Secondary structure prediction with support vector machines

Motivation: A new method that uses support vector machines (SVMs) to predict protein secondary structure is described and evaluated. The study is designed to develop a reliable prediction method using an alternative technique and to investigate the applicability of SVMs to this type of bioinformatics problem. Methods: Binary SVMs are trained to discriminate between two structural classes. The binary classifiers are combined in several ways to predict multi-class secondary structure. Results: The average three-state prediction accuracy per protein (Q3) is estimated by cross-validation to be 77.07 ± 0.26% with a segment overlap (Sov) score of 73.32 ± 0.39%. The SVM performs similarly to the 'state-of-the-art' PSIPRED prediction method on a non-homologous test set of 121 proteins despite being trained on substantially fewer examples. A simple consensus of the SVM, PSIPRED and PROFsec achieves significantly higher prediction accuracy than the individual methods. Availability: The SVM classifier is available from the authors. Work is in progress to make the method available on-line and to integrate the SVM predictions into the PSIPRED server.

Benchmarking secondary structure prediction for fold recognition

If secondary structure predictions are to be incorporated into fold recognition methods, an assessment of the effect of specific types of errors in predicted secondary structures on the sensitivity of fold recognition should be carried out. Here, we present a systematic comparison of different secondary structure prediction methods by measuring frequencies of specific types of error. We carry out an evaluation of the effect of specific types of error on secondary structure element alignment (SSEA), a baseline fold recognition method. The results of this evaluation indicate that missing out whole helix or strand elements, or predicting the wrong type of element, is more detrimental than predicting the wrong lengths of elements or overpredicting helix or strand. We also suggest that SSEA scoring is an effective method for assessing accuracy of secondary structure prediction and perhaps may also provide a more appropriate assessment of the “usefulness” and quality of predicted secondary structure, if secondary structure alignments are to be used in fold recognition.

Rapid protein domain assignment from amino acid sequence using predicted secondary structure

The elucidation of the domain content of a given protein sequence in the absence of determined structure or significant sequence homology to known domains is an important problem in structural biology. Here we address how successfully the delineation of continuous domains can be accomplished in the absence of sequence homology using simple baseline methods, an existing prediction algorithm (Domain Guess by Size), and a newly developed method (DomSSEA). The study was undertaken with a view to measuring the usefulness of these prediction methods in terms of their application to fully automatic domain assignment. Thus, the sensitivity of each domain assignment method was measured by calculating the number of correctly assigned top scoring predictions. We have implemented a new continuous domain identification method using the alignment of predicted secondary structures of target sequences against observed secondary structures of chains with known domain boundaries as assigned by Class Architecture Topology Homology (CATH). Taking top predictions only, the success rate of the method in correctly assigning domain number to the representative chain set is 73.3%. The top prediction for domain number and location of domain boundaries was correct for 24% of the multidomain set (±20 residues). These results have been put into context in relation to the results obtained from the other prediction methods assessed

Secondary structure analysis of the dissimilatory sulphite reductase in Desulfovibrio desulfuricans

The complete sequences of the dsrA and dsrB genes coding for the α− and β−subunits, respectively, of the sulphite reductase enzyme in Desulfovibrio desulfuricans were determined. Analyses of the amino acid sequences indicated a number of serohaem/Fe4S4 binding consensus sequences whilst predictive secondary structure analysis revealed a similar pattern of α−helix and β−strand structures between the two subunits which was indicative of gene duplication.

Hydrogelation and self-assembly of Fmoc-tripeptides: unexpected influence of sequence on self-assembled fibril structure, and hydrogel modulus and anisotropy

The self-assembly and hydrogelation properties of two Fmoc-tripeptides [Fmoc = N-(fluorenyl-9-methoxycarbonyl)] are investigated, in borate buffer and other basic solutions. A remarkable difference in self-assembly properties is observed comparing Fmoc-VLK(Boc) with Fmoc-K(Boc)LV, both containing K protected by N(epsilon)-tert-butyloxycarbonate (Boc). In borate buffer, the former peptide forms highly anisotropic fibrils which show local alignment, and the hydrogels show flow-aligning properties. In contrast, Fmoc-K(Boc)LV forms highly branched fibrils that produce isotropic hydrogels with a much higher modulus (G' > 10(4) Pa), and lower concentration for hydrogel formation. The distinct self-assembled structures are ascribed to conformational differences, as revealed by secondary structure probes (CD, FTIR, Raman spectroscopy) and X-ray diffraction. Fmoc-VLK(Boc) forms well-defined beta-sheets with a cross-beta X-ray diffraction pattern, whereas Fmoc-KLV(Boc) forms unoriented assemblies with multiple stacked sheets. Interchange of the K and V residues when inverting the tripeptide sequence thus leads to substantial differences in self-assembled structures, suggesting a promising approach to control hydrogel properties.

Protein structure prediction servers at University College London

A number of state-of-the-art protein structure prediction servers have been developed by researchers working in the Bioinformatics Unit at University College London. The popular PSIPRED server allows users to perform secondary structure prediction, transmembrane topology prediction and protein fold recognition. More recent servers include DISOPRED for the prediction of protein dynamic disorder and DomPred for domain boundary prediction.

The PSIPRED protein structure prediction server

The PSIPRED protein structure prediction server allows users to submit a protein sequence, perform a prediction of their choice and receive the results of the prediction both textually via e-mail and graphically via the web. The user may select one of three prediction methods to apply to their sequence: PSIPRED, a highly accurate secondary structure prediction method; MEMSAT 2, a new version of a widely used transmembrane topology prediction method; or GenTHREADER, a sequence profile based fold recognition method.

Molecular archaeology of Flaviviridae untranslated regions: duplicated RNA structures in the replication enhancer of flaviviruses and pestiviruses emerged via convergent evolution.

RNA secondary structures in the 3'untranslated regions (3'UTR) of the viruses of the family Flaviviridae, previously identified as essential (promoters) or beneficial (enhancers) for replication, have been analysed. Duplicated enhancer elements are revealed as a global feature in the evolution of the 3'UTR of distantly related viruses within the genera Flavivirus and Pestivirus. For the flaviviruses, duplicated structures occur in the 3'UTR of all four distantly related ecological virus subgroups (tick-borne, mosquito-borne, no known vector and insect-specific flaviviruses (ISFV). RNA structural differences distinguish tick-borne flaviviruses with discrete pathogenetic characteristics. For Aedes- and Culex-associated ISFV, secondary RNA structures with different conformations display numerous short ssRNA direct repeats, exposed as loops and bulges. Long quadruplicate regions comprise almost the entire 3'UTR of Culex-associated ISFV. Extended duplicated sequence and associated RNA structures were also discovered in the 3'UTR of pestiviruses. In both the Flavivirus and Pestivirus genera, duplicated RNA structures were localized to the enhancer regions of the 3'UTR suggesting an adaptive role predominantly in wild-type viruses. We propose sequence reiteration might act as a scaffold for dimerization of proteins involved in assembly of viral replicase complexes. Numerous nucleotide repeats exposed as loops/bulges might also interfere with host immune responses acting as a molecular sponge to sequester key host proteins or microRNAs.

Layer-by-layer electrostatic entrapment of protein molecules on superparamagnetic nanoparticle: new strategy to enhance adsorption capacity and maintain biological activity

There is a recent interest to use inorganic-based magnetic nanoparticles as a vehicle to carry biomolecules for various biophysical applications, but direct attachment of the molecules is known to alter their conformation leading to attenuation in activity. In addition, surface immobilization has been limited to monolayer coverage. It is shown that alternate depositions of negatively charged protein molecules, typically bovine serum albumin (BSA) with a positively charged aminocarbohydrate template such as glycol chitosan (GC) on magnetic iron oxide nanoparticle surface as a colloid, are carried out under pH 7.4. Circular dichroism (CD) clearly reveals that the secondary structure of the entrapped BSA sequential depositions in this manner remains totally unaltered which is in sharp contrast to previous attempts. Probing the binding properties of the entrapped BSA using small molecules (Site I and Site II drug compounds) confirms for the first time the full retention of its biological activity as compared with native BSA, which also implies the ready accessibility of the entrapped protein molecules through the porous overlayers. This work clearly suggests a new method to immobilize and store protein molecules beyond monolayer adsorption on a magnetic nanoparticle surface without much structural alteration. This may find applications in magnetic recoverable enzymes or protein delivery.

Direct repeats in the flavivirus 3' untranslated region; a strategy for survival in the environment?

Previously, direct repeats (DRs) of 20-70 nucleotides were identified in the 3' untranslated regions (3'UTR) of flavivirus sequences. To address their functional significance, we have manually generated a pan-flavivirus 3'UTR alignment and correlated it with the corresponding predicted RNA secondary structures. This approach revealed that intra-group-conserved DRs evolved from six long repeated sequences (LRSs) which, as approximately 200-nucleotide domains were preserved only in the genomes of the slowly evolving tick-borne flaviviruses. We propose that short DRs represent the evolutionary remnants of LRSs rather than distinct molecular duplications. The relevance of DRs to virus replication enhancer function, and thus survival, is discussed.

Benchmarking consensus model quality assessment for protein fold recognition

Background: Selecting the highest quality 3D model of a protein structure from a number of alternatives remains an important challenge in the field of structural bioinformatics. Many Model Quality Assessment Programs (MQAPs) have been developed which adopt various strategies in order to tackle this problem, ranging from the so called "true" MQAPs capable of producing a single energy score based on a single model, to methods which rely on structural comparisons of multiple models or additional information from meta-servers. However, it is clear that no current method can separate the highest accuracy models from the lowest consistently. In this paper, a number of the top performing MQAP methods are benchmarked in the context of the potential value that they add to protein fold recognition. Two novel methods are also described: ModSSEA, which based on the alignment of predicted secondary structure elements and ModFOLD which combines several true MQAP methods using an artificial neural network. Results: The ModSSEA method is found to be an effective model quality assessment program for ranking multiple models from many servers, however further accuracy can be gained by using the consensus approach of ModFOLD. The ModFOLD method is shown to significantly outperform the true MQAPs tested and is competitive with methods which make use of clustering or additional information from multiple servers. Several of the true MQAPs are also shown to add value to most individual fold recognition servers by improving model selection, when applied as a post filter in order to re-rank models. Conclusion: MQAPs should be benchmarked appropriately for the practical context in which they are intended to be used. Clustering based methods are the top performing MQAPs where many models are available from many servers; however, they often do not add value to individual fold recognition servers when limited models are available. Conversely, the true MQAP methods tested can often be used as effective post filters for re-ranking few models from individual fold recognition servers and further improvements can be achieved using a consensus of these methods.

Use of a novel Hill-climbing genetic algorithm in protein folding simulations

We have developed a novel Hill-climbing genetic algorithm (GA) for simulation of protein folding. The program (written in C) builds a set of Cartesian points to represent an unfolded polypeptide's backbone. The dihedral angles determining the chain's configuration are stored in an array of chromosome structures that is copied and then mutated. The fitness of the mutated chain's configuration is determined by its radius of gyration. A four-helix bundle was used to optimise simulation conditions, and the program was compared with other, larger, genetic algorithms on a variety of structures. The program ran 50% faster than other GA programs. Overall, tests on 100 non-redundant structures gave comparable results to other genetic algorithms, with the Hill-climbing program running from between 20 and 50% faster. Examples including crambin, cytochrome c, cytochrome B and hemerythrin gave good secondary structure fits with overall alpha carbon atom rms deviations of between 5 and 5.6 Angstrom with an optimised hydrophobic term in the fitness function. (C) 2003 Elsevier Ltd. All rights reserved.

Use of essential and molecular dynamics to study gamma B-crystallin unfolding after non-enzymic post-translational modifications

Essential and Molecular Dynamics (ED/MD) have been used to model the conformational changes of a protein implicated in a conformational disease-cataract, the largest cause of blindness in the world-after non-enzymic post-translational modification. Cyanate modification did not significantly alter flexibility, while the Schiff's base adduct produced a more flexible N-terminal domain, and intra-secondary structure regions, than either the cyanate adduct or the native structure. Glycation also increased linker flexibility and disrupted the charge network. A number of post-translational adducts showed structural disruption around Cys15 and increased linker flexibility; this may be important in subsequent protein aggregation. Our modelling results are in accord with experimental evidence, and show that ED/MD is a useful tool in modelling conformational changes in proteins implicated in disease processes. (C) 2003 Published by Elsevier Ltd.

Self-assembly in aqueous solution of a modified amyloid beta peptide fragment

The self-assembly in films dried from aqueous solutions of a modified amyloid beta peptide fragment is studied. We focus on sequence A beta(16-20), KLVFF, extended by two alanines at the N-terminus to give AAKLVFF. Self-assembly into twisted ribbon fibrils is observed, as confirmed by transmission electron microscopy (TEM). Dynamic light scattering reveals the semi-flexible nature of the AAKLVFF fibrils, while polarized optical microscopy shows that the peptide fibrils crystallize after an aqueous solution of AAKLVFF is matured over 5 days. The secondary structure of the fibrils is studied by FT-IR, circular dichroism and X-ray diffraction (XRD), which provide evidence for beta-sheet structure in the fibril. From high resolution TEM it is concluded that the average width of an AAKLVFF fibril is (63 +/- 18) nm, indicating that these fibrils comprise beta-sheets with multiple repeats of the unit cell, determined by XRD to have b and c dimensions 1.9 and 4.4 nm with an a axis 0.96 nm, corresponding to twice the peptide backbone spacing in the antiparallel beta-sheet. (C) 2008 Elsevier B.V. All rights reserved.

Interactions of KLVFF-PEG peptide conjugate with fibrinogen in neutral aqueous solutions

In this work we report on the interaction of KLVFF-PEG with fibrinogen (Fbg) in neutral aqueous solutions at 20 degrees C, for particular ratios of KLVFF-PEG to Fbg concentration, Delta = CKLVFF-PEG/C-Fbg- Our results show the formation of Fbg/KLVFF-PEG complexes for Delta > 0, such that there is not an extended network of complexes throughout the solution. In addition, cleaved protein and Fbg dimers are identified in the solution for Delta >= 0. There is a dramatic change in the tertiary structure of the Fbg upon KLVFF-PEG binding, although the KLVFF-PEG binds to the Fbg without affecting the secondary structure elements of the glycoprotein.