DAS Plans

From BioDAS
Revision as of 12:15, 15 October 2008 by Andy.jenkinson (talk)
Jump to: navigation, search

Future ideas for improvements to the DAS spec:

Hierarchical features

Features should have parents and children, this could be accomplished with additional tags. This can be seen as either replacement or complementary to groups. For example, groups could be seen as more generic logical and visual groupings (e.g. data from the same sample/tissue). Implementation of a proper hierarchy will enable support for assembly traversal (which isn't used at present and is intended to be removed).


Read/write DAS sources. This is currently being worked on. The specification is based on the DAS/2 writeback portion.

Annotating features

A possible tie-in with writeback, this could allow:

  1. users to add, edit, delete, tag or comment on features
  2. servers to postprocess or highlight other servers' features

Migrate 1.53E extensions

After some minor updates to ensure consistency with the rest of the spec, it is intended that extensions such as alignment be incorporated into the core spec.

Improvements for dense features

Especially for DAS sources with simple per-base annotations, speed is an issue. The maxbins extension goes a long way to helping, but there are other options:

  1. move type etc outside each feature to avoid replication
  2. allow a single feature to contain a set of scores across a region of sequence
  3. alternative content negotiation

Bridging coordinate systems

Alignments provide a way to bridge across 1:1 coordinate systems (e.g. UniProt protein sequence <--> Ensembl protein sequence). We can formalise this, and extend to cover genomic-protein bridging (i.e. 1:3 relationships).