesForAliasContaining in interface Registry

getGCAssemblyAccessionsForAliasBeginning

public Set<String> getGCAssemblyAccessionsForAliasBeginning(String alias)

Specified by:

getGCAssemblyAccessionsForAliasBeginning in interface Registry

getGCAssemblyAccessionsForAliasContaining

public Set<String> getGCAssemblyAccessionsForAliasContaining(String alias)

Specified by:

getGCAssemblyAccessionsForAliasContaining in interface Registry

getSpeciesForAliasBeginning

public Set<? extends Species> getSpeciesForAliasBeginning(String alias)

Specified by:

getSpeciesForAliasBeginning in interface Registry

getSpeciesForAliasContaining

public Set<? extends Species> getSpeciesForAliasContaining(String alias)

Specified by:

getSpeciesForAliasContaining in interface Registry

getBacterialSpeciesForAliasBeginning

public Set<? extends CollectionSpecies> getBacterialSpeciesForAliasBeginning(String alias)

Specified by:

getBacterialSpeciesForAliasBeginning in interface Registry

getBacterialSpeciesForAliasContaining

public Set<? extends CollectionSpecies> getBacterialSpeciesForAliasContaining(String alias)

Specified by:

getBacterialSpeciesForAliasContaining in interface Registry

getOldStyleBacterialSpeciesForAliasBeginning

public Set<? extends CollectionSpecies> getOldStyleBacterialSpeciesForAliasBeginning(String alias)

Specified by:

getOldStyleBacterialSpeciesForAliasBeginning in interface Registry

getOldStyleBacterialSpeciesForAliasContaining

public Set<? extends CollectionSpecies> getOldStyleBacterialSpeciesForAliasContaining(String alias)

Specified by:

getOldStyleBacterialSpeciesForAliasContaining in interface Registry

addSessionAlias

public boolean addSessionAlias(String ensembl_genus_species_name,
                      String alias)
                        throws Exception

Method for an adhoc alias to be added by the user in this session. Adds the alias to the aliases TreeSet of the cognate DBSpecies and to the ensemblNameForAliasHash hash of aliases in this Registry

Specified by:

addSessionAlias in interface Registry

Parameters:

ensembl_genus_species_name - Ensmebl databse dbVersion of the name, e.g. 'homo_sapiens'

alias - any string the user wants to use an alias for this

Returns:

boolean true if successful

Throws:

Exception - if update fails, with meaning full message

addSessionAlias

public boolean addSessionAlias(Species sp,
                      String alias)
                        throws Exception

Specified by:

addSessionAlias in interface Registry

Throws:

Exception

getSetOfSpeciesByAlias

public Collection<? extends Species> getSetOfSpeciesByAlias(String alias_or_name)

Returns a HashSet matching the given alias. If name is not a valid String, returns null, if no species are found, returns empty Set. If Name matches any DBSpecies, these are returned without searching for DBCollectionSpecies. Note if we hit a CollectionSpecies - any more recent Species with GCA (ENA) Identifiers will be returned before searching for older Species (lacking GCA IDs) which match the alias. If you want to be sure to search for all alias hits use getSetOfEverySpeciesByAlias(String alias_or_name).

Specified by:

getSetOfSpeciesByAlias in interface Registry

Parameters:

alias_or_name -

Returns:

a Collection of DBSpecies.

getSetOfEverySpeciesByAlias

public Collection<? extends Species> getSetOfEverySpeciesByAlias(String alias_or_name)

Returns a HashSet of all DBSpecies and old and newstyle CollectionSpecies matching the given alias. If name is not a valid String, returns null, if no species are found, returns empty Set.

Specified by:

getSetOfEverySpeciesByAlias in interface Registry

Parameters:

alias_or_name -

Returns:

a Collection of DBSpecies.

getSpeciesByAlias

public DBSpecies getSpeciesByAlias(String alias_or_name)
                            throws NonUniqueException

Wraps getSetOfSpeciesByAlias(String alias_or_name) to return a single species if the returned Collection has only one member. If no species are found, returns null. If more than one species is found, it throws a NonUniqueException which holds the HashSet of results. Note if we hit a CollectionSpecies - more recent Species with GCA (ENA) Identifiers will be returned in preference to older Species lacking IDs.

Specified by:

getSpeciesByAlias in interface Registry

Parameters:

alias_or_name -

Returns:

Throws:

NonUniqueException

getSpeciesByAlias

public DBSpecies getSpeciesByAlias(String alias_or_name,
                          String version)
                            throws NonUniqueException

Wraps getSetOfEverySpeciesByAlias(String alias_or_name) to return a single species if the returned Collection has only one member with the desired alias and version. If no species are found, returns null. If more than one species is found for the datasource version, it throws a NonUniqueException which holds the HashSet of results.

Specified by:

getSpeciesByAlias in interface Registry

Parameters:

alias_or_name -

Returns:

Throws:

NonUniqueException

getMostRecentSpeciesByAlias

public DBSpecies getMostRecentSpeciesByAlias(String alias_or_name)
                                      throws NonUniqueException

Returns the most recent Species matching the alias. Return null if no match, throws NonUniqueException which holds the HashSet of results if more than one, equally recent species is found.

Specified by:

getMostRecentSpeciesByAlias in interface Registry

Parameters:

alias_or_name -

Returns:

Throws:

NonUniqueException

getSpeciesByEnsemblName

public DBSpecies getSpeciesByEnsemblName(String ensemblName)

Specified by:

getSpeciesByEnsemblName in interface Registry

getCSpeciesByGCAccessionStem

public DBCollectionSpecies getCSpeciesByGCAccessionStem(String key)

getMostRecentCollection

public DBCollection getMostRecentCollection(String name)

Specified by:

getMostRecentCollection in interface Registry

getCollection

public DBCollection getCollection(String name,
                         String dbVersion)

Specified by:

getCollection in interface Registry

getCollectionRegistriesByName

public Collection<DBCollection> getCollectionRegistriesByName(String name)

Specified by:

getCollectionRegistriesByName in interface Registry

getCollectionsByDBVersion

public Collection<DBCollection> getCollectionsByDBVersion(String dbVersion)

Specified by:

getCollectionsByDBVersion in interface Registry

getCollections

public List<DBCollection> getCollections()

Specified by:

getCollections in interface Registry

getSpecies

public List<DBSpecies> getSpecies()

Fetches the list of all DBSpecies objects for this Registry.

Specified by:

getSpecies in interface Registry

Returns:

getCollectionSpecies

public List<DBCollectionSpecies> getCollectionSpecies()

Returns the combined list of DBCollectionSpecies that are indexed by dbstylename (old_style_ and those indexed by GCAssemblyAccession (new style, post v16)

Specified by:

getCollectionSpecies in interface Registry

Returns:

getGCAccessionedCollectionSpecies

public List<DBCollectionSpecies> getGCAccessionedCollectionSpecies()

Utility method to return only those DBCOllectionSpecies that possess and are indexed by a GCAssemblyAccession(Stem or Chain)

Returns:

getNamedCollectionSpecies

public List<DBCollectionSpecies> getNamedCollectionSpecies()

Utility method to return only those DBCOllectionSpecies that do not possess and are therefore not indexed by a GCAssemblyAccession but bey dbstylename

Returns:

getDatasourceType

public DBConnection.DataSource getDatasourceType()

Specified by:

getDatasourceType in interface Registry

makeDatabase

public DBDatabase makeDatabase(String db_name)
                        throws ConfigurationException

Specified by:

makeDatabase in interface Registry

Throws:

ConfigurationException

setSpeciesMetadata

public void setSpeciesMetadata(CoreDatabase db)
                        throws DAOException

should only be called for non current singlespecies core databases

Specified by:

setSpeciesMetadata in interface Registry

Parameters:

db -

Throws:

DAOException

getDefaultTranscriptionEngine

public TranscriptionEngine getDefaultTranscriptionEngine()

note by default translation by this default engine is configured to convert non-Methionine initiations to Methionine

Specified by:

getDefaultTranscriptionEngine in interface Registry

getTranscriptionEngine

public TranscriptionEngine getTranscriptionEngine(Integer codonTable)

Specified by:

getTranscriptionEngine in interface Registry

getRenamedDBs

public HashMap<String,String> getRenamedDBs()

Specified by:

getRenamedDBs in interface Registry

isSpeciesInPanHomology

public boolean isSpeciesInPanHomology(Species querySpecies,
                             String dbVersion)

Specified by:

isSpeciesInPanHomology in interface Registry

getPanComparaSpecies

public Set<DBSpecies> getPanComparaSpecies(String dbVersion)