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

BACKGROUND: One 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. RESULTS: We 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. CONCLUSIONS: Computational 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.

Reconstruction of the Temperature Index Series of China in 1368-1911 based on REACHES database.

This study reports the methodology for reconstructing anomalous temperature index series of China in 1368-1911 based on the REACHES database which digitizes the Chinese records quoted in the Compendium of Meteorological Records of China in the Last 3000 Years. The reconstruction adopts an ordinal scale index approach ranging from -2 (very cold) to 1 (warm). Based on the grading criteria, a total of 12,871 records were retrieved through a standard coding system established at REACHES. Sensitivity experiments were performed to test robustness of the index system and a reasonability test was conducted to develop an appropriate method for deriving areal mean temperature index. The reconstructed series were validated through comparison with early instrumental data from Global Historical Climatology Network which shows good correlations and reliability of the REACHES reconstructed index data. Annual and semi-annual (winter and summer) temperature index data series were released for the whole domain as well as the 3- and 15-subregion geographical domains in China.

Historical weather data for climate risk assessment.

Weather- and climate-related hazards are responsible for monetary losses, material damages, and societal consequences. Quantifying related risks is, therefore, an important societal task, particularly in view of future climate change. For this task, climate risk assessment increasingly uses model chains, which mainly build on data from the last few decades. The past record of events could play a role in this context. New numerical techniques can make use of historical weather data to simulate impacts quantitatively. However, using historical data for model applications differs from using recent products. Here, we provide an overview of climate risk assessment methodologies and of the properties of historical instrumental and documentary data. Using three examples, we then outline how historical environmental data can be used today in climate risk assessment by (1) developing and validating numerical model chains, (2) providing a large statistical sample which can be directly exploited to estimate hazards and to model present risks, and (3) establishing "worst-case" events which are relevant references in the present or future. The examples show that, in order to be successful, different sources (reanalyses, digitized instrumental data, and documentary data) and methods (dynamical downscaling and analog methods) need to be combined on a case-by-case basis.

Construction of the REACHES climate database based on historical documents of China.

This paper describes the methodology of an ongoing project of constructing an East Asian climate database REACHES based on Chinese historical documents. The record source is Compendium of Meteorological Records of China in the Last 3000 Years which collects meteorology and climate related records from mainly official and local chronicles along with a small number of other documents. We report the digitization of the records covering the period 1644-1795. An example of the original records is translated to illustrate the typical contents which contain time, location and type of events. Chinese historical times and location names are converted into Gregorian calendar and latitudes and longitudes. A hierarchical database system is developed that consists of the hierarchies of domains, main categories, subcategories, and further details. Historical events are then digitized and categorized into such a system. Code systems are developed at all levels such that the original descriptive entries are converted into digitized records suitable for treatment by computers. Statistics and characteristics of the digitized records in the database are described.