Topology Filter

The sequences of each family with a single GPCR-like 7TM core were aligned, and every residue was replaced by its predicted topology state. Over the columns occupied in at least 30% of the sequences, runs of at least eight columns with a helix as the consensus state define the family's consensus helices. A sequence's topology score is the mean, over those helices, of the Jaccard similarity between the consensus helix columns and the columns where the sequence carries the same helix state, and sequences scoring below 0.8 were removed. Each family is scored in its own alignment, which holds only its own members: the sequences shared with a more specific family were assigned to it beforehand (GPR1 and Vomeronasal 1 are not in the Class A alignment, nor Vomeronasal 2 in the Class C alignment). The table shows what the filter removed from each family; the map below shows the topology alignment of the family selected, with the removed sequences tinted.