Protein loop prediction

HWT van Vlijmen, M Karplus. PDB-based protein loop prediction Parameters for selection and methods for optimization. I Mol Biol 267 975-1001, 1997. 

U. Lessel, D. Schomburg. Importance of anchor group positioning in protein loop prediction. Proteins. 1999, 37, 56-64. 

I Wojcik, I-P Mornon, I Chomilier. New efficient statistical sequence-dependent structure prediction of short to medium-sized protein loops based on an exhaustive loop classification. I Mol Biol 289 1469-1490, 1999. 

The loop prediction algorithm implemented in the Protein Local Optimization Program (PLOP) is described in detail in [123]. During loop build-up,... 

We have tested the loop prediction algorithms on two sets of protein loops of known structure (see [101]). The first set is composed of the 57 nine-residue loops that were originally compiled by Fiser et al. [126] and by Xiang et al. [127]. The 35 13-residue loop set is the same as the one investigated by Zhu et al. [124],... 

In protein secondary structure prediction, where a three-category (a, P, and coil or loop) prediction is made, the accuracy can be measured by a 3 x 3 accuracy table, as in Rost and Sander (1993). 

Figure 3.2 Predicted 12-transmembrane domain model of rat Oatplal. Conserved amino acids are indicated in blue. Conserved and charged amino acids (D, E, K, R) are given in red, and conserved cysteines (C) are marked with green. Three potential N-glycosylation sites (Y) are present on extracellular protein loops. The OATP superfamily signature is indicated at the border ofthe extracellular loop 3 and the transmembrane domain 6.

Abbreviations used in table MC - Monte Carlo aa - amino acid vdW - van der Waals potential Ig - immunoglobulin or antibody CDR - complementarity-determining regions in antibodies RMS -root-mean-square deviation r-dependent dielectric - distance-dependent dielectric constant e - dielectric constant MD - molecular dynamics simulation self-loops - prediction of loops performed by removing loops from template structure and predicting their conformation with template sequence bbdep - backbone-dependent rotamer library SCMF - self-consistent mean field PDB - Protein Data Bank Jones-Thirup distances - interatomic distances of 3 Ca atoms on either side of loop to be modeled. 

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. 

Current prediction techniques allow a detailed and reasonable structure to be generated only for small or constrained systems. Five- or 10-residue protein loops or molecules such as enkephalin represent the limit of current techniques. A larger database of protein conformational motifs may increase the success of prediction, but brute force computational approaches will remain impractical for some time. The incorporation of a priori knowledge, whether at an atomic level or through abstraction at a higher level of organization, appears to be the best means available to understand and predict protein conformation. 

As discussed above, side-chain movement or protein flexibility plays an important role in a docking process. Usually, uncertainly in side-chain placement or loop modelhng arises in protein stmctures predicted through homology modelling. Only a few docking tools consider these, like FlexE. This is a tool which considers the protein structure variations or flexibility to dock flexible ligands [22] (Fig. 4.51). 

GA s have received much attention in recent years. In chemistry, they can be used for a search in conformational space which very often involves combinations of many parameters. In particular, there were several attempts to use them for protein structure prediction (reviewed in 48). Recently, GA s were suggested to use in three rather different aspects of RNA structurerconformational search for stem-loop structures 49), prediction of optimal and suboptimal secondary structures 50) and simulation of RNA folding pathways 44,45). In the case of RNA folding simulation, a GA is also very attractive because it allows to simulate the process, in addition to obtaining a final solution. 

Rummey C, Metz G (2007) Homology models of dipeptidyl peptidases 8 and 9 with a focus on loop predictions near the active site. Proteins 66 160-171... 

---