Secondary-Structure Prediction

JPRED [66,67] is a neural network-based program for predicting protein secondary structure sequence residues are assigned to one of three secondary-structure elements (alpha helix, beta sheet, or random coil). 

MEMSAT [70] is a program for predicting the secondary structure and topology (helical orientation) of integral membrane proteins. 

The first step, as alluded to above, is the development of possible loop conformations which connect the regions of secondary structure The loops which do not fit into the well-defined category of a-helices or (1-sheets have been fairly well characterized using the data base of proteins for which the three-dimensional structure is known [15,16], The identification of specific loop conformations provides insight into the possible orientations, or at least provides limitations on the possible orientations, of the various secondary structural elements. The second step is then analysis of the array of amino acids within the secondary structural elements with attention to the environment in which the amino acids would be found. It is clear that a cluster of hydrophobic amino acids would not likely be projecting into the aqueous solution, and more likely projecting into the core of the protein. This analysis provides additional restrictions to the number of possible arrangements in which the secondary structural elements may be found. 

Another approach is to map the arrangement of secondary structural elements onto the known tertiary structures of other proteins. Currently, approximately one hundred unique protein folds have been identified. There is some question as to if this is an upper limit. If this is indeed the case, then the protein of unknown structure must adopt a known topological fold. The secondary structural elements are mapped onto the template of the different known protein structures. The best fits, as judged by the environmental factors (solvent accessibility) of the individual amino acids, are then further analyzed as probable folds. This procedure is referred to as threading the secondary elements into three-dimensional structures [28], 

---

Rule-based Approaches Using Secondary Structure Prediction... 

Ihe rule-based approach to protein structure prediction is obviously very reliant on th quality of the initial secondary structure prediction, which may not be particularly accurate The method tends to work best if it is known to which structural class the protein belongs this can sometimes be deduced from experimental techniques such as circular dichroism... 

Cuff IA and G J Barton 1999. Evaluation and Improvement of Multiple Sequence Methods for P Secondary Structure Prediction. Proteins Structure, Function and Genetics 34 508-519. 

For example, Stolorz et al. [88] derived a Bayesian formalism for secondary structure prediction, although their method does not use Bayesian statistics. They attempt to find an expression for / ( j. seq) = / (seq j.)/7( j.)//7(seq), where J. is the secondary structure at the middle position of seq, a sequence window of prescribed length. As described earlier in Section II, this is a use of Bayes rule but is not Bayesian statistics, which depends on the equation p(Q y) = p(y Q)p(Q)lp(y), where y is data that connect the parameters in some way to observables. The data are not sequences alone but the combination of sequence and secondary structure that can be culled from the PDB. The parameters we are after are the probabilities of each secondary structure type as a function of the sequence in the sequence window, based on PDB data. The sequence can be thought of as an explanatory variable. That is, we are looking for... 

Thompson and Goldstein [89] improve on the calculations of Stolorz et al. by including the secondary structure of the entire window rather than just a central position and then sum over all secondary strucmre segment types with a particular secondary structure at the central position to achieve a prediction for this position. They also use information from multiple sequence alignments of proteins to improve secondary structure prediction. They use Bayes rule to fonnulate expressions for the probability of secondary structures, given a multiple alignment. Their work describes what is essentially a sophisticated prior distribution for 6 i(X), where X is a matrix of residue counts in a multiple alignment in a window about a central position. The PDB data are used to form this prior, which is used as the predictive distribution. No posterior is calculated with posterior = prior X likelihood. 

JE Gibrat, J Garnier, B Robson. Eurther developments of protein secondary structure prediction using information theory. New parameters and consideration of residue pairs. J Mol Biol 198 425-443, 1987. 

LH Holley, M Karplus. Protein secondary structure prediction with a neural network. Proc Natl Acad Sci USA 86 152-156, 1989. 

JM Chandoma, M Karplus. The importance of larger data sets for protein secondary structure prediction with neural networks. Protein Sci 5 768-774, 1996. 

During this process of designing sequence changes, models were built and assessed to ensure that there were no obvious steric clashes and that the hydrophobic core was well packed. Furthermore, secondary structure prediction was also used to monitor the progress of change and to choose among different possible substitutions. The final sequence (see Table 17.3) contains 28 changes it had 50% identity to B1 and the similarity to Rop had increased from 5.4% identity to 41%. 

Barton, G.J. Protein secondary structure prediction. Curr. Opin. Struct. Biol. 5 372-376, 1995. 

Rost, B., Sander, C., Schneider, R. Redefining the goals of protein secondary structure prediction. /. Mol. Biol. 

Building sequence profiles or Hidden Markov Models to perform more sensitive homology searches. A sequence profile contains information about the variability of every sequence position, improving structure prediction methods (secondary structure prediction). Sequence profile searches have become readily available through the introduction of PsiBLAST [4]... 

Natural mutation of amino acids in the core of a protein can stabilize the same fold with different complementary amino acid types, but they can also cause a different fold of that particular portion. If the sequence identity is lower than 30% it is much more difficult to identify a homologous structure. Other strategies like secondary structure predictions combined with knowledge-based rules about reciprocal exchange of residues are necessary. If there is a reliable assumption for common fold then it is possible to identify intra- and intermolecular interacting residues by search for correlated complementary mutations of residues by correlated mutation analysis, CMA (see e.g., http //www.fmp-berlin.de/SSFA). 

Secondary structural predictions about NPAs, and direct biophysical measurements, have demonstrated that the NPAs are rich in a-helix, with no p-structure either predicted from secondary structure prediction algorithms, or detected by circular dichroism (Kennedy et al, 1995b). In this they are the antithesis of the similarly sized cLBPs and lipocalins. The predictions are that each individual NPA unit protein will fold into four main regions of helix, and it has been speculated that the tertiary structure is as a four-bundle helix protein, similar to other invertebrate carrier proteins (Sheriff et al., 1987). 

Currently, there exists an enormous and growing deficit between the number of polypeptides whose amino acid sequence has been determined and the numbers of polypeptides whose three-dimensional structure has been resolved. Given the complexities of resolving three-dimensional structure experimentally, it is not surprising that scientists are continually attempting to develop methods by which they could predict higher order structure from amino acid sequence data. Although modestly successful secondary structure predictive approaches have been developed, no method by which tertiary structure may be predicted from primary data has thus far been developed. 

Over 20 different methods of secondary structure prediction have been reported (Table 2.6). The approaches taken fall into two main categories ... 

Table 2.6 Some secondary structure predictive methods currently used. Refer to text for further details...

SECONDARY STRUCTURE PREDICTION OF NA+/CL--COUPLED NEUROTRANSMITTER TRANSPORTERS... 

If the /3-rich conformation of outer membrane proteins is really the determinant of their localization, the prediction system of protein localization should evaluate the possibility of an input protein being the [3 type. Fortunately, this appears easier than ordinary secondary structure prediction of globular proteins. Several authors have proposed prediction methods. Here, a method that is conceptually simple and two other recently published methods are briefly described. 

In this respect, the CUE domain is not a isolated case. There are a number of other domain families, each of them only defined in the bioinformatical sense, that have significant matches within established UBA or CUE domain regions. Based on this similarity and on secondary-structure predictions, it can be expected that all of those domain types assume the typical UBA-like three-helix bundle fold. However, it is not clear if all of those domains also bind to ubiquitin, or if they have evolved to different binding properties. Many of the UBA-like domain classes are unpublished. Nevertheless, they should be briefly discussed here, as they are a logical extension of the UBA/CUE paradigm. 

---