From E‐MAPs to module maps: dissecting quantitative genetic interactions using physical interactions

Recent technological breakthroughs allow the quantification of hundreds of thousands of genetic interactions (GIs) in Saccharomyces cerevisiae. The interpretation of these data is often difficult, but it can be improved by the joint analysis of GIs along with complementary data types. Here, we describe a novel methodology that integrates genetic and physical interaction data. We use our method to identify a collection of functional modules related to chromosomal biology and to investigate the relations among them. We show how the resulting map of modules provides clues for the elucidation of function both at the level of individual genes and at the level of functional modules.
Extended synopsis High‐throughput profiling technologies, providing quantitative measures on a genome‐wide scale, are revolutionizing our ability to study molecular cell biology. These experiments also pose immense analytical challenges, because of the amount of measurements produced, and the inherent noise of the cellular and the experimental systems. Consequently, there is a wide gap today between the potential of the data and the amount of useful biological insights that are derived from them. Closing this gap requires dedicated computational tools, taking into account the intrinsic properties of the performed experiments and combining their results with information from other sources. Here, we focus on the analysis of large‐scale genetic interaction (GI) data, utilizing physical protein interaction networks. A pair of genes x and y is said to exhibit a GI if the phenotype of the double deletion strain Δx,y is different from the phenotype expected given the phenotypes of the single mutants Δx and Δy. The measured phenotype is usually the growth rate (or fitness) of each strain. GIs can be classified into alleviating or aggravating interactions (Schuldiner et al, 2005; Segre et al, 2005; St Onge et al, 2007). In an aggravating interaction, the fitness of the double mutant is lower than expected based on the fitness of the singe gene deletion strains. The extreme case of an aggravating interaction is synthetic lethality, where each single mutant is viable but the double mutant is not. In an alleviating interaction, on the other hand, the double mutant is healthier than expected. The nature of the GI between a pair of genes reflects the relationship between their cellular roles. An aggravating GI usually implies that the genes have parallel roles, where the deletion of one gene is compensated for by the retained activity of the other. Pathways with parallel or complementing roles are thus expected to exhibit a dense pattern of aggravating interactions between their genes. An alleviating GI is expected to be observed between genes that act as part of the same cellular machinery, where the deletion of one or both genes has essentially the same effect on fitness. Finally, genes with a similar function are expected to exhibit similar interactions with the rest of the genes, making the similarity of GI profiles a useful clue for function prediction (Schuldiner et al, 2005; Ye et al, 2005; Collins et al, 2007a, 2007b). Hence, large‐scale quantitative mapping of GIs can help in revealing functional relationships both on the level of individual genes and on the level of functional modules. Recent technological developments have greatly expanded the ability to map GIs in the yeast Saccharomyces cerevisiae. The Epistatic MiniArray Profiles (E‐MAP) method (Schuldiner et al, 2005) is capable of reliably and efficiently quantifying a large number of GIs. The largest E‐MAP experiment published to date contains over 180 000 measurements. To date, E‐MAPs were analyzed using hierarchical clustering followed by visual inspection, limiting the utilization of this immense resource. We describe a novel computational method that exploits a physical interaction (PI) network and enables a deeper understanding of the biology underlying the observed GIs. Our approach uses the genetic data alongside a PI network and identifies functional modules, groups of genes that act in concert to perform a common cellular function. The combined analysis of PI and GI networks is appealing, as they provide complementary information on coordinated activity and functional similarity of gene pairs (Beyer et al, 2007). Kelley and Ideker (2005) pioneered the integration of the genetic and PI data for ‘explanation’ of GIs. This study has shown that most of the aggravating GIs connect proteins found in different modules in the physical network. We have subsequently extended their approach and showed how additional insights can be gleaned from the integrative analysis (Ulitsky and Shamir, 2007a, 2007b). However, the focus of both works was limited to analysis of module pairs, which were frequently overlapping, and relations between pairs were not explicitly addressed. Our new approach is holistic and exploits the quantitative nature of the E‐MAP data. Instead of seeking a single pathway pair at a time, we seek to identify a partition of the genes into disjoint modules, such that each module is a connected physical subnetwork with a coherent GI profile. Moreover, our method explicitly accounts for the relationships between modules. Pairs of modules in the partition are classified as either complementing or neutral, based on the nature of the GIs between them. Our approach has two main merits over previous local methods: (a) it is capable of identifying weaker signals and (b) it provides a global view of the relationships between the modules. In the article, we first show that the modules obtained using our approach compare favorably to existing alternatives in terms of functional enrichment. We then use our method to thoroughly dissect the chromosome biology E‐MAP (Collins et al, 2007a, 2007b) encompassing 186 500 interactions between 754 genes, and identify 62 disjoint modules. The vast majority of the modules show high match to known protein complexes or biological processes. We use this observation to predict the function of four uncharacterized genes. The GIs we used were measured in optimal growth conditions. Using published phenotype data, we verified that the modules remain cohesive functional units in other conditions. Finally, our analysis shows that the connection between genetic complementarity and functional similarity is much more significant on the module level than on the gene level. Following this observation, we study in detail three pairs of modules that exhibit a significant aggravating interaction pattern and that we predict to be involved in similar functions. As quantitative GI mapping technologies become a widely used tool for dissection of complex pathways in S. cerevisiae and in other organisms, it is expected that multiple maps will become available in the near future, and that eventually genome‐wide GI data will be available for many model organisms. Our methodology provides a basis for analysis of future GI data and can also serve as a basis for further improvement of network integration methods.

Object Details

• Location: Deutsche Nationalbibliothek Frankfurt am Main
• Extent: Online-Ressource
• Language: Englisch

References and Relationships

• Bibliographic citation: From E‐MAPs to module maps: dissecting quantitative genetic interactions using physical interactions ; volume:4 ; number:1 ; year:2008 ; extent:12 ; Molecular systems biology ; 4, Heft 1 (2008) (gesamt 12)

Contributors, Places and Time

• Creator: Ulitsky, Igor ; Shlomi, Tomer ; Kupiec, Martin ; Shamir, Ron

Further information

• DOI: 10.1038/msb.2008.42
• URN: urn:nbn:de:101:1-2023073106230538196378
• Rights: Open Access; Der Zugriff auf das Objekt ist unbeschränkt möglich.
• Last update: 14.08.2025, 11:04 AM CEST