Over the past several decades, biology has become an increasingly data-driven science. Due in large part to new techniques that allow massive collection of biological data, including next-generation sequencing and high-throughput experimental screening, many of the limitations currently facing the field are in the organization and interpretation of these data. In this dissertation, I present several computational methods and resources designed to organize and perform functional inference on these systems-level biological data sources. In Chapters 2 and 3, I describe the construction of a database and web tool to aid in foundational genomics research by providing predictions of interacting protein domains in interactomes and all-by-all conversions of popular variant identification formats. In Chapter 4, I describe the construction of the first whole-interactome protein interaction network in the fission yeast S. pombe, and, through comparisons with other complete networks in human and the budding yeast S. cerevisiae, demonstrate principles of functional evolution. Finally, in Chapters 5 and 6, I propose two new methods for functional genomic inference—an algorithm to predict cancer driver genes and mutations through 3D atomic clustering of somatic mutations and an ensemble machine learning method to predict the 3D interfaces of protein interactions by taking into account the evolutionary relationships and biophysical properties of proteins. Taken together, this suite of computational resources will help researchers interpret biological function on a genomic scale.