Neutral Networks in Protein Space: A Computational Study Based on Knowledge-Based Potentials of Mean Force

Background: Protein space is explored by means of an inverse folding procedure that makes use of knowledge-based potentials of mean force. Results: Computer simulations indicate that amino acid sequences folding into a common shape are distributed homogenously forming extended percolating networks that span the entire sequence space. Conclusions: The existence of very long neutral paths on all examined protein structures indicates the existence of neutral networks percolating protein space. The same qualitative results were obtained for some, but not all, restricted amino acid alphabets. In this respect, the sequence-structure map of proteins seems to be very similar to the nucleic acid case.