Detecting coordinated regulation of multi-protein complexes using logic analysis of gene expression.

BACKGROUND: Many of the functional units in cells are multi-protein complexes such as RNA polymerase, the ribosome, and the proteasome. For such units to work together, one might expect a high level of regulation to enable co-appearance or repression of sets of complexes at the required time. However, this type of coordinated regulation between whole complexes is difficult to detect by existing methods for analyzing mRNA co-expression. We propose a new methodology that is able to detect such higher order relationships. RESULTS: We detect coordinated regulation of multiple protein complexes using logic analysis of gene expression data. Specifically, we identify gene triplets composed of genes whose expression profiles are found to be related by various types of logic functions. In order to focus on complexes, we associate the members of a gene triplet with the distinct protein complexes to which they belong. In this way, we identify complexes related by specific kinds of regulatory relationships. For example, we may find that the transcription of complex C is increased only if the transcription of both complex A AND complex B is repressed. We identify hundreds of examples of coordinated regulation among complexes under various stress conditions. Many of these examples involve the ribosome. Some of our examples have been previously identified in the literature, while others are novel. One notable example is the relationship between the transcription of the ribosome, RNA polymerase and mannosyltransferase II, which is involved in N-linked glycan processing in the Golgi. CONCLUSIONS: The analysis proposed here focuses on relationships among triplets of genes that are not evident when genes are examined in a pairwise fashion as in typical clustering methods. By grouping gene triplets, we are able to decipher coordinated regulation among sets of three complexes. Moreover, using all triplets that involve coordinated regulation with the ribosome, we derive a large network involving this essential cellular complex. In this network we find that all multi-protein complexes that belong to the same functional class are regulated in the same direction as a group (either induced or repressed).

Characterization and prediction of protein-protein interactions within and between complexes.

Databases of experimentally determined protein interactions provide information on binary interactions and on involvement in multiprotein complexes. These data are valuable for understanding the general properties of the interaction between proteins as well as for the development of prediction schemes for unknown interactions. Here we analyze experimentally determined protein interactions by measuring various sequence, genomic, transcriptomic, and proteomic attributes of each interacting pair in the yeast Saccharomyces cerevisiae. We find that dividing the data into two groups, one that includes binary interactions within protein complexes (stable) and another that includes binary interactions that are not within complexes (transient), enables better characterization of the interactions by the different attributes and improves the prediction of new interactions. This analysis revealed that most attributes were more indicative in the set of intracomplex interactions. Using this data set for training, we integrated the different attributes by logistic regression and developed a predictive scheme that distinguishes between interacting and noninteracting protein pairs. Analysis of the logistic-regression model showed that one of the strongest contributors to the discrimination between interacting and noninteracting pairs is the presence of distinct pairs of domain signatures that were suggested previously to characterize interacting proteins. The predictive algorithm succeeds in identifying both intracomplex and other interactions (possibly the more stable ones), and its correct identification rate is 2-fold higher than that of large-scale yeast two-hybrid experiments.

Utilizing logical relationships in genomic data to decipher cellular processes.

The wealth of available genomic data has spawned a corresponding interest in computational methods that can impart biological meaning and context to these experiments. Traditional computational methods have drawn relationships between pairs of proteins or genes based on notions of equality or similarity between their patterns of occurrence or behavior. For example, two genes displaying similar variation in expression, over a number of experiments, may be predicted to be functionally related. We have introduced a natural extension of these approaches, instead identifying logical relationships involving triplets of proteins. Triplets provide for various discrete kinds of logic relationships, leading to detailed inferences about biological associations. For instance, a protein C might be encoded within an organism if, and only if, two other proteins A and B are also both encoded within the organism, thus suggesting that gene C is functionally related to genes A and B. The method has been applied fruitfully to both phylogenetic and microarray expression data, and has been used to associate logical combinations of protein activity with disease state phenotypes, revealing previously unknown ternary relationships among proteins, and illustrating the inherent complexities that arise in biological data.

How reliable are experimental protein-protein interaction data?

Data of protein-protein interactions provide valuable insight into the molecular networks underlying a living cell. However, their accuracy is often questioned, calling for a rigorous assessment of their reliability. The computation offered here provides an intelligible mean to assess directly the rate of true positives in a data set of experimentally determined interacting protein pairs. We show that the reliability of high-throughput yeast two-hybrid assays is about 50%, and that the size of the yeast interactome is estimated to be 10,000-16,600 interactions.