Amino principal component analysis

The reason for the correlation between the localization and the amino acid composition was sought by Andrade et al. (1998). They examined the amino acid composition of proteins with known localization and three-dimensional structure in three ways total composition, surface composition, and interior composition. The principal component analysis showed the best correlation between the surface composition and the localization. Therefore, they concluded that the correlation is the result of evolutionary adaptation of proteins to the surrounding environment. 

Figure 9(b) shows the data points plotted in the scattering diagram on PC 1 and PC 2 by the principal component analysis. The first principal axis reflects bitterness and sweetness. The second principal axis reflects sourness and umami. Amino acids are classified clearly into five groups by the taste sensor. 

A later study using the same data set of 20 peptides showed that similar results could be obtained using a set of three descriptors, z values, based on an earlier analysis of changes in peptide properties due to amino acid substitutions. These were found using principal component analysis (75) of the effects of... 

Z-scales are obtained by principal component analysis of physico-chemical properties of monomers. E.g., the first Z-scales of amino adds describe hydrophobicity (z1), steric bulk/polarisability (z2) and polarity (z3) of the amino acids. 

Statistical Analysis The juice composition, as individual amino acids, free amino nitrogen and ratios of certain amino acids and groups of them were entered into a principal component analysis (PCA) using the SAS (Cary, NC) statistical software. The ratios considered were based on the thinking that relative proportions within the amino pool, rather than absolute quantities between alternative substrates might be related to the level of sulfide formation. 

The electronic properties of amino acid side chains are summarized in Table 3, and they represent a wide spectrum of measures. The NMR data are derived experimentally (37). The dipole (38), C mull, inductive, field, and resonance effects were derived from QM calculations (15). The VHSE5 (39) and Z3 (25) scales were developed for use in quantitative structure-activity relationship analysis of the biologic activity of natural and synthetic peptides. Both were derived from principal components analysis of assorted physico-chemical properties, which included NMR chemical shift data, electron-ion interaction potentials, charges, and isoelectric points. Therefore, these scales are composites rather than primary measures of electronic effects. The validity of these measures is indicated by their lack of overlap with hydrophobicity and steric parameters and by their ability to predict biologic activity of synthetic peptide analogs (25, 39). Finally, coefficients of electrostatic screening by amino acid side chains (ylocal and Ynon-local) were derived from an empirical data set (40), and they represent a composite of electronic effects. 

These are autocovariances and cross-covariances calculated from sequential data with the aim of transforming them into uniform-length descriptors suitable for QSAR modeling. ACC transforms were originally proposed to describe peptide sequences [Wold, Jonsson et al, 1993 Sjbstrbm, Rannar et al., 1995 Andersson, Sjostrom et al., 1998 Nystrom, Andersson et al., 2000]. To calculate ACC transforms, each amino acid position in the peptide sequence is defined in terms of three orthogonal z-scores, derived from a Principal Component Analysis (PC A) of 29 physico-chemical properties of the 20 coded amino acids. 

Ten principal properties were calculated by Principal Component Analysis on 188 physicochemical properties for the 20 coded amino acids [Kidera, Konisci et al., 1985a, 1985b]. These 10 properties were called KOKOS descriptors by Pogliani on the basis of the Authors names [Pogliani, 1994a] they describe most of the conformational, bulk, hydrophobicity, a-helix, and (3-structure properties of amino acids. 

To calculate ACC transforms of peptide sequences, each amino acid in the peptide sequence was described by three orthogonal z-scores (Table B3), derived from a —> Principal Component Analysis on 29 physico-chemical properties of the 20 coded amino a.cids (Hellberg,... 

SSIA descriptors (Scores of Structural Information for Amino acids) are z-scores derived from Principal Component Analysis on 3D VAIP descriptors for the 20 coded amino acids [Zhou, Zhou et al, 2006]. 

T-scale is a five-dimensional vectorial descriptor (Table B5) derived from Principal Component Analysis on 67 topological indices of 135 amino acids [Tian, Zhou et al, 2007]. 

A report where quantitative 2D NMR was applied to differentiation of beer brands was published by Khatib et al. They used 2D /-resolved NMR to improve resolution of components present in 2-butanol extracts of several beer brands. With application of principal component analysis they distinguished six lagers from each other based on their amino acid contents. They also discussed how the amount of compounds that have an important effect to the taste, like tyrosol, can be estimated with the method. 

The relationships between the amino acids are borne out by a principal component analysis which projects the 9-D space onto a plane (Figure 17.1 b). About 90% of the total variance of the data is accounted for by the two components which can be inter-... 

These studies use proton NMR spectroscopy to measure amino acids and other metabolites. The changes of the individual components of the NMR profile can yield information on the regional effects in the kidney (Holmes, Bonner, and Nicholson 1997 Holmes et al. 1998 Lindon, Holmes, and Nicholson 2004 Robertson et al. 2005). Several investigators have applied principal component analysis to improve the identification of affected regions of the nephron. As for other renal tests, the timing and collection procedures are critical to the application in addition, several of the measured metabolites are affected by other organ toxicities, particularly hepatotoxicity. 

In the literature, a large number of substituent descriptors have been reported. In order to use this information for substituent selection, appropriate statistical methods may be used. Pattern recognition or data reduction techniques, such as principal component analysis (PCA) or cluster analysis (CA) are good choices. As explained in Section V in more detail, PCA consists of condensing the information in a data table into a few new descriptors made of linear combinations of the original ones. These new descriptors are called principal components or latent variables. This technique has been applied to define new descriptors for amino acids, as well as for aromatic or aliphatic substituents, which are called principal properties (PPs). The principal properties can be used in factorial design methods or as variables in QSAR analysis. 

Amino acids were characterized by a principal component analysis of their side-chain properties [170, 171] first, for 20 coded amino acids three principal components Zj, Z2 and Zj were derived (interpreted as being related to hydrophilicity, side-chain bulk, and electronic properties) [170] and afterwards new z scales resulted [171] from a partial least squares (PLS) analysis of the side-chain properties of these amino acids and additional 35 noncoded (unnatural) amino acids. The use of these scales (instead of the original variables) was recommended for structure-activity analyses. 

To construct the protein descriptors, z-scales were used, z-scales, originally developed as a descriptor of amino acids, contain three variables designated Zj, z, and Zj [14]. The z parameters were determined by principal component analysis of 29 physicochemical parameters characterizing 20 natural amino acids. The first, second, and third principal components, corresponding to z, z, and z, respectively, could be tentatively interpreted as hydrophobicity, steric, and electronic properties. 

To create a set of variants we identified variable amino acid residues by comparing proteinase K with homologous sequeiwes from the serine protease family using standard phylogenetic analysis, structural conq)arisons and principal component analysis. Principal component analysis can be used to reduce the high dimensional complexity of sequence variations by creating new composite dimensions which account for the majority of the difierences within a... 

The principal component analysis also clusters related functional measures. We selected three representative activities based on the functional clustering shown in Figure 6 for further analysis activity towards AAPL-p-NA at pH 7.0, absolute activity towards AAPL-p-NA following 5 minutes at 65°C and activity towards casein. For each of these activities we constructed PLSR models similar to that shown in Figure 3, and calculated the regression coefficients for each amino acid variation as shown for thermal tolerance in Figure 4. The changes calculated to contribute positively to each property are shown in Table 1. 

The genetic code, regardless of whether it is a product of a "frozen accident" [1] or a deterministic interaction between the nucleotides and the amino acids [2], displays an apparent correlation between the nucleotides found at particular codon positions and the physico-chemical properties of the protein amino acid residues encoded by the nucleotides [2-10]. A variety of analytical methods have been employed to quantitatively examine these relationships. SjdstrOm and Wold [5], for example, have used Principal Component Analysis (PCA) to relate twenty physical properties of the amino acids to the genetic code. They find that 58% of the variance in the data can be accounted for by considering just three factors. In order of importance, the predominant contributions are 1) hydrophobicity, 2) molar volume, and 3) electronic descriptors (e.g. pKgS and NMR chemical shifts). Fig. 1 presents a concise display of... 

A cluster analysis of the amino acid structures by PCA of the A -matrix is shown in Figure 6.5a note that PCA optimally represents the Euclidean distances. The score plot for the first two principal components (preserving 27.1% and 20.5% of the total variance) shows some clustering of similar structures. Four structure pairs have identical variables 1 (Ala) and 8 (Gly), 5 (Cys) and 13 (Met), 10 (He) and 11 (Leu), and 16 (Ser) and 17 (Thr). Objects with identical variables of course have identical scores, but for a better visibility the pairs have been artificially... 

---