Machine Learning and Chemical Imaging to Elucidate Enzyme Immobilization for Biocatalysis

Abstract

Biocatalysis has rapidly become an essential tool in the scientific and industrial communities for the development of efficient, safe, and sustainable chemical syntheses. Immobilization of the biocatalyst, typically an engineered enzyme, offers significant advantages, including increased enzyme stability and control, resistance to environmental change, and enhanced reusability. Determination and optimization of the spatial and chemical distribution of immobilized enzymes are critical for proper functionality; however, analytical methods currently employed for doing so are frequently inadequate. Machine learning, in the form of multivariate curve resolution, with Raman hyperspectral imaging is presented herein as a potential method for investigating the spatial and chemical distribution of evolved pantothenate kinase immobilized onto two diverse, microporous resins. An exhaustive analysis indicates that this method can successfully resolve, both spatially and spectrally, all chemical species involved in enzyme immobilization, including the enzyme, both resins, and other key components. Quantitation of the spatial coverage of immobilized enzymes, a key parameter used for process development, was accomplished. Optimal analytical parameters were determined by the evaluation of different excitation wavelengths. Exploratory chemometric approaches, including principal component analysis, were utilized to investigate the chemical species embedded within the data sets and their relationships. The totality of this information can be utilized for an enhanced understanding of enzyme immobilization processes and can allow for the further implementation of biocatalysis within the scientific and pharmaceutical communities.

Putting precision and elegance in enzyme immobilisation with bio-orthogonal chemistry

“…The machine learning was based on multivariate curve resolution-alternating least squares (MCR-ALS), which was used to explain pure response profiles of the chemical constituents within a complex. 245 Machine learning can be used to understand and improve enzyme immobilisation for biocatalytic processes, and is likely to become an active research area in the near future.…”

Modernized Machine Learning Approach to Illuminate Enzyme Immobilization for Biocatalysis

“…Previous work demonstrated that principal component analysis (PCA) can provide concentration profiles but not sufficiently resolved spectra for direct comparison with reference data. 36 Multivariate curve resolution-alternating least-squares (MCR-ALS) can provide concentration profiles and relevant spectra to unmix species involved in enzyme immobilization; 27 however, optimal MCR-ALS model selection and subsequent species identification remains challenging.…”

“…Machine learning has exhibited remarkable performance in analyzing vast spectroscopic data from biological samples with complex backgrounds. − Specifically, non-negative matrix factorization (NMF) is an unsupervised machine learning technique that is frequently employed for unmixing pure, individual chemical species present within mixtures. − NMF operates by factorizing an original non-negative data matrix into both non-negative basis vectors and corresponding weights via multiplicative algorithms that minimize the norm of the difference matrix between the original data matrix and its approximate reconstruction. − NMF is advantageous over other decomposition methods because the resolved components reflect true spectra and provide direct chemical information about a mixture rather than variance-based indirect loadings. However, selecting the appropriate number of components to construct an NMF model while avoiding over- or underfitting remains a challenge.…”

Raman spectroscopy and multivariate analysis for identification and classification of pharmaceutical pain reliever tablets

“…6 Raman spectroscopy has demonstrated its utility in an extensive variety of applications, some of which include disease diagnostic development efforts, [7][8][9][10] forensic science advancements, [11][12][13] protein structure and shape analysis, [14][15][16] and pharmaceutical investigations. [17][18][19][20][21][22] Raman spectroscopy poses many distinct advantages compared to other methods due to its ability to generate a spectral "fingerprint" that is representative of the overall biochemical composition of the sample. 7,[23][24][25] Due to the inherent specificity of the method, small changes that exist between similar samples can be detected.…”

“…As an analytical methodology, Raman spectroscopy is an exceptional tool that provides distinct and specific information describing the molecular composition of a sample through excitation of molecular vibrations using a monochromatic light source 6 . Raman spectroscopy has demonstrated its utility in an extensive variety of applications, some of which include disease diagnostic development efforts, 7–10 forensic science advancements, 11–13 protein structure and shape analysis, 14–16 and pharmaceutical investigations 17–22 . Raman spectroscopy poses many distinct advantages compared to other methods due to its ability to generate a spectral “fingerprint” that is representative of the overall biochemical composition of the sample 7,23–25 .…”