Neural network‐based prediction of mutation‐induced protein stability changes in Staphylococcal nuclease at 20 residue positions

Abstract

Protein-based therapeutics are playing an increasingly important role in the treatment of diseases, including diabetes and cancer. The viability of these treatments, however, are highly dependent on the stability of the therapeutic, since stability affects both the shelf life of the therapeutic as well as its active life in the body. Stability engineering can, therefore, be used to increase the effectiveness of protein-based therapeutics. Computational methods of protein stability prediction have been under development for about a decade, but complex molecular interactions make stability prediction difficult and computationally intensive. A rapid computational method of protein stability prediction is developed using feed-forward neural networks and used to predict mutation-induced stability changes in Staphylococcal nuclease. The input to the neural network consisted of sequences of evolutionarily based amino acid similarity scores that were obtained through the comparison of the amino acids in a mutation containing sequence to their positional counterparts in the baseline wild-type amino acid sequence. A training set was created which consisted of similarity score sequences, for which the stabilities of the corresponding amino acid sequences were known, paired with the relative stabilities of the sequences to that of the baseline. Back-propagation of error was used to train the network to output accurate relative stability scores for the sequences in the training set. Neural network-based relative stability predictions for 55 sequences containing mutation combinations not found in the training set had an accuracy of 92.8%.

Amino acid sequence autocorrelation vectors and bayesian‐regularized genetic neural networks for modeling protein conformational stability: Gene V protein mutants

“…Taking into account that conformational stability is a more complex protein property in comparison to other physical stability measurements such as protein melting point, the accuracy over 85% of our approach for modeling the stability of gene V protein mutants is remarkably good. In this sense, our result is about the same range of about 90% obtained by Frenz for the ANN‐prediction of the relative stability of Staphylococcal nuclease mutants using similarity score vectors 30. In addition, the statistical quality of our ensemble model is in concordance with the report of Marrero–Ponce et al36 in which they extended topological indexes to the study of biological macromolecules.…”

“…In this regard, some X‐ray structural‐independent protein stability prediction methods have gained attention. The main advantages of such methods are they just use amino acid sequence information for predicting protein stability and are extremely less computational intensive in comparison with free energy function based methods 30. In this context, Levin and Satir31 successfully evaluated the functional significance of mutations on hemoglobin using amino acid similarity matrixes.…”

“…In this context, Levin and Satir31 successfully evaluated the functional significance of mutations on hemoglobin using amino acid similarity matrixes. Recently, Frenz30 reported an artificial neural network‐based model for predicting the stability of Staphylococcal nuclease mutants using amino acid similarity scores as network inputs.…”

Protein radial distribution function (P-RDF) and Bayesian-Regularized Genetic Neural Networks for modeling protein conformational stability: Chymotrypsin inhibitor 2 mutants

“…In this context, Levin and Satir [17] successfully evaluated the functional significance of mutations on hemoglobin using amino acid similarity matrixes. Recently, Frenz [18] reported an Artificial Neural Networkbased model for predicting the stability of Staphylococcal Nuclease mutants using amino acid similarity scores as network inputs.…”

Structural analysis and prediction of protein mutant stability using distance and torsion potentials: Role of secondary structure and solvent accessibility

“…In addition, a simple neighborhood analysis potential21 was used specifically for the hydrophobic core mutations, which uses four‐body potentials derived from protein packing parameters. Neural network22, 23 and support vector machine‐based methods24 were also recently implemented successfully, which facilitated considerable improvement over predictions based on above empirical methods.…”