Developing a Standardized AI-ready Pipeline for Organizing, Representing, and Training on Massive Datasets of Integrated Genetic Circuits

Engineering biology has the potential to revolutionize nearly all sectors of the economy. However, this field lags far behind other engineering disciplines in terms of predictive design. One of the primary challenges is our ability to predict how the biological components needed for engineering biology will behave when integrated together - referred to as biological circuits. This project supports a graduate student for a full-time research internship at the National Institute of Standards and Technology (NIST) Living Measurement Systems Foundry (LMSF) to advance AI-ready data standardization and integration for design and analysis of massive sets of more than 100,000 biological circuits. Working onsite at NIST LMSF, the Scholar-in-Residence will collaborate with experimental scientists to refine and apply Knox, a software platform for design and analysis of biological circuits. By combining high-quality experimental measurements with machine learning (ML) models, the project aims to accelerate the Design-Build-Test-Learn (DBTL) cycle, reducing experimental costs and making biological engineering more reproducible and accessible. The project strengthens the national bioeconomy infrastructure by creating open, standards-aligned computational tools for trustworthy biological engineering. By enabling faster and more accurate prediction of circuit behavior, Knox can shorten development timelines and improve the reliability and shareability of engineered biological systems. A public Knox instance, along with AI-ready datasets, and design rules, will lower barriers for researchers and educators working with biological circuits. The deployment of Knox at NIST LMSF will provide a sustainable computational infrastructure for measurement-based standards development. The Scholar-in-Residence offers a unique training experience at the intersection of computational modeling, experimental measurement, and standards science, preparing a highly skilled researcher to contribute to the future biotechnology workforce. By promoting reproducibility, transparency, and responsible design practices, the project supports safe and secure growth of the synthetic biology ecosystem. This project advances NSF’s priorities in Artificial Intelligence and Biotechnology. The project focuses on applying and refining the Knox platform, a grammar-based and ML-driven system for designing and analyzing large combinatorial spaces of biological circuits. The Scholar-in-Residence will analyze experimental results, identify key metadata influencing biological circuit performance, train predictive ML models, extract mechanistic design rules, and validate Knox’s generated designs and rules experimentally. The project advances fundamental understanding of biological circuit behavior by integrating NIST LMSF measurement data with interpretable ML. By refining Knox’s grammar, metadata schema, and predictive modeling capabilities, the work will generate mechanistic insights into sequence-structure-function relationships for RNA circuits. Experimental validation of Knox-derived designs and rules will close the DBTL cycle, demonstrating the platform’s ability to learn from data and produce generalizable design rules. Beyond RNA circuits, the project contributes broadly to computational biology and sequence-based modeling. Knox’s grammar-based representation and interpretable ML pipeline are generalizable to other domains where sequences encode function, including regulatory genomics, protein engineering, and other forms of biological circuit design. The platform’s architecture also parallels methods used in natural language processing, enabling cross-domain transfer of modeling strategies and interpretability techniques. The resulting workflows, model architectures, and rule sets will contribute to emerging standards for reproducible, transparent, and machine-readable biological design, aligning with the missions of both NSF and NIST. This award is funded by a collaborat NSF Award ID: 2622542 | Program: 01002627DB NSF RESEARCH & RELATED ACTIVIT,4082CYXXDB NSF TRUST FUND | Principal Investigator: Douglas Densmore | Institution: Trustees of Boston University, BOSTON, MA | Award Amount: $120,857 View on NSF Award Search: https://www.nsf.gov/awardsearch/show-award/?AWD_ID=2622542 View on Research.gov: https://www.research.gov/awardapi-service/v1/awards/2622542.html