Tools for Knowledge Discovery in Clinico-Genomic Data
Knowledge Discovery (KD) is the process of constructing valid, new, interesting, and potentially useful knowledge from data. In this context, the ACGT Knowledge Discovery Tools will provide the services and infrastructure necessary for the scientific exploitation of clinico-genomic data. Examples of KD tools include the R statistical environment, clustering, frequent itemset mining, visual analysis and much more.
The framework for the development and implementation of KD tools in ACGT is defined by the needs of three types of user groups, namely end-users (primarily clinicians), biomedical researchers and data miners. Each user group is defined by a different focus, different levels of bio-medical knowledge and data analysis expertise, and different research interests. The development of the tools follows a scenarios-based approach to guarantee the practical relevance of the developed tools.
In particular, the ACGT KD tools will be designed to cover the following objectives
Extensibility and reusability: the platform will be easily extensible to new tasks and existing solutions should be easily reusable and transferable to similar analysis problems. This goal will be approached by supporting a semantically rich description of services and workflows, including a repository for the storage and retrieval of workflows, a tool for workflow recommendation based on a semantic description of the data, and the support of open standards and interfaces.
Performance: the system must be performant enough to facilitate large analysis and optimization tasks, which calls for an efficient use of the ACGT Grid architecture. Features that will be supported are the efficient distribution of computational expensive tasks and the shipment of algorithms to efficiently process large data sets.
Usability: attention will be given to make the system easy to use for inexperienced users, but also powerful enough for the experienced expert.
