Graphical models for protein function and structure prediction

In this chapter, the authors review bioinformatics applications of two emerging graphical models, the Markov random field (MRF) and the conditional random field (CRF). The main advantage of these two methods is that they can represent dependencies of variables using graphs. Since many biological data can be described as graphs, both methods have gained increasing attention in the bioinformatics community. They first briefly describe the MRF and the CRF in comparison with the hidden Markov model (HMM). What follows are applications of the two graphical models, focusing on gene prediction, protein function prediction, and protein structure prediction. These applications benefit from the graphical models by being able to represent dependencies between graph nodes, which contributed to improvement of prediction accuracy. They discuss some applications of the MRF and CRF on gene prediction, protein function prediction, and protein structure prediction.