Computational Systems Biology Methods and Protocols.7z

the formula 1 þ^1 exin Wu et al. [9]. Zhao et al. [10] incorporated the

output of the PSSM with amino acid physicochemical properties, as

discussed in the following section.

2.2.6 Physicochemical
Properties

The specificity and diversity of protein structure and function are

largely attributed to the various properties of each of the 20 amino

acids in the sequence [37]. The physicochemical properties of amino

acids surrounding serine/threonine residues have been reported to

influence their O-GlcNAcylation. Therefore, some predictors have

applied the physicochemical and biochemical properties of amino

acids to predict protein O-GlcNAcylation sites [8, 10]. The AAindex

database (http://www.genome.jp/aaindex/)[43] includes amino

acid mutation matrices and amino acid indexes. Version 9.0 contains

544 physicochemical properties. An amino acid index is a set of

20 numerical values that denote various physicochemical properties

of amino acids. Physicochemical properties have been used success-

fully to predict several protein modifications [44–48]. Zhao et al.

[10] selected 17 informative physicochemical properties to encode

protein peptides in PGlcS method. They multiplied 20 PSSM values

of each residue by the corresponding physicochemical property of

the 20 amino acids and summed the properties of each residue

according to the following equation:

Fki¼

X^20

j¼ 1

PkjMij

wherePkjrepresents the value of thekth physicochemical property

for thejth amino acid type (k¼1, 2,...,L) andMijrepresents the

raw value of theith position for thejth amino acid type in the PSSM

(i¼1, 2,..., 15).

2.2.7 Secondary
Structure

To predict the secondary structure of all the amino acid residues,

the full-length protein sequences with O-GlcNAcylated sites were

submitted to PSIPRED [38]. PSIPRED outputs sequences com-

posed of the characters “C,” “H,” and “E,” where “C” represents

coil, “H” represents helix, and “E” represents strand. The orthog-

onal binary coding scheme was used to transform these secondary

structure terms into numeric vectors; for instance, a helix (H) is

encoded as “001,” a strand (E) is encoded as “010,” and a coil

(C) is encoded as “100” in PGlcS [10].

2.2.8 Disorder DISOPRED2, which is one of the leading servers for predicting
natively disordered regions in proteins [39], was applied to obtain
disorder information for the amino acid sequences. A disordered
structure is encoded as “0 1,” while an ordered structure is encoded
as “1 0” in PGlcS [10].

Computational Prediction of Protein O-GlcNAc Modification 241