Cytoscape cyberinfrastructure – quality network analysis done quicker and cheaper

Summary The Cytoscape Cyberinfrastructure (CI) combines Cytoscape and a microservices architecture to empower network biologists to create and share novel and reproducible web-based workflows quickly and cheaply. Using highly reusable, evolvable, and scalable computational and visualization services and user interface elements, it extends Cytoscape and its successful development model to Internet scale. Abstract Cytoscape’s primary mission is to support the analysis and visualization of networks that describe biological processes and structure. As networks have become more useful and relevant over the last decade, Cytoscape usage has mushroomed – 3,000 startups per day and 14,000 downloads per month worldwide. Predictably, as network analysis has become more common, we have seen more demand for novel and reproducible network-based workflows, network sharing and collaboration, and analytics and capacity to manipulate ever larger, more complex, and more hyperlinked and multiscale networks. As a standalone workstation-based desktop application, Cytoscape is limited in its ability to address these opportunities due to limited memory and CPU resources (e.g., processing Network-based GWAS and Stratification), and is often difficult and time consuming to customize. In the past year, we have answered this challenge by developing the Cytoscape Cyberinfrastructure (CI), an Internet-scale distributed system based on Microservices [1]. The CI’s mission is to enable and encourage network biologists to create and deploy high quality, innovative and scalable network-based computation, collaboration and visualization services using the tool chain of their choice. The CI links services via a light weight REST-based aspect-oriented interchange protocol (called CX), which enables tailored data streams while supporting service innovation via evolvable standards. The CI service broker (“Elsa”) enables biologists to write and publish biologically relevant services (e.g., enrichment calculations or heat propagation) comparatively easily and without concern for scaling or tracking job status. It provides the CyREST [2] interface layer to expose Cytoscape’s rich functionality as services, and it provides a framework ("CytoWidgets") to enable the creation, delivery, reuse, and evolution of Javascript-based user interface modules that package service functionality within both web-based and desktop-based applications. We demonstrate the CI in action by showing how applications (written in Java, IPython, R, and Javascript) call useful production services (written in the Java, Go, and Python languages) to execute real world workflows. Particularly, we preview the NDEx Valet CytoWidget embedded within Cytoscape to provide a state of the art chooser and importer for shared NDEx-based networks [3]. We also show how to quickly and cheaply new and novel web-based, network-oriented workflows can be built using NDEx Valet and network display and manipulation widgets. Finally, we demonstrate how the CI brokers long-running, CPU and memory intensive services. Future work includes adding clustering, analysis, layout, publishing and display microservices, interfaces to Galaxy and Taverna workflows, and support for the Publishing of the Future initiative. References [1] http://martinfowler.com/articles/microservices.html [2] K. Ono et al. CyREST: Turbocharging Cytoscape Access for External Tools via a RESTful API. F1000 Research. 2015 Aug 5;4:478. [3] D. Pratt et al. NDEx, the Network Data Exchange. Cell Systems 1, no. 4 (n.d.): 302–5