The problem. Synthetic biology is organized around the Design–Build–Test–Learn cycle, but “where exactly does software add value?” is usually hand-waved. This review answers it computation-first, walking the whole loop and naming the tools at each stage.

The idea. It surveys the DBTL toolchain node by node. Design: biological CAD, pathway and genome design, sequence/part selection. Build: DNA-assembly automation and biofoundry robotics. Test: analytics and omics readouts that quantify what was built. Learn: ML models, active learning, and design-of-experiments that turn test data into the next design. Crucially, it foregrounds the hand-offs — the data formats and interfaces where one stage feeds the next — which is precisely where a software/data layer either exists or doesn’t.

Why it matters. This is the most useful positioning document I’ve read for the “bioinformatics shop eyeing synbio” question. The wet-lab Build needs a foundry; the Design and Learn ends, and every hand-off between stages, are software and data problems — exactly what a pipelines-and-ML team can own without owning a lab. It reframes DBTL from a biology diagram into a systems-integration diagram, which is a language I already speak.

Verdict. As a 2024 survey it’s current and refreshingly computation-centric, though (being a review) it’s a map, not evidence — the real test is whether the hand-off seams it identifies are as ownable as they look. My plan: extract its DBTL diagram and annotate each node with “foundry-required” vs. “a data shop can own this,” as a positioning artifact.