(375b) Artificial Intelligence-Based Bacteria Species Distinction Based on Raman Microscopy.

A biomanufacturing setup harnesses the manufacturing potential of animal, plant or microbial cells to produce pharmaceutical, food and feed products. Biocentric manufacturing has many advantages over its traditional counterparts, with the primary being its renewable and sustainable nature.[1] However, most industrial bioprocesses are empirical processes which are strictly single culture processes. This design choice restricts taking advantage of co-cultural biopathways.[2] One of the main obstacles to implementing co-culture bioprocesses is a reliable analytical technique which will be able to distinguish between various species and give a quantitative outlook of the co-culture dynamic. Vibration spectroscopy can be used to provide label-free non-destructive chemical information, with recent studies showing Raman microscopy to have great promise as an analytical tool.[3] Being able to collect and analyze this type of data will aid in more robust model construction and discovery, monitoring and process control.

In this work we will aim to build an image processing and classification protocol which can distinguish between Pseudomonas putida and Corynebacterium glutamicum based not only on the Raman spectra bit also physical attributes gleamed from the hyperspectral dataset. These bacteria were chosen as they have both been shown to have great potential in depolymerizing lignin and converting it to cis, cis – muconinc acid.[4] Utilizing machine learning techniques for dimensionality reduction and clustering, and chemometric multi curve resolution techniques, the transformed combined spectra can be decomposed to quantify concentrations of each bacterium in the co-culture sample. This work will illustrate the methodology to deconvolute Raman signals which are not linearly separable which is a critical first step towards monitoring microbial co-cultures. The applications for this methodology are not limited to co-culture biomanufacturing setups but can be used for medical diagnostics and quality control.

References

[1] J. M. Clomburg, A. M. Crumbley, and R. Gonzalez, “Industrial biomanufacturing: The future of chemical production,” Science, vol. 355, no. 6320, p. aag0804, Jan. 2017, doi: 10.1126/science.aag0804.

[2] S. Kyriakopoulos et al., “Kinetic Modeling of Mammalian Cell Culture Bioprocessing: The Quest to Advance Biomanufacturing,” Biotechnol. J., vol. 13, no. 3, p. 1700229, 2018, doi: 10.1002/biot.201700229.

[3] K. A. Esmonde-White, M. Cuellar, C. Uerpmann, B. Lenain, and I. R. Lewis, “Raman spectroscopy as a process analytical technology for pharmaceutical manufacturing and bioprocessing,” Anal. Bioanal. Chem., vol. 409, no. 3, pp. 637–649, 2017, doi: 10.1007/s00216-016-9824-1.

[4] N.-Z. Xie, H. Liang, R.-B. Huang, and P. Xu, “Biotechnological production of muconic acid: current status and future prospects,” Biotechnol. Adv., vol. 32, no. 3, pp. 615–622, May 2014, doi: 10.1016/j.biotechadv.2014.04.001.