Principles of Protein Structural Organization - H. Schultz 1982

Prediction of secondary structure from amino acid sequence
Application of predictive methods

Simple prediction Methods are the most popular. Predictive approaches typically involve cumbersome calculations using extensive tables (such as Nagano's method [353]), meaning that their reproduction requires obtaining a computer program, tables, and parameters derived from a benchmark set From the Authors. While some methods can be applied without computing equipment, they are so intricate that they are exceptionally time-consuming (for instance, Lim's method [380]). Authors rarely employ approximations that could simplify the application of a predictive method. Nevertheless, several straightforward and consequently most popular methods exist. Among them, the method of Chou and Fasman [340] appears to be the most widely used. As a simple illustration, let us consider this method's predictions for the first 24 residues of adenylate kinase.

The first step of the Chou and Fasman method involves identifying helix and sheet initiation sites. Predictions of the helix and ß-Structure are conducted independently and thus performed simultaneously. The Amino Acid Sequence, the values characterizing the propensity of residues to form helices and ß-structures (Tables 6.1 and 6.2), as well as the designations H, h, I, i, b, B, are presented in Fig. 6.2.

The First stage consists of locating the initiation centers. Helix initiation requires the presence of a hexapeptide containing four helix-forming residues h^a or H^a (I^a is equivalent to half of h^a), along with no more than one helix breaker b^a or B^a. Four such Peptides, defining two nucleations, are shown in Fig. 6.2. Similarly, ß-structure initiation occurs when a pentapeptide contains three h^β or H^β residues and no more than one b^β and B^β residue. As can be seen from Fig. 6.2, there are four such peptides defining a single initiation center. Both centers (helix and sheet) are located near residue 12. To resolve this ambiguity, the highest average propensities are compared. Since the propensity for the ß-structure (1.56) is higher than that for the helix (1.09), this segment is assigned as the ß-structure initiation center.

Class="center">

Fig. 6.2. Prediction of a-helix, ß-structure, and reverse turn (rt) for the 24 N-terminal residues of adenylate kinase [389] using the Chou-Fasman method [340].

The P_a and P_β parameters characterizing the propensity of residues to occur in an a-helix and a ß-structure (a- and ß-potentials of residues, respectively) are taken from Tables 6.1 and 6.2*. Evaluation criteria are given in the boxed inset. For a-helix prediction, hexapeptides with an average propensity greater than 1.00 are indicated by solid lines. Pentapeptides with an average ß-propensity greater than 1.00 are marked similarly. Tetrapeptide terminal regions of helices and pleated sheets with an average propensity above 1.00 are marked with dash-dotted lines, and those below 1.00 with dashed lines. The propensity of a residue to occur in a chain turn depends on its position within the tetrapeptide under consideration. Thus, four propensities are determined for each residue. All of them are multiplied by 10³. The rt-potential is the product of the residue propensities in the tetrapeptide. In the figure, they are circled and shown on the same lines as the tetrapeptide rt-propensities. All potentials are multiplied by 10⁺⁶. Tetrapeptides with potentials exceeding 50∙10^-6 are underlined with solid lines, and those slightly below this value with a dashed line.

Helix and ß-structure nucleations propagate in both directions. At the next stage, the helix nucleation centered at residue 4 and the pleated sheet nucleation centered at residue 12 propagate in both directions until the average tetrapeptide propensity drops below 1.00. The rules do not specify precisely which of the residues in the terminating tetrapeptide should be included in the helix and ß-structure, respectively. Apparently, it is appropriate to include only the H, h, and I residues. Thus, a helix is predicted for residues 1–7, and a ß-structure for residues 10–14.

Prediction of reverse turns. To predict turns, Chou and Fasman [340] applied the method of Lewis et al. [326] with an expanded base set. The reverse turn potential of a tetrapeptide was defined as the product of the turn propensities of the four residues at positions i, i + 1, i + 2, and i + 3 (Sec. 6.1). The corresponding values are shown in Fig. 6.2. Using a threshold of 50 ∙ 10^-6, turns were predicted for residues 16–23.

A comparison between the predicted and observed secondary structures in this case reveals good agreement: the predicted helix is shorter by one residue, while the ß-sheet and turn are predicted accurately. However, this example is not entirely representative because the a, ß, and rt potentials in this case are exceptionally high and distinct. Usually, in real structures, the situation is less definite and predictions are considerably less accurate.