Abstract
Modular structures in biological networks are ubiquitous and well-described, yet this organization does not capture the complexity of genes individually influencing many modules. Pleiotropy, the phenomenon of a single genetic locus with multiple phenotypic effects, has previously been measured according to many definitions, which typically count phenotypes associated with genes. We take the perspective that, because genes work in complex and interconnected modules, pleiotropy can be treated as a network-derived characteristic. Here, we use the complete network of yeast genetic interactions (GI) to measure pleiotropy of nearly 2700 essential and nonessential genes. Our method uses frequent item set mining to discover GI modules, annotates them with high-level processes, and uses entropy to measure the functional spread of each gene's set of containing modules. We classify genes whose modules indicate broad functional influence as having high pleiotropy, while genes with focused functional influence have low pleiotropy. These pleiotropy classes differed in a number of ways: high-pleiotropy genes have comparatively higher expression variance, higher protein abundance, more domains, and higher copy number, while low pleiotropy genes are more likely to be in protein complexes and have many curated phenotypes. We discuss the implications of these results regarding the nature and evolution of pleiotropy.
