Generalized Centroid Estimators in Bioinformatics

Michiaki Hamada,Kiyoshi Asai,Wataru Iwasaki,Hisanori Kiryu,Raya Khanin

doi:10.1371/journal.pone.0016450

Abstract

In a number of estimation problems in bioinformatics, accuracy measures of the target problem are usually given, and it is important to design estimators that are suitable to those accuracy measures. However, there is often a discrepancy between an employed estimator and a given accuracy measure of the problem. In this study, we introduce a general class of efficient estimators for estimation problems on high-dimensional binary spaces, which represent many fundamental problems in bioinformatics. Theoretical analysis reveals that the proposed estimators generally fit with commonly-used accuracy measures (e.g. sensitivity, PPV, MCC and F-score) as well as it can be computed efficiently in many cases, and cover a wide range of problems in bioinformatics from the viewpoint of the principle of maximum expected accuracy (MEA). It is also shown that some important algorithms in bioinformatics can be interpreted in a unified manner. Not only the concept presented in this paper gives a useful framework to design MEA-based estimators but also it is highly extendable and sheds new light on many problems in bioinformatics.

Highlights

In estimation problems in bioinformatics, the space of solutions is generally large and often high-dimensional
A number of fundamental problems in bioinformatics, such as alignment of biological sequences, prediction of secondary structures of RNA sequences, prediction of biological networks, and estimation of phylogenetic trees, are classified into estimation problems whose solutions are in a high-dimensional binary space
In order to make the ccentroid estimator applicable to other estimation problems, we introduce an abstract form of the c-centroid estimator, which is defined on general binary spaces and designed to fit to the commonly used accuracy measures

Summary

Introduction

In estimation problems in bioinformatics, the space of solutions is generally large and often high-dimensional. A number of fundamental problems in bioinformatics, such as alignment of biological sequences, prediction of secondary structures of RNA sequences, prediction of biological networks, and estimation of phylogenetic trees, are classified into estimation problems whose solutions are in a high-dimensional binary space. Such problems are generally difficult to solve, and the estimates are often unreliable. The popular solutions for these problems, such as for the secondary structure of RNA with minimum free energy, are the maximum likelihood (ML) estimators. Noticing the drawbacks of the ML estimators, Carvalho and Lawrence have proposed the centroid estimator, which represents an ensemble of all the possible solutions and minimizes the expected Hamming loss of the prediction [1]

Methods

Results

Discussion

Conclusion