运用PDB中的同源信息提高NetTurnP的蛋白质β转角预测精度

β转角作为一种蛋白质二级结构类型在蛋白质折叠、蛋白质稳定性、分子识别等方面具有重要作用．现有的β转角预测方法,没有将PDB等结构数据库中先前存在的同源序列的结构信息映射到待预测的蛋白质序列上．PDB存储的结构已超过70 000,因此对一条新确定的序列,有较大可能性从PDB中找到其同源序列．本文融合PDB中提取的同源结构信息(对每一待测序列,仅使用先于该序列存储于PDB中的同源信息)与NetTurnP预测,提出了一种新的β转角预测方法BTMapping,在经典的BT426数据集和本文构建的数据集EVA937上,以马修斯相关系数表示的预测精度分别为0.56、0.52,而仅使用NetTurnP的为0.50、0.46,以Q_total表示的预测精度分别为81.4%、80.4%,而仅使用NetTurnP的为78.2%、77.3%．结果证实同源结构信息结合先进的β转角预测器如NetTurnP有助于改进β转角识别．BTMapping程序及相关数据集可从http://www.bio530.weebly.com获得．     

Type I and II β-turns prediction using NMR chemical shifts

A method for predicting type I and II β-turns using nuclear magnetic resonance (NMR) chemical shifts is proposed. Isolated β-turn chemical-shift data were collected from 1,798 protein chains. One-dimensional statistical analyses on chemical-shift data of three classes β-turn (type I, II, and VIII) showed different distributions at four positions, (i) to (i + 3). Considering the central two residues of type I β-turns, the mean values of C_ο, C_α, H_N, and N_H chemical shifts were generally (i + 1) > (i + 2). The mean values of C_β and H_α chemical shifts were (i + 1) < (i + 2). The distributions of the central two residues in type II and VIII β-turns were also distinguishable by trends of chemical shift values. Two-dimensional cluster analyses on chemical-shift data show positional distributions more clearly. Based on these propensities of chemical shift classified as a function of position, rules were derived using scoring matrices for four consecutive residues to predict type I and II β-turns. The proposed method achieves an overall prediction accuracy of 83.2 and 84.2 % with the Matthews correlation coefficient values of 0.317 and 0.632 for type I and II β-turns, indicating that its higher accuracy for type II turn prediction. The results show that it is feasible to use NMR chemical shifts to predict the β-turn types in proteins. The proposed method can be incorporated into other chemical-shift based protein secondary structure prediction methods.     

Using predicted shape string to enhance the accuracy of <Emphasis Type="Bold">γ</Emphasis>-turn prediction

Numerous methods for predicting γ-turns in proteins have been developed. However, the results they generally provided are not very good, with a Matthews correlation coefficient (MCC) ≤0.18. Here, an attempt has been made to develop a method to improve the accuracy of γ-turn prediction. First, we employ the geometric mean metric as optimal criterion to evaluate the performance of support vector machine for the highly imbalanced γ-turn dataset. This metric tries to maximize both the sensitivity and the specificity while keeping them balanced. Second, a predictor to generate protein shape string by structure alignment against the protein structure database has been designed and the predicted shape string is introduced as new variable for γ-turn prediction. Based on this perception, we have developed a new method for γ-turn prediction. After training and testing the benchmark dataset of 320 non-homologous protein chains using a fivefold cross-validation technique, the present method achieves excellent performance. The overall prediction accuracy Q _total can achieve 92.2% and the MCC is 0.38, which outperform the existing γ-turn prediction methods. Our results indicate that the protein shape string is useful for predicting protein tight turns and it is reasonable to use the dihedral angle information as a variable for machine learning to predict protein folding. The dataset used in this work and the software to generate predicted shape string from structure database can be obtained from anonymous ftp site freely.     

Classification and Prediction of β-Turn Types

Although a β-turn consists of only four amino acids, it assumes many different types in proteins. Is this basically dependent on the tetrapeptide sequence alone or is it due to a variety of interactions with the other part of a protein? To answer this question, a residue-coupled model is proposed that can reflect the sequence-coupling effect for a tetrapeptide in not only a β-turn or non-β-turn, but also different types of a β-turn. The predicted results by the model for 6022 tetrapeptides indicate that the rates of correct prediction for β-turn types I, I′, II, II′, VI, and VIII and non-β-turns are 68.54%, 93.60%, 85.19%, 97.75%, 100%, 88.75%, and 61.02%, respectively. Each of these seven rates is significantly higher than $\frac{1}{7}$ = 14.29%, the completely randomized rate, implying that the formation of different β-turn types or non-β-turns is considerably correlated with the sequences of a tetrapeptide.     

Predicting beta-turns in proteins using support vector machines with fractional polynomials

Background

β-turns are secondary structure type that have essential role in molecular recognition, protein folding, and stability. They are found to be the most common type of non-repetitive structures since 25% of amino acids in protein structures are situated on them. Their prediction is considered to be one of the crucial problems in bioinformatics and molecular biology, which can provide valuable insights and inputs for the fold recognition and drug design.

Results

We propose an approach that combines support vector machines (SVMs) and logistic regression (LR) in a hybrid prediction method, which we call (H-SVM-LR) to predict β-turns in proteins. Fractional polynomials are used for LR modeling. We utilize position specific scoring matrices (PSSMs) and predicted secondary structure (PSS) as features. Our simulation studies show that H-SVM-LR achieves Qtotal of 82.87%, 82.84%, and 82.32% on the BT426, BT547, and BT823 datasets respectively. These values are the highest among other β-turns prediction methods that are based on PSSMs and secondary structure information. H-SVM-LR also achieves favorable performance in predicting β-turns as measured by the Matthew's correlation coefficient (MCC) on these datasets. Furthermore, H-SVM-LR shows good performance when considering shape strings as additional features.

Conclusions

In this paper, we present a comprehensive approach for β-turns prediction. Experiments show that our proposed approach achieves better performance compared to other competing prediction methods.

     

Prediction of β-Turns

Kohonen's self-organization model, a neural network model, is applied to predict the β-turns in proteins. There are 455 β-turn tetrapeptides and 3807 non-β-turn tetrapeptides in the training database. The rates of correct prediction for the 110 β-turn tetrapeptides and 30,229 non-β-turn tetrapeptides in the testing database are 81.8% and 90.7%, respectively. The high quality of prediction of neural network model implies that the residue-coupled effect along a polypeptide chain is important for the formation of reversal turns, such as β-turns, during the process of protein folding.     

Introduction of a new scheme for classifying β-turns in protein structures

Protein β-turn classification remains an area of ongoing development in structural biology research. While the commonly used nomenclature defining type I, type II and type IV β-turns was introduced in the 1970s and 1980s, refinements of β-turn type definitions have been introduced as recently as 2019 by Dunbrack, Jr and co-workers who expanded the number of β-turn types to 18 (Shapovalov et al, PLOS Computat. Biol., 15, e1006844, 2019). Based on their analysis of 13 030 turns from 1074 ultrahigh resolution (≤1.2 Å) protein structures, they used a new clustering algorithm to expand the definitions used to classify protein β-turns and introduced a new nomenclature system. We recently encountered a specific problem when classifying β-turns in crystal structures of pentapeptide repeat proteins (PRPs) determined in our lab that are largely composed of β-turns that often lie close to, but just outside of, canonical β-turn regions. To address this problem, we devised a new scheme that merges the Klyne-Prelog stereochemistry nomenclature and definitions with the Ramachandran plot. The resulting Klyne-Prelog-modified Ramachandran plot scheme defines 1296 distinct potential β-turn classifications that cover all possible protein β-turn space with a nomenclature that indicates the stereochemistry of i + 1 and i + 2 backbone dihedral angles. The utility of the new classification scheme was illustrated by re-classification of the β-turns in all known protein structures in the PRP superfamily and further assessed using a database of 16 657 high-resolution protein structures (≤1.5 Å) from which 522 776 β-turns were identified and classified.     

Prediction of β-turns in proteins by 1-4 and 2-3 correlation model

A 1–4 and 2–3 residue-correlation model is proposed to predict the β-turns in proteins. The average rate of correct prediction for the 455 β-turn tetrapeptides and 4018 non-β-turn tetrapeptides in the training data base is 80.1%, and that for the 223 β-turn tetrapeptides and 12562 non-β-turn tetrapeptides in the testing data base is 80.9%. Compared with the rates of correct prediction based on the residue-independent model reported previously, the quality of prediction is significantly improved by the new model, implying that the correlation effect between the 1st and the 4th residues and that between the 2nd and 3rd residues along a tetrapeptide are important for forming a β-turn in a protein during the process of its folding. © 1997 John Wiley & Sons, Inc. Biopoly 41: 673–702, 1997     

Prediction of β-turns and β-turn types by a novel bidirectional Elman-type recurrent neural network with multiple output layers (MOLEBRNN)

Prediction of β-turns from amino acid sequences has long been recognized as an important problem in structural bioinformatics due to their frequent occurrence as well as their structural and functional significance. Because various structural features of proteins are intercorrelated, secondary structure information has been often employed as an additional input for machine learning algorithms while predicting β-turns. Here we present a novel bidirectional Elman-type recurrent neural network with multiple output layers (MOLEBRNN) capable of predicting multiple mutually dependent structural motifs and demonstrate its efficiency in recognizing three aspects of protein structure: β-turns, β-turn types, and secondary structure. The advantage of our method compared to other predictors is that it does not require any external input except for sequence profiles because interdependencies between different structural features are taken into account implicitly during the learning process. In a sevenfold cross-validation experiment on a standard test dataset our method exhibits the total prediction accuracy of 77.9% and the Mathew's Correlation Coefficient of 0.45, the highest performance reported so far. It also outperforms other known methods in delineating individual turn types. We demonstrate how simultaneous prediction of multiple targets influences prediction performance on single targets. The MOLEBRNN presented here is a generic method applicable in a variety of research fields where multiple mutually depending target classes need to be predicted. Availability: http://webclu.bio.wzw.tum.de/predator-web/.     

Vibrational analysis of peptides,polypeptides, and proteins. V. Normal vibrations of β-turns

The normal modes have been calculated for β-turns of types I, II, III, I′, II′, and III′. The complete set of frequencies is given for the first three structures; only the amide I, II, and III modes are given for the latter three structures. Calculations have been done for structures with standard dihedral angles, as well as for structures whose dihedral angles differ from these by amounts found in protein structures. The force field was that refined in our previous work on polypeptides. Transition dipole coupling was included, and is crucial to predicting frequency splittings in the amide I and amide II modes. The results show that in the amide I region, β-turn frequencies can overlap with those of the α-helix and β-sheet structures, and therefore caution must be exercised in the interpretation of protein bands in this region. The amide III modes of β-turns are predicted at significantly higher frequencies than those of α-helix and β-sheet structures, and this region therefore provides the best possibility of identifying β-turn structures. Amide V frequencies of β-turns may also be distinctive for such structures.     

Improving the performance of β-turn prediction using predicted shape strings and a two-layer support vector machine model

Background

The β-turn is a secondary protein structure type that plays an important role in protein configuration and function. Development of accurate prediction methods to identify β-turns in protein sequences is valuable. Several methods for β-turn prediction have been developed; however, the prediction quality is still a challenge and there is substantial room for improvement. Innovations of the proposed method focus on discovering effective features, and constructing a new architectural model.     

Improved method for predicting beta-turn using support vector machine

MOTIVATION: Numerous methods for predicting beta-turns in proteins have been developed based on various computational schemes. Here, we introduce a new method of beta-turn prediction that uses the support vector machine (SVM) algorithm together with predicted secondary structure information. Various parameters from the SVM have been adjusted to achieve optimal prediction performance. RESULTS: The SVM method achieved excellent performance as measured by the Matthews correlation coefficient (MCC = 0.45) using a 7-fold cross validation on a database of 426 non-homologous protein chains. To our best knowledge, this MCC value is the highest achieved so far for predicting beta-turn. The overall prediction accuracy Qtotal was 77.3%, which is the best among the existing prediction methods. Among its unique attractive features, the present SVM method avoids overtraining and compresses information and provides a predicted reliability index.     

Prediction of Beta-Turn in Protein Using E-SSpred and Support Vector Machine

β-Turn is a secondary protein structure type that plays an important role in protein configuration and function. Here, we introduced an approach of β-turn prediction that used the support vector machine (SVM) algorithm combined with predicted secondary structure information. The secondary structure information was obtained by using E-SSpred, a new secondary protein structure prediction method. A 7-fold cross validation based on the benchmark dataset of 426 non-homologous protein chains was used to evaluate the performance of our method. The prediction results broke the 80% Q _total barrier and achieved Q _total = 80.9%, MCC = 0.44, and Q _predicted higher 0.9% when compared with the best method. The results in our research are coincident with the conclusion that β-turn prediction accuracy can be improved by inclusion of secondary structure information.     

Structural features determining the site specificity of a rat liver cAMP-independent protein kinase.

The Seryl and Threonyl residues affected in α_s1 and in β-caseins by rat liver “casein kinase TS” (a cytosolic cAMP-independent protein kinase) have been identified. All of them, as well as the residues affected by the same enzyme in α_s2-casein are characterized by an acidic group two residues to their C terminus and by being located within predicted β-turns. Several other potential sites of phosphorylation, according to their primary structure, but located outside predicted β-turns, are not significantly labeled by the protein kinase. It seems conceivable therefore that both a definite aminoacid sequence including a critical acidic residue, and the existence of a β-turn are required for the activity of this protein kinase.     

γ-turn types prediction in proteins using the two-stage hybrid neural discriminant model

Due to the slightly success of protein secondary structure prediction using the various algorithmic and non-algorithmic techniques, similar techniques have been developed for predicting γ-turns in proteins by Kaur and Raghava [2003. A neural-network based method for prediction of γ-turns in proteins from multiple sequence alignment. Protein Sci. 12, 923-929]. However, the major limitation of previous methods was inability in predicting γ-turn types. In a recent investigation we introduced a sequence based predictor model for predicting γ-turn types in proteins [Jahandideh, S., Sabet Sarvestani, A., Abdolmaleki, P., Jahandideh, M., Barfeie, M, 2007a. γ-turn types prediction in proteins using the support vector machines. J. Theor. Biol. 249, 785-790]. In the present work, in order to analyze the effect of sequence and structure in the formation of γ-turn types and predicting γ-turn types in proteins, we applied novel hybrid neural discriminant modeling procedure. As the result, this study clarified the efficiency of using the statistical model preprocessors in determining the effective parameters. Moreover, the optimal structure of neural network can be simplified by a preprocessor in the first stage of hybrid approach, thereby reducing the needed time for neural network training procedure in the second stage and the probability of overfitting occurrence decreased and a high precision and reliability obtained in this way.     

Beta-turns in proteins

The X-ray atomic co-ordinates from 29 proteins of known sequence and structure were utilized to elucidate 459 β-turns in regions of chain reversals. Tetrapeptides whose αC_iαC_{(i + 3)} distances were below 7 Å and not in a helical region were characterized as β-turns. In addition, β-turns were considered to have hydrogen bonding if their computed O_(i)N_{(i + 3)} distances were ≤3.5 Å. The torsion angles of 26 proteins containing 421 β-turns were examined and classified into 11 bend types based on the (φ, ψ) dihedral angles of the i + 1 and i + 2 bend residues. The average frequency of β-turns is 32% as compared to the 38% helices and 20% β-sheets in the 29 proteins. The most frequently occurring bend residues are Asn, Cys, Asp in the first position, Pro, Ser, Lys in the second position, Asn, Asp, Gly in the third position, and Trp, Gly, Tyr in the fourth position. Residues with the highest β-turn potential in all four positions are Pro, Gly, Asn, Asp, and Ser with the most hydrophobic residues (i.e. Val, IIe, and Leu) showing the lowest bend potential. However, in the region just beyond the β-turns, hydrophobic residues occur with greater frequency than do hydrophilic residues. An environmental analysis of β-turn neighboring residues shows that reverse chain folding is stabilized by anti-parallel β-sheets as well as helix-helix and α-β interactions. The β-turn potential at the 12 positions adjacent to and including the bend were plotted for the 20 amino acids and showed dramatic positional preferences, which may be classified according to the nature of the side-chains. An examination of the 27 β-turns in elastase showed that 21 were found in identical positions as those in α-chymotrypsin. However, only 37 of the 84 bend residues were conserved, indicating that structural similarity may persist despite differences in sequence homology. A survey of residues occupying bend types I′, II′ and III′ showed that Gly appeared most frequently in the third position in bend types I′ and III′ as well as in the second position in bend types II′ and III′. Fourteen hydrogenbonded type II bends were found without a Gly at the third position, contrary to the energy calculations. Eight type VI bends with a cis Pro at the third position were also elucidated.     

Effect of Stereochemistry (Z and E) and Position of α,β-Dehydrophenylalanine (ΔPhe) on β-turn Stability

Several model peptides containing α, β-dehydrophenylalanine (ΔPhe) in both Z and E configurations were studied for β-turn stability at the AM1 level of theory. Both configurations of ΔPhe are well able to stabilize β-turns in the backbone. However, the β-turns for peptides bearing Z-ΔPhe are energetically more stable than the E-counterparts. The difference in energies between the global minima of these peptides having the Z and E configuration of ΔPhe, is dictated by the size and stereochemistry of residues flanking ΔPhe. One distinct feature of E-ΔPhe is that it pushes peptides to adopt a Type II β-turn with the ΔPhe residue in the (i + 1) position of the turn. This unique feature may be exploited in peptide design.     

A search for the ideal type I β-turn

In 1968 C. Venkatachalam (Biopolymers, Vol. 6, pp. 1425–1436) predicted the ideal forms of β-turns (type I, type II, etc.) based entirely on theoretical calculations. Subsequently, over a thousand x-ray structures of different globular proteins have been analyzed, with results suggesting that the most important form among the hairpin conformers is the type I β-turn. For the latter type of hairpin conformation, the original computations had predicted ϕ_i+1 = −60°, ψ_i+1 = −30°, ϕ_i+2 = −90°, and ψ_i+2 = 0° as backbone torsion angle values, and these have been used from that time as reference values for the identification of the type I β-turn. However, it has never been clarified whether these “ideal” backbone torsion angle values exist in real structures, or whether these torsion angles are only “theoretical values.” Using the most recent release of the Protein Data Bank (1994), a survey has been made to assign amino acid pairs that approach the ideal form of the type I β-turn. The analysis resulted in four sequences where the deviation from ideal values for any main-chain torsion angles was less than 2°. In order to determine whether such a backbone fold is possible only in proteins owing to fortuitous cooperation of different folding effects, or whether it occurs even in short peptides, various attempts have been made to design the optimal amino acid sequence. Such a peptide model compound adopting precisely the predicted torsion angle values [ϕ_i+1 = −60°, ψ_i+1 = −30°, ϕ_i+2 = −90°, and ψ_i+2 = 0°] could provide valuable information. The solid state conformation of cyclo[(δ) Ava-Gly-Pro-Thr (O¹Bu)-Gly] reported herein, incorporating the -Pro-Thr- subunit, yields values suggesting that the “ideal” type I β-turn is even possible for a peptide where there are no major environmental effects present. © 1996 John Wiley & Sons, Inc.     

Critical amino acids responsible for converting specificities of proteins and for enhancing enzyme evolution are located around β-turn potentials: Data-based prediction

Various reports have described that amino acid substitutions can alter substrate, positional, inhibitory, and target gene specificities of proteins. By using the method of Chou and Fasman, the present work predicts that critical amino acids for converting these specificities are located around β-turns. Residues responsible for the alterations of substrate specificities of trypsin,l-lactate dehydrogenase, aspartate aminotransferase, β-lactamase, and cytochrome P-450 are found to exist within regions predicted as β-turns. The ratios of hydroxylation and oxygenation positions of substrates by cytochrome P-450 and lipoxygenase, respectively, are varied by changes of the protein structures, probably around turn conformations. Inhibitory specificities of bovine pancreatic trypsin inhibitor and α₁-antitrypsin and target gene specificity of glucocorticoid receptor are converted by changing turn structures. Occurrence of β-turn probabilities can be predicted around the amino acid alteration positions of an evolutionally antecedent protein of a nylon degradation enzyme. These findings will have relevance to work on protein engineering and enzyme evolution.     

DNA sequence selectivity of hairpin polyamide turn units

A class of hairpin polyamides linked by 3,4-diaminobutyric acid, resulting in a β-amine residue at the turn unit, showed improved binding affinities relative to their α-amino-γ-turn analogs for particular sequences. We incorporated β-amino-γ-turns in six-ring polyamides and determined whether there are any sequence preferences under the turn unit by quantitative footprinting titrations. Although there was an energetic penalty for G·C and C·G base pairs, we found little preference for T·A over A·T at the β-amino-γ-turn position. Fluorine and hydroxyl substituted α-amino-γ-turns were synthesized for comparison. Their binding affinities and specificities in the context of six-ring polyamides demonstrated overall diminished affinity and no additional specificity at the turn position. We anticipate that this study will be a baseline for further investigation of the turn subunit as a recognition element for the DNA minor groove.