Kvik : interactive exploration of genomic data from the NOWAC postgenome biobank

Abstract

Recent technological advances provide large amounts of data for epidemiological analyses that can provide novel insights in the dynamics of carcinogenesis. These analyses are often performed without prior hypothesis and therefore require an exploratory approach. Realizing exploratory analysis requires the development of new systems that provide interactive exploration and visualization of large-scale scientific datasets.

This thesis presents Kvik, an interactive system for exploring the dynamics of carcinogenesis through integrated studies of biological pathways and genomic data. Kvik is designed as a three-tiered application, an architecture that is commonly used for peta-scale applications. It provides researchers with a lightweight web application for navigating through biological pathways from the KEGG database integrated with genomic data from the NOWAC postgenome biobank.

In collaboration with researchers from the NOWAC systems epidemiology group, we have described the requirements for such a system, and by using an iterative approach we implemented Kvik through small development cycles, involving the end-users in the development process. Throughout the project we have gained valuable interdisciplinary experience in developing systems for use in explorative analysis of carcinogenesis.

Through an evaluation of the exploration tasks and workflow of an end-user, we demonstrate that Kvik has the capability of interactive exploration of genomic data and biological pathways.

We believe Kvik is important to enable novel discoveries from the data produced in the NOWAC systems epidemiology project. It provides epidemiology researchers with access to powerful compute and storage resources enabling the use of advanced statistical methods for the analysis. Finally, from our experiences in developing Kvik, we provide use cases and requirements for future analysis, computation and storage systems developed in our research group and by others.