A new set of amino acid descriptors and its application in peptide QSARs

Abstract

In this work, a new set of amino acid descriptors, i.e., VHSE (principal components score Vectors of Hydrophobic, Steric, and Electronic properties), is derived from principal components analysis (PCA) on independent families of 18 hydrophobic properties, 17 steric properties, and 15 electronic properties, respectively, which are included in total 50 physicochemical variables of 20 coded amino acids. Using the stepwise multiple regression (SMR) method combined with partial least squares (PLS), the VHSE scales are then applied to QSAR studies of bitter-tasting dipeptides (BTD), angiotensin-converting enzyme (ACE) inhibitors, and bradykinin-potentiating pentapeptides (BPP). To validate the predictive power of resulting models, external validation are also performed. A comparison of the results to those obtained with z scores and other two-dimensional (2D) or three-dimensional(3D) descriptors shows that the VHSE scales are comparable for parameterizing the structural variability of the peptide series.

Predicting the Flexibility Profile of Ribosomal RNAs

“…The 0.311 used here as the threshold is because this value is the default setting in the widely used chemometrics program SIMCA-P (http://www.umetrics.com/simca) and has been demonstrated to be reasonable for high-dimensional PLS modeling. [43][44][45] (ii) For the LSSVM regression, a coarse-grained gridsearching scheme using the root-mean-square error of cross-validation (RMSCV) as the objective function was carried out to determine the optimum combination of regularization g and kernel parameter s 2 . In this procedure, the natural logarithms of g and s 2 were tuned simultaneously in a grid ranging from 0 to 10 with step size of 1 and the combination of g and s 2 that give rise to the RMSCV minimum was ultimately determined.…”

Peptide binding specificities of HLA-B*5701 and B*5801

“…VHSE, a set of amino acid descriptors, was derived from the research of Mei et al [19]. Fifty physico-chemical properties of 20 coded amino acids, consisting of 18 hydrophobic properties, 17 steric properties, and 15 electronic properties, were used for principal component analysis (PCA).…”

“…That is to say, the hydrophobic, steric, and electronic properties of 20 coded amino acids can be characterized by 8 principal component scores with minimal loss of information. These 8 score vectors are called the VHSE (principal components score vectors of hydrophobic, steric, and electronic properties) descriptors [19]. For the 20 amino acids, VHSE1 and VHSE2 are related to hydrophobic properties, VHSE3 and VHSE4 to steric properties, and VHSE5 to VHSE8 to electronic properties (Table 1).…”

Structural Parameter Characterization and Bioactivity Simulation Based on Peptide Sequence

“…Collantes et al [8] established two computable 3-D descriptors, Isotropic Surface Area (ISA) and Electronic Charge Index (ECI), on the base of three-dimensional structural characters of amino acid side chains. Recently, Mei et al [9] obtained VHSE descriptors which were derived from the PCA individually on hydrophobic, steric, and electronic properties of 50 physicochemical variables of 20 coded amino acids. The descriptors showed a good parameterization for the structural variability in the peptides.…”

“…To update, many studies indicated that the interaction between the drug molecules and the acceptor mainly displays in the nonbonding effects, such as hydrophobic properties, steric properties, and electronic properties and the hydrogen bond contribution and so on. Therefore, in this paper, we proposed a novel set of amino acid descriptors based on our previous work [9,12,13]. The novel amino acid descriptors, Hydrophobic, Electronic, Steric, and Hydrogen (HESH) (principal component score vector of HESH bond contribution properties), were derived from Principal Components Analysis (PCA) on 171 physicochemical properties of 20 natural amino acids.…”