On the mutual information between interaction perplexity and function of proteins

We are exploring a novel information theory based approach to analyse the relation of the number of protein interactions (interaction perplexity) and protein function. The interaction perplexity is translated into a smooth sigmoid function, indicating if the protein is a hub candidate or not (hub likelihood). The overall mutual information between the hub likelihood and the protein function exhibits a clear maximum at an interaction perplexity of 21. Moreover the conditional probability of observing a certain protein function given the hub likelihood shows discriminative power. In addition our findings support that the mutual information between hub likelihood and protein function depends upon the cell compartment under investigation, it´s maximum ranging up to 29% of the achievable information value at interaction perplexities, ranging from 15 to 50.