|
|
Challenges
In order to set up a fully-automated distributed transparent approach for structure-based proteome annotation, several key technical problems had to be overcome:
- No single UK group has the facilities and expertise to manage the required annotation for the proteomes of dozens of major species. However, a mixture of areas of expertise across several groups will allow this.
- Proteome annotation methodology is not well established and thus attempting several different strategies with comparison of the results helps identify and solve annotation problems.
- Structural and functional annotation methods use different algorithms for prediction. Comparison of results from different approaches is necessary to assess the reliability of these predictions.
- New methodologies are constantly being developed and need to be evaluated and incorporated if necessary. DAS allows the new annotations to be easily integrated with, and compared to, those from existing methods.
The databases require regular, and frequent, updating to remain useful. Sharing of computing resources is essential to reliably support the computational load produced by certain steps in the annotation pipeline,
such as homology searching using PSI-BLAST. Therefore, e-science has its own set of challenges that will be addressed:
- The development of the best strategy for providing high performance computing from distributed resources.
- To provide end-users (i.e. biologists) with a powerful interface to multi-site.
|
