Figure 8
From: Addressing the unmet need for visualizing conditional random fields in biological data
Birds-eye views of the dependency network in ADK while being explored by a user to identify a subset suitable for building a CRF. Views from StickWRLD being used to refine a GPM to identify the most critical determinants of catalytic activity in the Adenylate Kinase lid domain. A) shows an overhead view of the 300-column, 21-category dependency network after it has already been filtered down to relatively large residuals. B), C) and D) show successive refinements using a statistical threshold cutoff. By the time p=0.001 has been reached in D), the majority of the interesting interactions have been lost.