Digital Microfluidic Biochips (DMFBs) have the potential to fundamentally transform biochemical disciplines through automation, miniaturization, and the ability to facilitate repeatable chemical experimentation. Programming DMFBs has historically been accomplished by writing low-level bit manipulations to select which electrodes should activate in sequence. Recent research on high-level programming languages and compilers for DMFBs have begun to address the programmability challenge, but important capabilities such as loading and executing pre-compiled libraries and function calls, are absent from the literature. A primary driver of this oversight is the lack of a memory hierarchy to store physical chemicals off-chip to jump to and from function calls. This paper addresses the complexities involved in compiling function calls within the technology's unique boundaries, and provides a proof-of-concept implementation from language to code generation, with solutions evaluated using a cycle-accurate DMFB simulator as well as physical execution on an open-hardware DMFB.