ACO-Based Bayesian Network Ensembles for the Hierarchical Classification of Ageing-Related Proteins

Abstract

Abstract. The task of predicting protein functions using computational techniques is a major research area in the field of bioinformatics. Casting the task into a classification problem makes it challenging, since the classes (functions) to be predicted are hierarchically related, and a protein can have more than one function. One approach is to produce a set of local classifiers; each is responsible for discriminating between a subset of the classes in a certain level of the hierarchy. In this paper we tackle the hierarchical classification problem in a local fashion, by learning an ensemble of Bayesian network classifiers for each class in the hierarchy and combining their outputs with four alternative methods: a) selecting the best classifier, b) majority voting, c) weighted voting, and d) constructing a meta-classifier. The ensemble is built using ABC-Miner, our recently introduced Ant-based Bayesian Classification algorithm. We use different types of protein representations to learn different classification models. We empirically evaluate our proposed methods on an ageingrelated protein dataset created for this research.

eAnt-Miner : An Ensemble Ant-Miner to Improve the ACO Classification

“…We recommend to apply these new methods to hierarchical data sets. We also recommend to employ the ensemble methods proposed in [16,3] on Ant-Miner variations to further improve the ACO based classification.…”

“…Khalid & Freitas [16] proposed a new ACO based Bayesian network ensembles with a new ABC-Miner [15] as the base classifier, in order to predict the ageing related protein functions in bioinformatics. They identified the importance of human ageing related protein function prediction as a major application of hierarchical classification.…”

A Novel Extended Hierarchical Dependence Network Method Based on Non-hierarchical Predictive Classes and Applications to Ageing-Related Data

“…Salama and Freitas [19] have already compiled an ageingrelated dataset for the hierarchical classification of ageingrelated proteins. We build upon their work by updating and expanding the dataset to contain more species and the features used in [22], which focused on the hierarchical classification of generic (not specifically ageing-related) proteins functions.…”

“…We call these overexpressed GO terms ageing-related GO terms, as they occur significantly more often than statistically expected by chance in our datasets of ageing-related proteins. Numeric Alignment-Independent Features: We extracted the following numeric features described in [19,22]: "Amino Acid Composition", "Composition", "Transition", "Distribution", and "Z-Values". Furthermore, all datasets (with all types of features) have two features: "Sequence Length" (the amino acid sequence length), and "Molecular Weight" (the molecular weight of the protein).…”