GCAD: A Computational Framework for Mammalian Genetic Program Computer-Aided Design

Genetic programs
can direct living systems to perform diverse,
prespecified functions. As the library of parts available for building
such programs continues to expand, computation-guided design is increasingly
helpful and necessary. Predictive models aid the challenging design
process, but iterative simulation and experimentation are intractable
for complex functions. Computer-aided design accelerates this process,
but existing tools do not yet capture the behavior of mammalian-specific
parts and population-level effects needed by mammalian synthetic biologists.
To address these needs, we developed a framework for mammalian genetic
program computer-aided design. Starting with a user-defined design
specification to quantify circuit performance, the framework uses
a genetic algorithm to search through possible designs. Circuit space
is defined by a library of experimentally characterized parts and
dynamical systems models for gene expression in a heterogeneous cell
population. We developed this genetic algorithm using a directed graph-based
formulation with biologically constrained rules to explore regulatory
connections and parts. We evaluated the framework for design problems
of varying complexity, including programs we describe as an amplifier,
signal conditioner, and pulse generator, demonstrating that the algorithm
can successfully find optimal circuit designs. Finally, we experimentally
evaluated selected circuits, demonstrating the path from a predicted
circuit design to experimental testing and highlighting the importance
of characterization in enabling predictive design. Overall, this framework
establishes general approaches that can be refined and expanded, accelerating
the design and implementation of mammalian genetic programs.