Table Of Contents

Previous topic

VCF files

Next topic

Ensembl Core

This Page

Chado

We have developed an InterMine data source that can use a GMOD Chado database as a source for an InterMine warehouse. The eventual aim is to allow import of any Chado database with some configuration. This will provide a web environment to perform rapid, complex queries on Chado databases with minimal development effort.

Converter

The converter for this source is the ChadoDBConverter class. This class controls which ChadoProcessors are run. A ChadoProcessor class corresponds to a chado module. For example, the sequence module is processed by the SequenceProcessor and the stock module is processed by the StockProcessor.

Chado tables

The chado-db source is able to integrate objects from a Chado database. Currently only tables from the Chado sequence module and Chado stock modules are read.

These tables are queried from the chado database:

feature
used to create objects in the ObjectStore
  • The default configuration only supports features that have a Sequence Ontology type (eg. gene, exon, chromosome)
  • Each new feature in InterMine will be a sub-class of SequenceFeature.
featureloc
used to create Location objects to set chromosomeLocation reference in each SequenceFeature
feature_relationship
used to find part_of relationships between features
  • this information is used to create parent-child references and collections
  • examples include setting the transcripts collection in the Exon objects and the gene reference in the Transcript class.
dbxref and feature_dbxref
used to create Synonym objects for external identifiers of features
  • the Synonym`s will be added to the `synonyms collection of the relevant SequenceFeature
featureprop
used to set fields in features based on properties
  • an example from the FlyBase database: the SequenceFeature.cytoLocation field is set using the cyto_range feature_prop
synonym and feature_synonym
used to create extra Synonym objects for chado synonyms and to set fields in features
  • the Synonym`s will be added to the `synonyms collection of the relevant SequenceFeature
cvterm and feature_cvterm
used to set fields in features and to create synonyms based on CV terms
pub, feature_pub and db
used to set the publications collection in the new SequenceFeature objects.

Additionally, the StockProcessor class reads the tables from the chado stock module, eg. stockcollection, stock, stock_genotype.

Default configuration

The default configuration of ChadoDBConverter is to query the feature table to only a limited list of types. The list can be changed by sub-classing the ChadoDBConverter class and overriding the getFeatureList() method. The featureloc, feature_relationship and pub tables will then be queried to create locations, parent-child relationships and publications (respectively).

Converter configuration

Sub-classes can control how the Chado tables are used by overriding the getConfig() method and returning a configuration map.

Source configuration

Example source configuration for reading from the ‘’C.elegans’’ Chado database:

<source name="chado-db-wormbase-c_elegans" type="chado-db" dump="true">
  <property name="source.db.name" value="wormbase"/>
  <property name="genus" value="Caenorhabditis"/>
  <property name="species" value="elegans"/>
  <property name="taxonId" value="6239"/>
  <property name="dataSourceName" value="WormBase"/>
  <property name="dataSetTitle" value="WormBase C.elegans data set"/>
</source>

Sub-classing the converter

The processor classes can be sub-classed to allow organism or database specific configuration. To do that, create your class (perhaps in bio/sources/chado-db/main/src/) set the processors property in your source element. For example for reading the FlyBase Chado database there is a FlyBaseProcessor which can be configured like this:

<source name="chado-db-flybase-dmel" type="chado-db">
...
        <property name="processors" value="org.intermine.bio.dataconversion.FlyBaseProcessor"/>
...

Current uses

FlyMine uses the chado-db source for reading the ‘’Drosophila’’ genomes from the FlyBase chado database. The FlyBaseProcessor sub-class is used for configuration and to handle FlyBase special cases.

modMine for the modENCODE project uses ChadoDBSource for reading ‘’D. melanogaster’’ from FlyBase and to read ‘’C. elegans’’ data from the WormBase chado database. The WormBaseProcessor sub-class is used for configuration when reading from WormBase.

Implementation notes for the chado-db source

The chado-db source is implemented by the ChadoDBConverter class which runs the ChadoProcessor`s that have been configured in the `project.xml. The configuration looks like this:

<source name="chado-db-some-database" type="chado-db">
  ...
  <property name="processors" value="org.intermine.bio.dataconversion.ChadoSequenceProcessor org.intermine.bio.dataconversion.StockProcessor"/>
  ...

ChadoDBConverter.process() will create an object for each ChadoProcessor in turn, then call ChadoProcessor.process().

Chado sequence module table processing

ChadoSequenceProcessor processes the sequence module from Chado. The process() method handles each table in turn by calling: processFeatureTable(), processFeatureCVTermTable() etc.

Each table processing method calls a result set method, eg. processFeatureTable() calls getFeatureTableResultSet() and then processes each row. The returned ResultSet may not always include all rows from the Chado table. For example the getFeatures() method returns a sub-set of the possible feature types and that list is used to when querying the feature table.

Generally each row is made into an appropriate object, eg. in processFeatureTable(), feature table rows correspond to BioEntity objects. Some rows of some tables are ignored (ie. not turned into objects) based on configuration.

Reading the feature table

Handled by ChadoSequenceProcessor.processFeatureTable()

For each feature it calls: ChadoSequenceProcessor.makeFeatureData(), which may be overridden by subclasses. If makeFeatureData() returns null (eg. because the sub-class does not need that feature) the row is discarded, otherwise processing of the feature continues.

Based on the configuration, fields in the BioEntity are set using feature.uniquename and feature.name from Chado.

If the residues column in the feature is set, create a Sequence object and add it to the new BioEntity.

Reading the featureloc table

Handled by ChadoSequenceProcessor.processLocationTable().

This method gets passed a result set with start position, end position and information from the featureloc table. For each row from the result set it will:

  • store a Location object
  • set chromosomeLocation in the associated SequenceFeature
  • set the chromosome reference in the SequenceFeature if the srcfeature from the featureloc table is a chromosome feature

Reading the feature_relationship table

Handled by ChadoSequenceProcessor.processRelationTable().

This method calls getFeatureRelationshipResultSet() to return the relations of interest. The relations will be used to create references and collections.

The method will automatically attempt to find and set the appropriate references and collections for part_of relations. As an example, if there is a part_of relation in the table between Gene and Transcript and there Gene.transcript reference or a Gene.transcripts collection, it will be set.

There are two modes of operation, controlled by the subjectFirst parameters. If true, order by the subject_id of the feature_relationship table so we get results like:

gene1_feature_id relation_type protein1_feature_id
gene1_feature_id relation_type protein2_feature_id
gene2_feature_id relation_type protein1_feature_id
gene2_feature_id relation_type protein2_feature_id

(Assuming the unlikely case where two genes are related to two proteins)

If subjectFirst is false we get results like:

gene1_feature_id relation_type protein1_feature_id
gene2_feature_id relation_type protein1_feature_id
gene1_feature_id relation_type protein2_feature_id
gene2_feature_id relation_type protein2_feature_id

The first case is used when we need to set a collection in the gene, the second if we need to set a collection in proteins.

Reading the cvterm table

Handled by ChadoSequenceProcessor.processFeatureCVTermTable()

Using the default chado source

  1. Add the chado database to your MINE_NAME.properties file, eg:
db.flybase.datasource.class=org.postgresql.ds.PGPoolingDataSource
db.flybase.datasource.dataSourceName=db.flybase
db.flybase.datasource.serverName=SERVER_NAME
db.flybase.datasource.databaseName=DATABASE_NAME
db.flybase.datasource.user=USER_NAME
db.flybase.datasource.password=SECRET_PASSWORD
db.flybase.datasource.maxConnections=10
db.flybase.driver=org.postgresql.Driver
db.flybase.platform=PostgreSQL

The chado database has to be on the local network.

  1. Add source to project XML file
<source name="chado-db" type="chado-db">
  <property name="source.db.name" value="flybase"/>
  <property name="organisms" value="7227"/>
  <property name="dataSourceName" value="FlyBase"/>
  <property name="converter.class" value="org.intermine.bio.dataconversion.ChadoDBConverter"/>
  <property name="processors" value="org.intermine.bio.dataconversion.SequenceProcessor"/>
</source>
  1. Run the build
$ cd MINE_NAME/integrate
$ (cd ../dbmodel && ant build-db -v); ant -Dsource=chado-db -v

See Database Building for more information on running builds.

This will load the data using the default chado loader. If you want to load more data you will have to write a custom chado converter. FlyMine uses a FlyBase chado “processor” to parse interactions, etc. See FlyBaseProcessor.java for an example.

Tripal

The Chado specific tables are not in the postgres default “public” schema of the database. Instead, Tripal puts it in a postgres schema named “chado”.

To workaround this, you would need to alter your Chado processor to run this query first, before running any SELECT statements:

ALTER DATABASE <dbname> SET search_path TO chado, public