Abstract
Network science empowers researcher across different disciplines to study complex relationships between a huge number of elements and reveal the underling wiring diagram of a system. Therefore, network visualization tools have become a very important component in various data analysis workflows (Barabási, 2011). Gephi is an open source software for explorative analysis and visualization of large-scale networks. It facilitates real-time visualizations and analysis to improve the exploratory process by high quality layout algorithms and intuitive user interface (Bastian, Heymann, & Jacomy, 2009). Due to universality of network science, their representation emerges of different data sources from nature or society. Therefore, the need to efficiently connect to rich information spaces and create complex repeatable workflows exceeds the capabilities of automation features in any point-and-click-based network visualization and analysis software available. The functional-first programming languages F# is specialized for type save information-rich programming and domain driven explorative programming (Don Syme, 2006) (Donald Syme, Battocchi, Takeda, Malayeri, & Petricek, 2013). This enables data collection, processing and analysis tasks leveraging the flexibility of the programming environments.
We created FSharpGephiStreamer to enable automated access to the Gephi network and visualization models directly from within the F# interactive programming environment. The functional approach used for the implementation of FSharpGephiStreamer provides a short grammar that models the direct graph data streaming process to transfer the data over the Gephi Rest API. This allows an immediate insert, update or delete of individual graph elements directly in the Gephi network and visualization model without the necessity to create the network on the client side. The client-host communication works exception-free with error handling using error type monads. FSharpGephiStreamer closes the gap between F# and the functionality of the Gephi software project. Hence, effectively improving the integration of Gephi into large scale data analysis pipelines and increasing workflow reproducibility
Kevin Schneider
Doctoral student, Computational Systems Biology
My research interests include large-scale meta-analysis of biological datasets, as well as applications from network science and machine learning on the aforementioned.