Analyzing simulations of biochemical systems with feature-based visual analytics

• We apply feature-based concepts to drive visual analytics.
• We include analytic, visual, and interactive means to facilitate insight generation.
• We implemented our approach in a multi-display multi-user visualization environment.
• We apply our feature-based visual analytics solution to a real-world simulation problem.

Spatial simulations of biochemical systems are carried out to gain insight into nature's underlying mechanisms. However, such simulations are usually difficult to set up and they generate large and complex data. In order to help scientists understand their models and the data generated by the simulations, appropriate visual support can be a decisive factor. In this paper, we apply and extend ideas of feature-based visualization to develop a visual analytics approach to analyze data of reaction–diffusion system simulations. Our approach enables simulation experts to interactively specify meaningful features, which are automatically extracted and tracked via analytical means. Events in the features’ evolution over time are detected as well. Features and events are visualized via dedicated 3D and 2D views, which in combination portray the interplay of the spatial, temporal, and structural aspects of the simulation data. Our approach is being implemented in the context of a multi-view multi-display visualization environment. We demonstrate how researchers can analyze spatio-temporal distributions of particles in a multi-step activation model with spatial constraints. The visual analytics approach helped to identify interesting behavior of the spatial simulation, which was previously only speculated about, and to examine and discuss competing hypotheses regarding possible reasons for the behavior.

Figure optionsDownload high-quality image (263 K)Download as PowerPoint slide