The Distributed Annotation System

1
The Distributed Annotation System

• Dowell, R.D., Jokerst, R.M., Day, A., Eddy, S.R.,
  and Stein, L. (2001)

2
DAS

• Distributed Annotation System (DAS) is an effort
  to share annotation amongst multiple groups.
• An XML specification is used to enforce
  compatibility.
• Byproduct of NIH NHGRI C. elegans genome project

3
Ideal Annotation System

• Should give individual experts the ability to
  contribute the collective annotation.
• They should have complete control over their
  annotations.
• These annotations should not need approval from a
  central authority.

4
Problems with Centralization

• Centralized repositories such as GenBank act
  primarily as storage databases for sequence.
• Annotation in these systems is made difficult and
  in some cases nearly impossible.

5
Curation

• Databases have emerged to act as curators for
  particular communities
• Swissprot
• RefSeq
• WormPD
• ACeDB

6
Curation (continued)

• Individuals submit annotation to a central
  curation group.
• Approved annotations are included in official
  releases.
• Limited curators cause bottlenecks.

7
The DAS Solution

• DAS allows experts to post annotations.
• Rather than presenting to a centralized group for
  consideration, the annotation is added to one of
  many third party servers.
• The client is then responsible for accessing the
  server and annotation.

8
Implementation

• 1 Reference Server
• DAS relies on a common reference sequence
• 1 or more Annotation Servers
• Responsible for list of annotations across a
  defined region of the genome.
• 1 Client
• Allows the user to request annotations from
  specific servers across a specific reference
  sequence.

9
Reference Server

• Responsible for supplying the reference map and
  the reference sequence.
• Reference Map
• These may be entire chromosomes, a collection of
  contigs, or even individual clones.
• The entry point is the top level structure.
• Substructure is defined by start and stop points
  within the entry point.
• Substructure is recursive.
• Reference Sequence
• The underlying DNA.

10
Annotation Server

• Responsible for supplying annotations for a
  specified region of the reference sequence.
• Annotations are related to regions of the genome
  by relative start and stop points.

11
Annotations

• The details of an annotation are kept very non
  specific using the General Feature Format (GFF)
• Annotators are encouraged to provide links to a
  more applicable description.
• An annotation can be associated with a glyph
  which describes how it should be displayed to the
  user.

12
Annotations (continued)

• DAS does not impose a controlled vocabulary for
  annotation descriptions.
• An annotation does have three attributes
• Type - biological significance
• Ex tRNA, snoRNA, miscRNA
• Method - how annotation was discovered
• Ex any specific software or typing methods
• Category - functional category
• Ex Homology, variation, transcribed

13
Client

• The client is responsible for allowing the user
  to examine a particular part of the genome for
  annotations located on multiple third party
  servers.
• Only the relative position is considered, so no
  semantic contradictions are considered.
• The current tools are optimal for visualizing and
  indexing annotations.

14
Why XML?

• Strongly typed, extensible data exchange format.
• At the cost of non-trivial bandwidth demands.
• A user can easily request more data than their
  connection can reasonably handle.
• This is why very little information on an
  annotation is returned.
• Data compression may resolve in the future.

15
The Future of DAS

• As of 2001, this papers publication, Ensembls
  database of human genome annotations supports
  DAS.
• As of 2006, DAS servers are running at WormBase,
  FlyBase, Ensembl, TIGR, and UCSC.
• Biodas.org