An Ensemble Method for Predicting Subnuclear Localizations from Primary Protein Structures

Abstract

BackgroundPredicting protein subnuclear localization is a challenging problem. Some previous works based on non-sequence information including Gene Ontology annotations and kernel fusion have respective limitations. The aim of this work is twofold: one is to propose a novel individual feature extraction method; another is to develop an ensemble method to improve prediction performance using comprehensive information represented in the form of high dimensional feature vector obtained by 11 feature extraction methods.Methodology/Principal FindingsA novel two-stage multiclass support vector machine is proposed to predict protein subnuclear localizations. It only considers those feature extraction methods based on amino acid classifications and physicochemical properties. In order to speed up our system, an automatic search method for the kernel parameter is used. The prediction performance of our method is evaluated on four datasets: Lei dataset, multi-localization dataset, SNL9 dataset and a new independent dataset. The overall accuracy of prediction for 6 localizations on Lei dataset is 75.2% and that for 9 localizations on SNL9 dataset is 72.1% in the leave-one-out cross validation, 71.7% for the multi-localization dataset and 69.8% for the new independent dataset, respectively. Comparisons with those existing methods show that our method performs better for both single-localization and multi-localization proteins and achieves more balanced sensitivities and specificities on large-size and small-size subcellular localizations. The overall accuracy improvements are 4.0% and 4.7% for single-localization proteins and 6.5% for multi-localization proteins. The reliability and stability of our classification model are further confirmed by permutation analysis.ConclusionsIt can be concluded that our method is effective and valuable for predicting protein subnuclear localizations. A web server has been designed to implement the proposed method. It is freely available at http://bioinformatics.awowshop.com/snlpred_page.php.

A new method for identification of pre-microRNAs based on hybrid features

“…The empirical mode decomposition is a time-frequency analysis and was originally proposed by Huang et al [25] for the study of ocean waves. The EMD method has been used by our group to simulate geomagnetic field data [45] and to predict protein subnuclear localization [46]. In EMD, the base functions, which are called intrinsic mode functions (IMFs), are obtained adaptively from the original signal.…”

“…In EMD, the base functions, which are called intrinsic mode functions (IMFs), are obtained adaptively from the original signal. The principle and details of Hilbert spectral analysis can be found in [25,46]. Combining the sequences of the pre-microRNAs and the physical and chemical characteristics of the dinucleotides, the feature extraction method based on the Hilbert-Huang transform is described as follows:…”

Progress in Machine Learning-based Predicting Subcellular Localizations of Proteins with Multiple Sites

“…Freund and Schapire proposed the famous AdaBoost algorithm, and Breiman proposed the Bagging algorithm, respectively. Despite the successes have been gained [26], the ensemble learning is still a very hot topic nowadays, and many problems are not solved successfully yet. There is a huge study space needs to research deeply.…”

Characterization of the nuclear matrix targeting sequence (NMTS) of the BPV1 E8/E2 protein — the shortest known NMTS

“…To date, no clear rules or determinants predicting NM targeting have been published; predictions of NM localization have not been very successful. 7 Disruption of NM targeting often leads to changes in protein function, highlighting the relationship between localization and function. 8,9 Alternative transcription and alternative splicing generates alternative protein isoforms and is one factor beside others that regulate the subcellular localization and subnuclear compartmentalisation of proteins.…”

“…15 However, most subnuclear localization signals or the targeting signals for specific compartments are not well defined and cannot be predicted with high confidence. 7 There are increasing list of changes in protein localization due to alternative splicing. 16 Alternative splicing has been shown to affect sub-cellular localization and the subnuclear compartmentalisation of viral 17 and cellular proteins (ING4 18,19 and WT1 20 ); some of these proteins are related to diseases, such as CIZ1.…”