Protein secondary structure: Entropy, correlations and prediction

Abstract

Is protein secondary structure primarily determined by local interactions between residues closely spaced along the amino acid backbone, or by non-local tertiary interactions? To answer this question we have measured the entropy densities of primary structure and secondary structure sequences, and the local inter-sequence mutual information density. We find that the important inter-sequence interactions are short ranged, that correlations between neighboring amino acids are essentially uninformative, and that only 1/4 of the total information needed to determine the secondary structure is available from local inter-sequence correlations. Since the remaining information must come from non-local interactions, this observation supports the view that the majority of most proteins fold via a cooperative process where secondary and tertiary structure form concurrently. To provide a more direct comparison to existing secondary structure prediction methods, we construct a simple hidden Markov model (HMM) of the sequences. This HMM achieves a prediction accuracy comparable to other single sequence secondary structure prediction algorithms, and can extract almost all of the inter-sequence mutual information. This suggests that these algorithms are almost optimal, and that we should not expect a dramatic improvement in prediction accuracy. However, local correlations between secondary and primary structure are probably of under-appreciated importance in many tertiary structure prediction methods, such as threading.