Prediction of conformational B-cell epitopes from 3D structures by random forests with a distance-based feature

Zhang, Wen; Xiong, Yi; Zhao, Meng; Zou, Hua; Ye, Xinghuo; Liu, Juan

doi:10.1186/1471-2105-12-341

Cited by 93 publications

(75 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We selected four conformational and four linear epitope predictors as our base learners. The conformational predictors were DiscoTope 2.0 [9], ElliPro [10], SEPPA 2.0 [11], and Bpredictor [18], and the linear epitope predictors were BepiPred [5], ABCpred [6], AAP [7], and BCPREDS [8]. We calculated the Pearson’s correlation coefficients for the prediction scores produced by the base prediction tools.…”

Section: Resultsmentioning

confidence: 99%

See 1 more Smart Citation

A meta-learning approach for B-cell conformational epitope prediction

Lin

et al. 2014

BMC Bioinformatics

View full text Add to dashboard Cite

BackgroundOne of the major challenges in the field of vaccine design is identifying B-cell epitopes in continuously evolving viruses. Various tools have been developed to predict linear or conformational epitopes, each relying on different physicochemical properties and adopting distinct search strategies. We propose a meta-learning approach for epitope prediction based on stacked and cascade generalizations. Through meta learning, we expect a meta learner to be able integrate multiple prediction models, and outperform the single best-performing model. The objective of this study is twofold: (1) to analyze the complementary predictive strengths in different prediction tools, and (2) to introduce a generic computational model to exploit the synergy among various prediction tools. Our primary goal is not to develop any particular classifier for B-cell epitope prediction, but to advocate the feasibility of meta learning to epitope prediction. With the flexibility of meta learning, the researcher can construct various meta classification hierarchies that are applicable to epitope prediction in different protein domains.ResultsWe developed the hierarchical meta-learning architectures based on stacked and cascade generalizations. The bottom level of the hierarchy consisted of four conformational and four linear epitope prediction tools that served as the base learners. To perform consistent and unbiased comparisons, we tested the meta-learning method on an independent set of antigen proteins that were not used previously to train the base epitope prediction tools. In addition, we conducted correlation and ablation studies of the base learners in the meta-learning model. Low correlation among the predictions of the base learners suggested that the eight base learners had complementary predictive capabilities. The ablation analysis indicated that the eight base learners differentially interacted and contributed to the final meta model. The results of the independent test demonstrated that the meta-learning approach markedly outperformed the single best-performing epitope predictor.ConclusionsComputational B-cell epitope prediction tools exhibit several differences that affect their performances when predicting epitopic regions in protein antigens. The proposed meta-learning approach for epitope prediction combines multiple prediction tools by integrating their complementary predictive strengths. Our experimental results demonstrate the superior performance of the combined approach in comparison with single epitope predictors.Electronic supplementary materialThe online version of this article (doi:10.1186/s12859-014-0378-y) contains supplementary material, which is available to authorized users.

show abstract

Section: Resultsmentioning

confidence: 99%

“…We conducted the experiments on several representative epitope predictors released in 2008–2014: SEPPA 2.0 (2014), DiscoTope 2.0 (2012), Bpredictor (2011), CBTOPE (2010), and ElliPro (2008). Each of them had been trained and tested by different data sets [9–11,18,24]. In each experiment, we selected one epitope predictor for comparison.…”

Section: Resultsmentioning

confidence: 99%

A meta-learning approach for B-cell conformational epitope prediction

Lin

et al. 2014

BMC Bioinformatics

View full text Add to dashboard Cite

show abstract

“…Zhang and coworkers (31) recently reported an improved method of predicting conformational epitopes based on a "thick surface patch" model including an "adjacent residue distance" feature. This concept of an epitope includes the idea of CRs layered on top of supporting amino acids but fails to appreciate that the structure of the immediately underlying amino acids is, in turn, dependent on even more amino acids comprising the complete protein domain.…”

Section: Discussionmentioning

confidence: 99%

Antibody and Antigen Contact Residues Define Epitope and Paratope Size and Structure

Stave¹,

Lindpaintner²

2013

The Journal of Immunology

View full text Add to dashboard Cite

A total of 111 Ag–Ab x-ray crystal structures of large protein Ag epitopes and paratopes were analyzed to inform the process of eliciting or selecting functional and therapeutic Abs. These analyses illustrate that Ab contact residues (CR) are distributed in three prominent CR regions (CRR) on L and H chains that overlap but do not coincide with Ab CDR. The number of Ag and Ab CRs per structure are overlapping and centered around 18 and 19, respectively. The CR span (CRS), a novel measure introduced in this article, is defined as the minimum contiguous amino acid sequence containing all CRs of an Ag or Ab and represents the size of a complete structural epitope or paratope, inclusive of CR and the minimum set of supporting residues required for proper conformation. The most frequent size of epitope CRS is 50–79 aa, which is similar in size to L (60–69) and H chain (70–79) CRS. The size distribution of epitope CRS analyzed in this study ranges from ∼20 to 400 aa, similar to the distribution of independent protein domain sizes reported in the literature. Together, the number of CRs and the size of the CRS demonstrate that, on average, complete structural epitopes and paratopes are equal in size to each other and similar in size to intact protein domains. Thus, independent protein domains inclusive of biologically relevant sites represent the fundamental structural unit bound by, and useful for eliciting or selecting, functional and therapeutic Abs.

show abstract

“…Random forest (RF) classifiers [48] have been shown to outperform SVM classifiers on several tasks, including DNAbinding site prediction [49], conformational B-cell epitope prediction [50], gene expression profile classification [51], and prediction of RNA-protein interaction partners [52]. We used the WEKA default settings of the RF classifier, except for the number of trees which was set to 50 instead of 10.…”

Section: Classification Methodsmentioning

confidence: 99%

Predicting protective bacterial antigens using random forest classifiers

EL-Manzalawy

Dobbs

Honavar

2012

Proceedings of the ACM Conference on Bioinformatics, Computational Biology and Biomedicine

View full text Add to dashboard Cite

Identifying protective antigens from bacterial pathogens is important for developing vaccines. Most computational methods for predicting protein antigenicity rely on sequence similarity between a query protein sequence and at least one known antigen. Such methods limit our ability to predict novel antigens (i.e., antigens that are not homologous to any known antigen). Therefore, there is an urgent need for alignment-free computational methods for reliable prediction of protective antigens.We evaluated the discriminative power of four different amino acid composition derived feature representations using three classification methods (Logistic Regression, Support Vector Machine, and Random Forest) on a cross validation data set of 193 protective bacterial antigens and 193 non-antigenic bacterial proteins. Our results show that, with all four data representations, Random Forest classifiers consistently outperform other classifiers. We compared HRF50, one of the best performing Random Forest classifiers with VaxiJen and SignalP on independent test sets derived from the Chlamydia trachomatis and Bartonella proteomes. Our results show that our HRF50 predictor outperforms VaxiJen and is competitive with SignalP and ANTIGENpro in predicting protective antigens. We further showed that when we combine SignalP with HRF50, the resulting method, which we call BacGen, yields performance that is comparable to or better than that of ANTIGENpro in predicting antigens in bacterial sequences. We conclude that amino acid sequence composition derived features can be effectively used to design alignment-free methods for predicting protein antigenicity using Random Forest classifiers. BacGen is available as an online server at

show abstract

Prediction of conformational B-cell epitopes from 3D structures by random forests with a distance-based feature

Cited by 93 publications

References 42 publications

A meta-learning approach for B-cell conformational epitope prediction

A meta-learning approach for B-cell conformational epitope prediction

Antibody and Antigen Contact Residues Define Epitope and Paratope Size and Structure

Predicting protective bacterial antigens using random forest classifiers

Contact Info

Product

Resources

About