Fig. 3From: Algorithms for matching partially labelled sequence graphsTop ranked contacts for 1aoz. a Raw data with the consensus plotted bold: green = distance-based, purple = topology-based. b Consensus plots filtered by residue exposure. Bold = 30% weight: green = distance-based, purple = topology-based. In both, the cumulative contact score error is plotted (Y-axis) against the rank of the pair (X-axis). The lower two panels constitute a stereo-pair of the structure with the domains coloured sequentially, purple, green, orange and the top predicted contacts linked by blue lines. The number of contacts to view was taken at the point where the purple line in (b) remained below a CCSE of 2Back to article page