Probabilistic models widely used for describing features of a protein sequence. Hidden Markov Models introduce a “regular grammar” that characterizes a set of biological sequences. These are generative models, which renders them highly applicable in biological sequence analysis. In general, a HMM is composed of a set of states that form a first order Markovian process, connected by means of the transition probabilities. Each state, has a unique probability distribution for generating (emitting) the symbols of the finite alphabet (nucleotides or amino-acids). The most widely used variant of Hidden Markov Model (HMM) is the profile HMM which models in a probabilistic manner the matches, inserts and deletions occurring in every column of a multiple sequence alignment. However, other variations are also common (i.e. the circular HMM).
Published in Chapter:
Computational Methods for the Prediction of GPCRs Coupling Selectivity
Nikolaos G. Sgourakis (Rensselaer Polytechnic Institute, USA), Pantelis G. Bagos (University of Central Greece, and University of Athens, Greece), and Stavros J. Hamodrakas (University of Athens, Greece)
Copyright: © 2009
|Pages: 15
DOI: 10.4018/978-1-60566-076-9.ch009
Abstract
GPCRs comprise a wide and diverse class of eukaryotic transmembrane proteins with well-established pharmacological significance. As a consequence of recent genome projects, there is a wealth of information at the sequence level that lacks any functional annotation. These receptors, often quoted as orphan GPCRs, could potentially lead to novel drug targets. However, typical experiments that aim at elucidating their function are hampered by the lack of knowledge on their selective coupling partners at the interior of the cell, the G-proteins. Up-to-date, computational efforts to predict properties of GPCRs have been focused mainly on the ligand-binding specificity, while the aspect of coupling has been less studied. Here, we present the main motivations, drawbacks, and results from the application of bioinformatics techniques to predict the coupling specificity of GPCRs to G-proteins, and discuss the application of the most successful methods in both experimental works that focus on a single receptor and large-scale genome annotation studies.