Fig. 2From: A proximity-based graph clustering method for the identification and application of transcription factor clustersThe construction of co-occurrence matrices. For each TFBS of a given TF, the other TFs within a 1000 base-pair window are recorded in that TF’s co-occurrence matrix. The amount of co-occurrence is inversely proportional to the sparsity of the rows; in the above example, TF B is most highly associated with TF D and least associated with TF ABack to article page