Revolutionizing Synthetic Biology with AI-Enhanced Manufacturing Workflows

Synthetic biology is a multidisciplinary field applying engineering principles to the rational design of living systems. Engineering biology can offer fundamental biological insights unattainable through traditional means, and provide new-to-nature functionalities for applications in biotechnology. The synthetic biology workflow of “Design, Build, Test and Learn” (DBTL) reflects the iterative nature of engineering living systems for which genomic data is incomplete for even the simplest organisms1, and often involves trial and error due to the complex nature of living systems.

A workflow to engineer organisms meets manufacturing challenges at each stage. Chemical DNA synthesis is limited to short oligonucleotides due to errors and yield declines inherent to the chemistry2. Sequence features such as repeats, secondary structure, and high %GC content are difficult to synthesise accurately. Synthesis providers have specific constraints and proprietary screening services, and submitting sequences for synthesis could entail rejection and recoding (where possible), time delays, or price premiums. Synthesis providers often provide tools to optimize codon usage for expression in a given organism, yet recoding non-coding DNA, which may harbour regulatory functionality, is hence unpredictable.

Different approaches exist to assemble short synthetic DNA into larger constructs3.Quality control, error correction, and integration into the host cell must occur before design specifications have been met, and the phenotypic viability of the final organism assessed. Differences in assembly methods and laboratory capabilities represent challenges to optimization, standardization and reproducibility. Two labs may assemble the same DNA construct through different methods. Sub-optimal methodologies would scale poorly for large manufacturing projects, hence accurate sequence-based prediction of approaches to determine the optimal represents a valuable goal of computer-aided manufacturing for synthetic biology.

Artificial intelligence and synthetic biology naturally complement each other 4,5. Biologists have been producing high volumes of multi-omic datasets which could train predictive models, generate novel designs and optimal experiments. This project will investigate AI applications to manufacturing, or ‘Build’ (the ‘B’ in DBTL) workflows in synthetic biology. Under the supervision of prominent experts in both synthetic biology and AI, the researcher will help develop methodologies to directly address some of the greatest technical challenges in synthetic biology.

The researcher will investigate automated decision-making systems for the conversion of discrete manufacturing tasks into machinable and manual instructions specific to a given lab’s resources and equipment, thereby increasing the accessibility and uptake of collaborators in distributed manufacturing projects. Meeting this objective will leverage the supervisors successes in translating assembly designs to human and machine-readable protocols6,7 for automation, and expertise in employing AI to address challenges in the biosciences8. Quantifiable metrics will inform the scheduling of jobs best suited for robotic or manual execution, based on cost-benefit analyses. We will define assembly experiments and dynamically produce “hardware agnostic” instructions for automated liquid handling robots to circumvent time lost manually entering instructions.

An AI-assisted virtual laboratory for engineering biology is in development, based on digital twins of the cellular entities and processes involved, and the human designers, with the goal to develop human-centric AIs and human-AI collaborations. formulating assistants which are useful to human scientists while leaving them in full control. For this, the assistants will need to infer their user’s goals and then recommend actions in a way they understand. In other words they would need models of human users to efficiently collaborate with them

References

1 Gibson, D. G. et al. Science 329, 52–6 (2010) 2 Beaucage, S. L. et al. Tetrahedron Lett. 22, 1859–1862 (1981) 3 Casini, A. et al. Nat. Rev. Mol. Cell Biol. 16, 568–576 (2015) 4 Beardall, W. A. V. et al. GEN Biotechnol. 1, 360–371 (2022) 5 Eslami, B. Y. M. et al. Commun. ACM 65, (2022) 6 Luo, Y. et al. ACS Synth. Biol. 11, 579–586 (2022) 7 James, J. S. et al. ACS Synth. Biol. 11, 587–595 (2022) 8 Peuter, S. D. et al. Proc. AAAI Conf. Artif. Intell. 37, 11551–11559 (2023)

Additional information

Additional information in Find A PhD