Ibidas: Querying Flexible Data Structures to Explore Heterogeneous Bioinformatics Data

Abstract

Nowadays, bioinformatics requires the handling of large and diverse datasets. Analyzing this data demands often significant custom scripting, as reuse of code is limited due to differences in input/output formats between both data sources and algorithms. This recurring need to write data-handling code significantly hinders fast data exploration.

We argue that this problem cannot be solved by just data integration and standardization alone. We propose that the integration-analysis chain misses a link: a query solution which can operate on diversely structured data throughout the whole bioinformatics workflow, rather than just on data available in the data sources. We describe how a simple concept (shared ’dimensions’) allows such a query language to be constructed, enabling it to handle flat, nested and multi-dimensional data. Due to this, one can operate in a unified way on the outputs of algorithms and the contents of files and databases, directly structuring the data in a format suitable for further analysis. These ideas have been implemented in a prototype system called Ibidas. To retain flexibility, it is directly integrated into a scripting language. We show how this framework enables the reuse of common data operations in different problem settings, and for different data interfaces, thereby speeding up data exploration.