Genomic Science Program
U.S. Department of Energy | Office of Science | Biological and Environmental Research Program

Computational Modeling to Enable Predictive Secure Biosystems Designs


Rodrigo Santibanez Palominos1* (, Juan Tibocha-Bonilla2, Bin Yang6, Chao Wu6, Jianping Yu6, Wei Xiong6, Karsten Zengler1,3,4,5, Michael Guarnieri6


1Department of Pediatrics, University of California–San Diego; 2Bioinformatics and Systems Biology Graduate Program, University of California–San Diego; 3Department of Bioengineering, University of California–San Diego; 4Center for Microbiome Innovation, University of California–San Diego; 5Program in Materials Science and Engineering, University of California–San Diego; 6National Renewable Energy Laboratory



Systems modeling is an integral part of the Integrative Modeling and Genome- scale Engineering for Biosystems Security (IMAGINE BioSecurity) Science Focus Area project that address biosafety concerns related to microbial biocontainment and performance stability. Researchers aim to develop integrative bioinformatics tools for the efficient modeling of metabolism and gene expression (ME-models) of bacterial systems and metabolic flux analysis (MFA). These ME-models and MFA analysis will be used for the predictive design of novel containment strategies by identifying critical metabolic reactions governing secure biosystems designs in engineered bacteria.


Genetically modified microorganisms (GMMs) are widely used in agriculture and bioenergy industries. Among current systems biology tools, computational methods can interrogate systems with unprecedented detail and high throughput. However, lagging behind is the application of these tools for secure biosystems designs. Here, the group presents coralME, FreeFlux, and EMUlator2ML, new computational tools aiming to design predictable and generalizable biocontainment and robustness stabilization strategies.

First, the team developed coralME to automatically reconstruct nearly finished ME-models from genome-scale metabolic models (M-models), which allowed researchers to complete four highly curated ME-models for bioeconomy relevant microorganisms Pseudomonas putida KT2440 (iPpu1686-ME), Synechocystis sp. PCC 6803 (iSyn1015-ME), Clostridium ljungdahlii DSM 13528 (iClj978-ME) and Mycoplasma mycoides JCVI-Syn3A (iMmy259-ME). As they include gene expression, the number of components and reactions grows to accommodate transcription and translation. For instance, iPpu1686-ME models 224 additional gene products (+15.32%) compared to its parental M-model, iJN1463, in a network of 14,426 reactions and 7,566 components, increasing in 392.86% and 251.42% the number of reactions and components, respectively. Additionally, coralME aided the development of 17 draft ME-models for diverse bacteria, covering eight phyla. Application of coralME to these 21 different organisms resulted in short reconstruction times, effectively reducing the reconstruction of ME-models from several months to minutes. The platform is ideally suited for reconstruction of feasible ME-models employing efficient troubleshooting and reporting methods that repair and guide the manual addition of reactions, respectively. Consequently, coralME has accelerated modeling and simulation, further allowing the prediction of hundreds of essential genes, microbe-microbe interactions, overflow metabolites, use and essentiality of enzyme cofactors, and proteome composition.

In parallel, the team have developed FreeFlux, an open-source Python package which offers comprehensive 13carbon-MFA analysis, boasting swift, reliable flux computation. EMUlator2ML, a machine learning framework, accelerates flux estimation, enhancing large-scale analysis and strain screening by “learning” intrinsic relationships between metabolite labeling patterns and metabolic flux, inferring fluxomic phenotypes from isotopically metabolomic datasets. It demonstrated the ability to use M-models to generate the training dataset with minimal reliance on experimental data. Finally, the group is introducing a biocontainment approach combining ensemble computational modeling with CRISPR interference (CRISPRi) to modulate GMM metabolism, targeting core robustness for growth instability. This approach enables identification of enzymatic targets sensitive to expression perturbations and establishing genetic circuits for enhanced performance and safety, with strains meeting the National Institutes of Health escape frequency standard, validated across various conditions.

In summary, researchers used ensemble modeling, FreeFlux, EMUlator2ML, and highly curated M-models and ME-models to elucidate microbial metabolism under variable conditions, metabolic interactions within microbiomes, and the productivity of GMMs under secure biosystems constraints in an iterative Design-Build-Test-Learn cycle. The resultant pipeline will enable rapid and predictive secure biosystems designs.


Wu, C., et al. 2022. “A Computational Framework for Machine Learning-Enabled 13C-Fluxomics,” ACS Synthetic Biology 11(1), 103–15. DOI:10.1021/acssynbio.1c00189.

Wu, C., et al. 2023. “FreeFlux: a Python Package for Time- Efficient Isotopically Nonstationary Metabolic Flux Analysis,” ACS Synthetic Biology 12(9). DOI:10.1021/acssynbio.3c00265.

Yang, B., et al. Submitted. “Advancing Biocontainment Design Through Computational Steering and CRISPRi-Driven Robustness Regulation,” Science Advances.

Funding Information

This research was supported by the DOE Office of Science, BER program, GSP, Secure Biosystems Design Science Focus Area IMAGINE BioSecurity: Integrative Modeling and Genome-scale Engineering for Biosystems Security, under contract number DE-AC36-08GO28308.