All-atom protein structure prediction

This review indicates that all-atom protein structure prediction with stochastic optimization methods becomes feasible with present-day computational resources. The fact that three proteins were reproducibly folded with different optimization methods to near-native conformation increases the confidence in the parameterization of our all-atom protein force field PFFOl. The... 

Summary. We recently developed an all-atom free energy force field (PFFOl) for protein structure prediction with stochastic optimization methods. We demonstrated that PFFOl correctly predicts the native conformation of several proteins as the global optimum of the free energy surface. Here we review recent folding studies, which permitted the reproducible all-atom folding of the 20 amino-acid trp-cage protein, the 40-amino acid three-helix HIV accessory protein and the sixty amino acid bacterial ribosomal protein L20 with a variety of stochastic optimization methods. These results demonstrate that all-atom protein folding can be achieved with present day computational resources for proteins of moderate size. 

Samudrala and Moult described a method for handling context sensitivity of protein structure prediction, that is, simultaneous loop and side-chain modeling, using a graph theory method [198, 209] and an all-atom distance-dependent statistical potential energy function [199]. Their program RAMP is listed in Table 5.6. 

R. Samudrala, J. Moult. An all-atom distance-dependent conditional probability discriminatory function for protein structure prediction. / Mol Biol. 1998, 275, 895-916. 

Given that elegant advances are being made in automated protein structure classification and even with the soon-to-be-initiated production stage of the PSI (called PSI-2), the difficulties inherent in protein crystallization imply that not all possible protein structures will be known in the near future. Accordingly, there is a need to predict at atomic resolution the three-dimensional (3-D) shape of novel designer proteins and proteins whose... 

Kmiecik, S., Gront, D., 8c Kolinski, A. (2007). Towards the high-resolution protein structure prediction. Fast refinement of reduced models with all-atom force field. BMC Structural Biology, 7, 43. 

Independent of the exact features of the model or criterion defining the protein s folded state, the computational demands of evaluating thermodynamic and kinetic properties of these models can be formidable. At the present time, the best methods combined with the most powerful computational engines are inadequate to fold an all-atom model of a protein in computo. As such, a careful choice of the computational method is essential. The development of new computational methods is infinite in its possibilities. The field of development of conformational optimization algorithms for proteins has shown rapid progress in recent years. This rapid development of new algorithms promises to continue. This article provides a snapshot of the field of protein structure prediction as a problem of conformational optimization. There is an emphasis on the most general and fundamental methods where further development appears to be most likely. The discussion is not intended to be a comprehensive review or even a survey of the most effective methods. The reader is referred to the references for a more comprehensive discussion. 

This section briefly reviews prediction of the native structure of a protein from its sequence of amino acid residues alone. These methods can be contrasted to the threading methods for fold assignment [Section II.A] [39-47,147], which detect remote relationships between sequences and folds of known structure, and to comparative modeling methods discussed in this review, which build a complete all-atom 3D model based on a related known structure. The methods for ab initio prediction include those that focus on the broad physical principles of the folding process [148-152] and the methods that focus on predicting the actual native structures of specific proteins [44,153,154,240]. The former frequently rely on extremely simplified generic models of proteins, generally do not aim to predict native structures of specific proteins, and are not reviewed here. 

Eortunately, a 3D model does not have to be absolutely perfect to be helpful in biology, as demonstrated by the applications listed above. However, the type of question that can be addressed with a particular model does depend on the model s accuracy. At the low end of the accuracy spectrum, there are models that are based on less than 25% sequence identity and have sometimes less than 50% of their atoms within 3.5 A of their correct positions. However, such models still have the correct fold, and even knowing only the fold of a protein is frequently sufficient to predict its approximate biochemical function. More specifically, only nine out of 80 fold families known in 1994 contained proteins (domains) that were not in the same functional class, although 32% of all protein structures belonged to one of the nine superfolds [229]. Models in this low range of accuracy combined with model evaluation can be used for confirming or rejecting a match between remotely related proteins [9,58]. 

T. Herges and W. Wenzel. An All-Atom Force Field for Tertiary Structure Prediction of Helical Proteins. Biophys. J., 87(5) 3100-3109, 2004. 

DNA (like in the human genome project), which codes all the proteins of the species, each protein given as a sequence of amino acids. The estimated number of proteins in the terrestrial biological organisms is of the order of 107, while the number of the 3D structures (in the atomic resolution) deposited in the Protein Data Bank (as of 24 June 2008) is 47,526. Thus, theoretical prediction of the 3D shape of proteins, being potentially much faster than the X-ray and NMR techniques, seems to be a must for contemporary biology. 

Motion in a protein may be modeled by computer using Isaac Newton s equation of motion, / = ma. This modeling requires the three-dimensional coordinates from an X-ray structure analysis as a starting point and some knowledge of interatomic potentials, so that only reasonable interatomic distances will be employed at all stages.Such molecular dynamics calculations lead to a prediction of where atoms will move in a short period of time, and result in the calculation of a time-dependent trajectory of all atoms. Initially each atom is moved in the direction of the force on it from other atoms and then, as each atom moves, its trajectory may change to accommodate this. In addition, this method aids in protein structure refinement,as was described in Chapter 10, although it is important to ensure that the model so refined still fits the electron density map. 

---