Bio::Chado::Schema

NESCent, Durham, NC

March 5-7, 2011

What is DBIx::Class?

• Object-relational mapping framework for Perl 5
• is now the de-facto standard
• very powerful query builder

What is DBIx::Class?

• query building (the magic of chainable ResultSets)
• cross-database deployment (using SQL::Translator in the backend)
• lots and lots of plugins availabe (dates, testing, ...)

see http://search.cpan.org/dist/DBIx-Class/lib/DBIx/Class/Manual.pod

Data Model

• a set of classes
• encapsulate the underlying storage
• providing a nicer, maintainable API for working with the data

This isn't really middleware, it's a code library.

And therefore, Bio::Chado::Schema

As of 0.08: 205 Perl 5 classes

• every table and view, every module
• Chado docs mirrored in the POD

http://search.cpan.org/dist/Bio-Chado-Schema

And therefore, Bio::Chado::Schema

As of 0.08: 205 Perl 5 classes

• every table and view, every module
• Chado docs mirrored in the POD

http://search.cpan.org/dist/Bio-Chado-Schema

( So why do we need this? )

Chado needs encapsulation

Chado's design makes for:

• complex queries
• steep learning curve

Chado needs encapsulation

complex queries

• generating
• storing
• automating

Chado needs encapsulation

steep learning curve

• codifying best practices
• unified, high-level documentation
• can help with database administration

BCS Usage

• open the schema. ( actual DB connection is deferred )

BCS Usage

• open the schema. ( actual DB connection is deferred )

my $chado = Bio::Chado::Schema->connect( 'dbi:Pg:...', $user, $pass );

BCS Usage

• open the schema. ( actual DB connection is deferred )

my $chado = Bio::Chado::Schema->connect( 'dbi:Pg:...', $user, $pass );

• get a ResultSet object representing a set of Rows

BCS Usage

• open the schema. ( actual DB connection is deferred )

my $chado = Bio::Chado::Schema->connect( 'dbi:Pg:...', $user, $pass );

• get a ResultSet object representing a set of Rows

$chado->resultset('Sequence::Feature');

BCS Usage

• open the schema. ( actual DB connection is deferred )

my $chado = Bio::Chado::Schema->connect( 'dbi:Pg:...', $user, $pass );

• get a ResultSet object representing a set of Rows

$chado->resultset('Sequence::Feature');
      ->search({ name => 'something' });

BCS Usage

• open the schema. ( actual DB connection is deferred )

my $chado = Bio::Chado::Schema->connect( 'dbi:Pg:...', $user, $pass );

• get a ResultSet object representing a set of Rows

$chado->resultset('Sequence::Feature');
      ->search({ name => 'something' });
      ->search_related('organism')

BCS Usage

• open the schema. ( actual DB connection is deferred )

my $chado = Bio::Chado::Schema->connect( 'dbi:Pg:...', $user, $pass );

• get a ResultSet object representing a set of Rows

$chado->resultset('Sequence::Feature');
      ->search({ name => 'something' });
      ->search_related('organism')

• get an actual data object

BCS Usage

• open the schema. ( actual DB connection is deferred )

my $chado = Bio::Chado::Schema->connect( 'dbi:Pg:...', $user, $pass );

• get a ResultSet object representing a set of Rows

$chado->resultset('Sequence::Feature');
      ->search({ name => 'something' });
      ->search_related('organism')

• get an actual data object

$chado->resultset('Sequence::Feature')
      ->find( 232432 );

BCS Usage

BCS Usage

• iterate over the actual data in a set

BCS Usage

• iterate over the actual data in a set

my $stuff = $chado->blahblahblah;
while( my $row = $stuff->next ) {
  # do something
}

BCS Usage

• iterate over the actual data in a set

my $stuff = $chado->blahblahblah;
while( my $row = $stuff->next ) {
  # do something
}

• traverse relationships

BCS Usage

• iterate over the actual data in a set

my $stuff = $chado->blahblahblah;
while( my $row = $stuff->next ) {
  # do something
}

• traverse relationships

say $some_feature->type->name;

BCS Usage: Joined Select

# get features via the potato organism, also joining in the cvterms table

BCS Usage: Joined Select

# get features via the potato organism, also joining in the cvterms table
my $potato_bacs =
     $chado->resultset('Organism::Organism')

BCS Usage: Joined Select

# get features via the potato organism, also joining in the cvterms table
my $potato_bacs =
     $chado->resultset('Organism::Organism')
           ->search({ species => 'Solanum tuberosum' })

BCS Usage: Joined Select

# get features via the potato organism, also joining in the cvterms table
my $potato_bacs =
     $chado->resultset('Organism::Organism')
           ->search({ species => 'Solanum tuberosum' })
           ->search_related( 'features',
                             { 'type.name' => 'BAC_clone'},
                             { 'join' => 'type' },
                           );

BCS Usage: Joined Select

# the equivalent bare SQL
my $potato_bacs = $dbh->selectall_arrayref( <<EOS, undef, 'Solanum tuberosum', 'BAC_clone');
SELECT features.feature_id
     , features.dbxref_id
     , features.organism_id
     , features.name
     , features.uniquename
     , features.residues
     , features.seqlen
     , features.md5checksum
     , features.type_id
     , features.is_analysis
     , features.is_obsolete
     , features.timeaccessioned
     , features.timelastmodified
FROM organism me
LEFT JOIN feature features
       ON features.organism_id = me.organism_id
JOIN cvterm type
       ON type.cvterm_id = features.type_id
WHERE type.name = 'BAC_clone' AND species = 'Solanum tuberosum'
EOS

BCS Usage: Loading

$chado->resultset( 'Cv::Cv' )
      ->find_or_create({ name => 'My Fake Ontology' })
      ->create_related(  'cvterm',
                         { name => 'MyFakeTerm' });

makes the SQL:

SELECT me.cv_id
     , me.name
     , me.definition
FROM cv me
WHERE ( me.name = 'my fake ontology' )

INSERT INTO cv ( name )
        VALUES ( ?    )

BCS Usage: Transactions

$chado->txn_do(sub {

    $chado->resultset('Cv::Cv')
          ->find_or_create({ name => 'My Fake Ontology' })
          ->create_related( 'cvterm', { name => 'MyFakeTerm' } );

    $chado->do_all_kinds_of_other_stuff;
});

The Real Advantages of DBIC

• easier to manipulate and assemble queries
• Don't Repeat Yourself

The Real Advantages of DBIC

• it's all objects. you can delegate to them, pass them around, etc.
• HOWEVER:

The Real Advantages of DBIC

• it's all objects. you can delegate to them, pass them around, etc.
• HOWEVER:
• usually you don't want to subclass them
• but, see DBIx::Class::Manual::Cookbook

The Real Advantages of DBIC

• complex joined queries (Chado queries) are very easy and compact

The Real Advantages of DBIC

• SQL syntax errors are much more difficult to make

Using DBIC with your own tables

• use DBIx::Class::Schema::Loader to dump a whole set
• make your own definitions

• your table: other_thing, foreign key feature_id to Chado feature table

package My::DBIC:::Layer::OtherThing;
use base 'DBIx::Class::Core';

__PACKAGE__->table('other_thing');
__PACKAGE__->add_columns(
  'other_thing_id' => { ... },
  'name'           => { ... },
  'definition'     => { ... },
  'feature_id'     => { ... },
);
__PACKAGE__->set_primary_key('other_thing_id');
__PACKAGE__->add_unique_constraint('ot_c1', ['name']);

__PACKAGE__->belongs_to(
  'feature',
  'Bio::Chado::Schema::Sequence::Feature',
  { 'foreign.feature_id' => 'self.feature_id' },
);

"Duct tape" your own schema onto BCS

• make an accessor 'other_things' that ties your own DBIC class to BCS

"Duct tape" your own schema onto BCS

• make an accessor 'other_things' that ties your own DBIC class to BCS

Bio::Chado::Schema::Sequence::Feature->has_many(
  'other_things',
  'My::DBIC::Layer::OtherThing',
  { 'foreign.feature_id' => 'self.feature_id' },
);

"Duct tape" your own schema onto BCS

• make an accessor 'other_things' that ties your own DBIC class to BCS

Bio::Chado::Schema::Sequence::Feature->has_many(
  'other_things',
  'My::DBIC::Layer::OtherThing',
  { 'foreign.feature_id' => 'self.feature_id' },
);

• add it to the BCS schema dynamically

"Duct tape" your own schema onto BCS

• make an accessor 'other_things' that ties your own DBIC class to BCS

Bio::Chado::Schema::Sequence::Feature->has_many(
  'other_things',
  'My::DBIC::Layer::OtherThing',
  { 'foreign.feature_id' => 'self.feature_id' },
);

• add it to the BCS schema dynamically

Bio::Chado::Schema->register_source('OtherThing', 'My::DBIC::Layer::OtherThing');

"Duct tape" your own schema onto BCS

• make an accessor 'other_things' that ties your own DBIC class to BCS

Bio::Chado::Schema::Sequence::Feature->has_many(
  'other_things',
  'My::DBIC::Layer::OtherThing',
  { 'foreign.feature_id' => 'self.feature_id' },
);

• add it to the BCS schema dynamically

Bio::Chado::Schema->register_source('OtherThing', 'My::DBIC::Layer::OtherThing');

• use it with the rest

"Duct tape" your own schema onto BCS

• make an accessor 'other_things' that ties your own DBIC class to BCS

Bio::Chado::Schema::Sequence::Feature->has_many(
  'other_things',
  'My::DBIC::Layer::OtherThing',
  { 'foreign.feature_id' => 'self.feature_id' },
);

• add it to the BCS schema dynamically

Bio::Chado::Schema->register_source('OtherThing', 'My::DBIC::Layer::OtherThing');

• use it with the rest

$chado->resultset('Sequence::Feature')->other_things;

Making a composite schema

package My::Schema;
# load our own classes
__PACKAGE__->load_namespaces;
__PACKAGE__->load_namespaces(
    result_namespace    => '+Bio::Chado::Schema::Result',
    resultset_namespace => '+Bio::Chado::Schema::ResultSet',
  );

Development Roadmap

• continue to fill in many_to_many rels as needed
• continue to add more (generic!) convenience methods
• automate more common querying and loading patterns

Acknowledgments

• Jonathan Leto (SGN)
• Aure Bombarely (SGN)
• Naama Menda (SGN)
• Siddhartha Basu (dictybase)
• Lukas Mueller (SGN)