Exploring the Monochromatic Landscape in Yeast using Genetic Interactions and Known Processes Reveals the Importance of Protein Complexes

Abstract

Background If perturbing two genes together has a stronger effect than expected based on both single perturbation effects, these two genes are said to genetically interact. Recently experimental methods have enabled to obtain quantitative positive and negative genetic interactions [1] where negative interactions indicate buffering between genes and positive interactions suggest that the genes are part of the same process. In such genetic interaction networks, some groups of genes interact in a monochromatic manner: interactions between them are mostly positive or mostly negative. It has been proposed that these groups are functionally related and enriched in protein complexes [2]. Nevertheless boundaries and system level relationships are still difficult to define from the network. Instead of starting from the network, our new approach starts from known biological processes. We propose to evaluate the current model and study the monochromatic purity of biological processes using the most comprehensive quantitative genetic interaction data set currently available in the budding yeast that includes measurements for 5.4 million pairs of genes [1]. Method We first study the monochromatic purity of biological processes and of the connections between processes. We use Gene Ontology (GO) to define biological processes in yeast. We assess the monochromatic purity of a process with a score based on the relative ratio of positive to negative interactions within that process. We generate random networks and compute zscores to identify statistically significant scores. We perform a statistical test (Fisher) to assess to which extent the connections between different processes are monochromatic. Then we study different features that could explain the monochromaticity (paralogs, protein complexes, ...). Performing the identification of monochromatic purity after removing the features indicates to which extent this feature is important for the monochromaticity. Results We show that 10% of the biological processes are monochromatic and identify 1% of the connections between processes as monochromatic. Among the different features we studied to explain this monochromaticity, protein complexes have a strong effect. In addition we show that 63% of the interactions are attributed to complexes whereas we expect only 49% given the number of genes present in complexes. Conclusion This work is the first systematic study of the monochromatic purity of biological processes and connections between them. It reveals that protein complexes play a disproportionately important role in the yeast genetic landscape. This result gives new insight into the organization of biological processes and can be used for complex member prediction.