Abstract
Networks provide a powerful representation of interacting components within complex systems, making them ideal for visually and analytically exploring big data. However, the size and complexity of many networks render static visualizations on typically-sized paper or screens impractical, resulting in proverbial ‘hairballs’. Here, we introduce a Virtual Reality (VR) platform that overcomes these limitations by facilitating the thorough visual, and interactive, exploration of large networks. Our platform allows maximal customization and extendibility, through the import of custom code for data analysis, integration of external databases, and design of arbitrary user interface elements, among other features. As a proof of concept, we show how our platform can be used to interactively explore genome-scale molecular networks to identify genes associated with rare diseases and understand how they might contribute to disease development. Our platform represents a general purpose, VR-based data exploration platform for large and diverse data types by providing an interface that facilitates the interaction between human intuition and state-of-the-art analysis methods.
Highlights
Networks provide a powerful representation of interacting components within complex systems, making them ideal for visually and analytically exploring big data
The Virtual Reality (VR) module is connected to the data analytics engine via a (4) user interface (UI) module that serves as a communication layer and offers the functionality of implementing arbitrary UI elements using standard web design libraries
We provide a (5) web module as a browser-based frontend designed for tasks that can be performed more conveniently on a conventional computer screen, such as data preprocessing or further inspection of the results of a VR exploration session
Summary
Networks provide a powerful representation of interacting components within complex systems, making them ideal for visually and analytically exploring big data. Networks have a distinct advantage compared to other computational methodologies for integrating and interpreting biological data: Their visual representation allows for a uniquely intuitive interpretation, enabling us to quickly identify potential local and global patterns in complex data that can be further assessed by advanced computational and statistical means (Fig. 1a): In molecular interaction networks, for example, highly connected hubs generally correspond to genes that play important roles in healthy and disease states, such as pleiotropic genes[5] or cancer driver genes[6]. The inherent multi-scale nature of biological processes can only be fully appreciated when the entire range from local to global network structures can be inspected continuously and interactively Until now, this is only possible for very limited network sizes of up to a few hundred nodes. As a proof of concept, we show how this exploration can be leveraged to investigate gene variants in the context of a molecular interaction network for identifying variants responsible for severe genetic diseases
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
