Grading amino acid properties increased accuracies of single point mutation on protein stability prediction

Abstract

BackgroundProtein stabilities can be affected sometimes by point mutations introduced to the protein. Current sequence-information-based protein stability prediction encoding schemes of machine learning approaches include sparse encoding and amino acid property encoding. Property encoding schemes employ physical-chemical information of the mutated protein environments, however, they produce complexity in the mean time when many properties joined in the scheme. The complexity introduces noises that affect machine learning algorithm accuracies. In order to overcome the problem we described a new encoding scheme that graded twenty amino acids into groups according to their specific property values.ResultsWe employed three predefined values, 0.1, 0.5, and 0.9 to represent 'weak', 'middle', and 'strong' groups for each amino acid property, and introduced two thresholds for each property to split twenty amino acids into one of the three groups according to their property values. Each amino acid can take only one out of three predefined values rather than twenty different values for each property. The complexity and noises in the encoding schemes were reduced in this way. More than 7% average accuracy improvement was found in the graded amino acid property encoding schemes by 20-fold cross validation. The overall accuracy of our method is more than 72% when performed on the independent test sets starting from sequence information with three-state prediction definitions.ConclusionsGrading numeric values of amino acid property can reduce the noises and complexity of input information. It is in accordance with biochemical concepts for amino acid properties and makes the input data simplified in the mean time. The idea of graded property encoding schemes may be applied to protein related predictions with machine learning approaches.

Relationship between amino acid changes in mitochondrial ATP6 and life-history variation in anguillid eels

“…Differences in properties of amino acid replacements between the ancestral and species sequences were assessed using 11 physico-chemical and four structural properties [23]. The numerical differences of properties were summed to produce an overall score.…”

“…Genotype was represented by different datasets: (i) total nonsynonymous (amino acid changing) and synonymous changes and dN/dS considering all 13 genes, (ii) total amino acid changes at ATP6 and (iii) within five amino acid positions fixed between the American and European eels possibly under positive selection [8,9]. Differences in properties of amino acid replacements between the ancestral and species sequences were assessed using 11 physico-chemical and four structural properties [23]. The numerical differences of properties were summed to produce an overall score.…”

Machine-learning-guided directed evolution for protein engineering

“…While general-purpose kernels can be applied to protein inputs, there are also kernels designed for use on proteins, including spectrum and mismatch string kernels, 30,31 which count the number of shared subsequences between two proteins, and weighted decomposition kernels, 32 which account for three-dimensional protein structure. Support vector machines have been used to predict protein thermostability, 33,34,35,36,25,27,26,37 enzyme enantioselectivity, 38 and membrane protein expression and localization. 39 Gaussian process models combine kernel methods with Bayesian learning to produce probabilistic predictions.…”

PHDcleav: a SVM based method for predicting human Dicer cleavage sites using sequence and secondary structure of miRNA precursors

“…Several studies have identified single nucleotide polymorphisms (SNPs) in the miRNAs and their target genes, which possibly affect miRNA biogenesis, expression of target genes, and contribute to diseases [15,40-42]. Jazdzewski et al found that polymorphism in hsa-mir-146a reduces the formation of pre- and mature-miRNA [40].…”