Although this method eliminated approximately 65% of the protein clusters' edge, no reduction in the number of known protein complexes and functional scores was observed.
The example of inter-cluster and intra-cluster edges of (C_i), the dashed lines are the inter-cluster edges and the solid lines in cluster (C_i) are the intra-cluster edges.
The number assigned to an edge shows the number of the cross-cluster edges.
Through multi-partite graphs, nodes can be distributed among various clusters with inter-cluster edges only.
Such slight changes ensure a negligible influence of the cluster edges on the calculated adsorption geometries.
Our technique is based on modularity clustering so as to decrease the number of cross-cluster edges.
Having static clusters, however, is sub-optimal as fixed clusters create cluster edges, instead of cell edges, in which the degradation could in fact become more pronounced.
Although this idea might lose some true inter-cluster edges, we may be able to control such lost to an acceptable level with a carefully designed clustering strategy.
Brandes et al. [44] proposed the coverage of the graph clustering based on the number of intra-cluster edges and inter-cluster edges as shown in Eq. 3, where (m(C_i, C_i)) and ({m}(C_i, bar{C_i})) are the numbers of intra-edges and inter-edges in the cluster (C_i) respectively.
In contrast, there are other graph clustering techniques [7, 8, 9] that are designed to reduce the number of cross-cluster edges.
