E-Book Overview
Comparative Genomics, Volume 2, provides a collection of robust protocols for molecular biologists beginning to use comparative genomic analysis tools in a variety of areas. Volume 2 contains the last three of seven sections. In the second volume, the fifth section describes a number of tools for comparative analysis of domain and gene families. These tools are particularly useful for predicting protein function as well as potential protein-protein interactions. In the sixth section, methods for comparing groups of genes and gene order are discussed, as are several tools for analyzing genome evolution. Finally, the seventh section deals with experimental comparative genomics. This section includes methods for comparing gene copy number across an entire genome, comparative genomic hybridization, SNP analysis, as well as genome-wide mapping and typing systems for bacterial genomes. Given the tremendous increase in available biosequence data over the past ten years, Comparative Genomics, Volume 1, is timely, comprehensive, and novel. A companion Volume 1 is also available from Humana Press.
E-Book Content
1 Computational Prediction of Domain Interactions Philipp Pagel, Normann Strack, Matthias Oesterheld, Volker Stümpflen, and Dmitrij Frishman
Summary Conserved domains carry many of the functional features found in the proteins of an organism. This includes not only catalytic activity, substrate binding, and structural features but also molecular adapters, which mediate the physical interactions between proteins or proteins with other molecules. In addition, two conserved domains can be linked not by physical contact but by a common function like forming a binding pocket. Although a wealth of experimental data has been collected and carefully curated for protein– protein interactions, as of today little useful data is available from major databases with respect to relations on the domain level. This lack of data makes computational prediction of domain– domain interactions a very important endeavor. In this chapter, we discuss the available experimental data (iPfam) and describe some important approaches to the problem of identifying interacting and/or functionally linked domain pairs from different kinds of input data. Specifically, we will discuss phylogenetic profiling on the level of conserved protein domains on one hand and inference of domain-interactions from observed or predicted protein–protein interactions datasets on the other. We explore the predictive power of these predictions and point out the importance of deploying as many different methods as possible for the best results.
Key Words: Conserved domains; function prediction; protein interactions; data integration.
1. Introduction Biological networks have long been an important topic in computational biology and they constitute the focal point of the emerging field of systems biology. Just like metabolic pathways, regulatory or transcriptional networks, From: Methods in Molecular Biology, vol. 396: Comparative Genomics, Volume 2 Edited by: N. H. Bergman © Humana Press Inc., Totowa, NJ
3
4
Pagel et al.
physical interactions (PPI), and functional associations between proteins can be represented in the form of large graphs. In many cases, protein function and, specifically, interactions are carried out by conserved domains, which are reused in different proteins throughout an organism. This modular architecture has been a focus of interest for a long time (1). The problem of formal description, biological annotation, and classification of these domains has been addressed by several groups leading to valuable resources such as SMART (2), BLOCKS (3), Pfam (4), and the recent integration endeavor InterPro (5). Because of their important role as biological modules carrying function and/or mediating ph