Practical Protein Chemistry - A. Darbre 1989

Prediction of Peptide and Protein Conformation
Current State of Prediction Methods

Over the past five years, numerous significant results have been achieved, steadily reinforcing the conviction that the protein folding problem will be solved in the near future. While it is impossible to provide an exhaustive Structure/133.html">Discussion of the Prospects in this field within The Scope of this review, we will outline a few relevant considerations. It should be noted that researchers hold diverse views regarding the primary direction of future investigations. For instance, Levitt and Scheraga argue that leveraging experimental data on the conformational properties of a protein is extremely beneficial for successful spatial folding simulations. NMR data are particularly informative in this regard, as they readily yield numerous interatomic distances within protein molecules. The challenge here is to achieve results using a minimal amount of experimental input. Naturally, as the theory advances, the required volume of such data will steadily decrease. Secondly, work in the author's laboratory demonstrated that tertiary structure prediction Methods are fully applicable to "miniproteins"—biological Peptides containing up to ten amino acid residues. The prediction accuracy for Proteins 100–200 residues in length should align with the specific goals of the research. For instance, it has been shown that sufficiently approximations are entirely adequate for designing synthetic peptide Vaccines which, upon conjugation with a carrier molecule, acquire immunogenic properties and elicit Antibodies against the target peptide. Furthermore, preliminary Spatial Structure modeling, performed long before comprehensive crystallographic data become available, is expected to prove valuable in biotechnological studies. Consequently, the growing body of research focusing on proteins with unknown tertiary structures holds far greater interest than academic or methodological studies concerning proteins whose detailed Conformations are already well established. Pharmacologists, neurochemists, immunologists, and genetic engineers have very little interest in computational models of proteins with experimentally well-known spatial structures; for these groups of scientists, predictive modeling of the proteins under investigation is of paramount importance.

Crucial Factors Determining the Qualitative and quantitative capabilities of protein molecule calculations include the speed performance of algorithms and software. Although increasing computational speed is primarily a technical issue, considerable efforts are being directed toward this goal in many biological laboratories. In the author's laboratory, for example, calculation speeds have increased at least a thousandfold compared to 1979. Perhaps one of the most powerful acceleration factors has been access to high-speed vector supercomputers such as the CRAY 1 or CYBER 205. For instance, utilizing the CYBER 205 computer increased calculation speed by a factor of 30 compared to its predecessor, the CDC 7600. Algorithms implemented on such computers, alongside simple program "vectorization," must exploit the advanced capabilities offered by vector processors. A substantial increase in computing speed has enabled a dramatic expansion in the number of permissible protein Folding Pathways, which in turn has provided the basis for several important Conclusions. It has become feasible to examine the details of the potential energy surface as a function of protein conformation. Results obtained by us and other researchers indicate that the calculated surface contains several broad basins, each harboring dozens, hundreds, or occasionally several thousand local minima. It has also transpired that Levitt's Analysis of the potential energy surface shape actually describes a single basin containing the native conformation. Moreover, the approaches developed by Levitt, Scheraga, and Robson proved remarkably similar in reaching the native conformation, provided that the correct basin was already identified. In contrast, the approach of Blundell et al. relies on examining and classifying all known conformational motifs and mapping them to protein Amino acid sequences. When predicting an unknown protein conformation, one must correctly select the structural motif corresponding to the specific Primary Structure, after which energy Minimization is performed under the assumption that the starting conformation lies near the basin containing the native Cell/13.html">Protein Structure.

Robson's work has been dedicated to searching for criteria whose fulfillment ensures that a protein conformation resides near the correct basin. It was demonstrated that potential or Free energy magnitude cannot be considered a definitive factor; for instance, the starting conformational point even within the correct basin may possess a very high energy value. Furthermore, several studies have shown that the bottoms of different basins can harbor points with energies comparable to that of the Native State. The primary determinant of the native conformation appears to be the favorable arrangement of nonpolar hydrophobic groups, which becomes even more pronounced when The Role of the solvent is properly accounted for in energy calculations. Building upon Sternberg's solvent model to account for intramolecular hydrogen bonding, we developed a method that ensures greater stability for the native conformation compared to more open conformers. Consequently, conformations outside the target basin exhibit higher free energy than those within the basin containing the native structure. The method has not yet been refined to the point of guiding the search directly into the region containing the native conformation; currently, it operates successfully primarily within the immediate vicinity of the target basin. Apparently, search trajectories should be directed in such a way as to avoid leading deep into dead-end basins. The general path to solving this problem is clear, although certain formidable theoretical hurdles remain to be overcome. These can potentially be resolved by modeling protein self-assembly processes in four-dimensional space rather than three-dimensional space. The computational time required for such modeling can be significantly reduced by incorporating recent advances in spatial topology.

It should be noted that the current advancement of protein conformation prediction methods is constrained less by imperfect potential Functions, the difficulties of accurately accounting for solvent effects, or deficiencies in software and hardware, and more by a shortage of fundamental new mathematical developments in this field. In this review, we have limited our discussion to software tools that optimize computational time when exploring conformational space. Nevertheless, achieving successful spatial folding of a protein chain still demands substantial computer resources, a limitation tied to the aforementioned slow progress in the relevant branches of mathematics. It must also be emphasized that enormous computational costs translate directly into significant calendar time, meaning that current successes in modeling protein folding processes remain only partial. For instance, while progress in engineering Structural domains in proteins is highly encouraging, triggering their association into a native-like structure has not yet been achieved (and achieving The formation of certain individual domains can also prove difficult).

In Conclusion, it must be stressed that a great deal of work lies ahead, but rapid progress in this field provides strong grounds for optimism. Unfortunately, it is not yet possible to present a striking example of folding a medium-sized protein based solely on its Amino Acid Sequence. However, it is fair to say that computational methods are successfully applied in biotechnological areas related to drug design, refinement of crystallographic structures, Homology-based protein structure modeling, and more. Moreover, such Applications of theoretical techniques are becoming increasingly routine and indispensable.