Computing centrality involves finding the most 'central' or important nodes in a network. Although potentially useful for biological networks, this can be challenging if the definition of importance is not obvious [1]. There are many different centrality algorithms with different importance definitions that return different results. This is immediately obvious in Figure 1(a), which shows the results of betweenness (red, [2]), closeness (yellow, [3]) and degree (blue, [4]) centrality on a bacterial co-occurence network [5]. Black nodes indicate mutual agreements. We color the top 20% of nodes found by each algorithm, and use appropriate color combinations for those found by two (i.e., red+yellow=orange for betweenness and closeness). As shown, due to spatial bias there is a wide variation making these results difficult to interpret or generalize. Betweenness tends to find nodes on the same path, closeness toward the middle of the network, and degree within the same strongly connected component.
