The strength of weird ties: positive genetic interactions have greater scientific impact but are under-represented in the literature

Abstract

Genetic interactions, where mutating two genes together affects a trait differently than expected from their individual effects, often reveal how genes work together. When the trait is cell growth, systematic studies show that negative interactions (worse than expected) usually link functionally related genes and are widely considered informative, whereas positive interactions (better than expected) tend to link seemingly unrelated genes and are often deemed less informative. However, the sociology of science suggests that transformative discoveries often arise from unexpected, cross-domain connections. To evaluate these competing perspectives, we linked large-scale yeast interaction data to the literature on each gene pair and its citations. Contrary to prevailing assumptions, papers on positively interacting gene pairs tend to be more highly cited. Yet positive interactions are significantly under-represented in the literature relative to their experimental prevalence. This gap persisted after a technology detecting both types equally well became available, implicating factors beyond mere discoverability. Research building on positive interactions also forms a more interconnected body of work. These findings reveal the underappreciated value of positive interactions. More broadly, they show how science-of-science insights can identify biases in biomedical research and provide concrete guidance on what to study next.