Protein Structure and Function: Application of Bioinformatics Methods - John Rigden 2014

Bioinformatics Methods for Studying the Structure and Function of Unordered Proteins
Prediction of Disorder
Prediction Based on Contact Potentials

Some prediction Methods are based on a unique principle that differs from those described above. According to this concept, IDPs cannot fold into ordered structures because The amino acid residues in their sequence interact too weakly to overcome the unfavorable Entropy decrease that accompanies the folding process. Several prediction methods rely on this principle, utilizing simple statistical patterns (FoldUnfold (Galzitskaya et al. 2006)), pairwise interaction potential comparisons (Ucon (Schlessinger et al. 2007)), or the estimation of the total interaction energy between chain residues (IUPred (Dosztanyi et al. 2005a, b)). Let us examine the latter method in greater detail.

To estimate the total pairwise interaction energy within a polypeptide chain, IUPred employs low-resolution force fields (statistical potentials) derived from Globular Proteins. The core idea is that THE CONTRIBUTION OF an individual residue to the total energy is determined not only by the residue type itself, but also by other amino acid residues—its potential partners—along the sequence. Since a probabilistic description of all possible interactions between every residue is difficult to process, the problem is simplified by introducing a quadratic expression for the Amino Acid Composition. The contribution of an individual amino acid residue is approximated by an energy prediction matrix that relates the energy contributions of residues i and j. The matrix parameters are determined via least-squares fitting against globular proteins. Using this approach, the average energy level obtained for disordered proteins (-0,07 arbitrary units) is unfavorable compared to the corresponding value for globular proteins (-0,81 arbitrary units). This indicates that the approach in question is informative when the studied proteins belong to macroscopic structural states (Fig. 5.3). When calculations consider only predetermined local neighboring amino acid residues in the sequence, this approach yields sequence-specific information regarding structural disorder, forming The basis of the IUPred algorithm (Dosztanyi et al. 2005a, b).

Class="center">Structure/structure.files/image028.jpg" width="331"/>

Fig. 5.3. Estimated pairwise interaction energies of globular proteins and IDPs. The total pairwise interaction energies of globular proteins (shown as gray +) and intrinsically disordered proteins (shown as black x) were estimated based on their amino acid composition and plotted as a function of sequence length. The negative direction corresponds to increased stability driven by pairwise amino acid interactions. Mean values, expressed in arbitrary energy units, indicate greater stability for globular proteins (-0,81 units) than for IDPs (-0,07 units) (reprinted from Dosztanyi et al. 2005b with permission from Elsevier)