A visual analytics approach to feature discovery and subspace exploration in protein flexibility matrices

Jing Yang,S. Barlowe

2011-01-01

Abstract:The vast amount of information generated by domain scientists makes the transition from data to knowledge difficult and often impedes important discoveries. For example, the knowledge gained from protein flexibility data sets can speed advances in genetic therapies and drug discovery. However, these models generate so much data that large scale analysis by traditional methods is almost impossible. This hinders biomedical advances. Visual analytics is a new field that can help alleviate this problem. Visual analytics attempts to seamlessly integrate human abilities in pattern recognition, domain knowledge, and synthesis with automatic analysis techniques. I propose a novel, visual analytics pipeline and prototype which eases discovery, comparison, and exploration in the outputs of complex computational biology datasets. The approach utilizes automatic feature extraction by image segmentation to locate regions of interest in the data, visually presents the features to users in an intuitive way, and provides rich interactions for multi-resolution visual exploration. Functionality is also provided for subspace exploration based on automatic similarity calculation and comparative visualizations. The effectiveness of feature discovery and subspace exploration is shown through a user study and user scenarios. Feedback from analysts confirms the suitability of the proposed solution to domain tasks.

Computer Science