Dissertation/ Thesis
The application of neural differential equations to enzymatic time-series
| Τίτλος: | The application of neural differential equations to enzymatic time-series |
|---|---|
| Συγγραφείς: | Heusdens, Danica Aimee |
| Συνεισφορές: | Rohwer, Johann, Stellenbosch University. Faculty of Science. Centre for Bioinformatics & Computational Biology. |
| Στοιχεία εκδότη: | Stellenbosch University |
| Έτος έκδοσης: | 2025 |
| Συλλογή: | Stellenbosch University: SUNScholar Research Repository |
| Θεματικοί όροι: | Enzymatic analysis, Differential equations -- Data processing, Neural networks (Computer science) -- Mathematical models, Time-series analysis -- Data processing, Saccharomyces cerevisiae, Enzyme kinetics, UCTD |
| Περιγραφή: | Thesis (MSc)--Stellenbosch University, 2025. ; Heusdens, D. A. 2025. The Application of Neural Differential Equations to Enzymatic Time-Series. Unpublished masters thesis. Stellenbosch: Stellenbosch University [online]. Available: https://scholar.sun.ac.za/items/1fb7a419-ca41-4aca-a096-aa7e4c0d8708 ; ENGLISH ABSTRACT: The aim of this thesis was to explore machine-learning models for parameter estimation in enzymatic systems, with focus on phosphoglycerate mutase (PGM) and enolase (ENO) in Saccharomyces cerevisiae. This thesis proposes a combination model of neural differential equations (NDE) and Hamiltonian Monte Carlo simulation (HMC). By utilizing time-series data obtained from the laboratory and processing it using the EnzymeML framework, this study demonstrated the combination of NDE-HMC and how it can enhance the accuracy of parameter estimation while reducing the computational demands compared to an HMC-only approach. This study employed the Catalax framework to train and validate the NDE, with the utilization of optimization algorithms such as L-BFGS and Nelder-Mead. The results show an improvement in computational efficiency and goodness of fit metrics for the NDE-HMC model compared to that of the HMC-only model. This thesis established a foundation for the utilization of machine-learning models to estimate enzyme kinetic parameters in biochemical reaction systems. ; AFRIKAANSE OPSOMMING: Geen opsomming beskikbaar. ; Masters |
| Τύπος εγγράφου: | thesis |
| Περιγραφή αρχείου: | viii, 74 pages : illustrations; application/pdf |
| Γλώσσα: | unknown |
| Relation: | https://scholar.sun.ac.za/handle/10019.1/132433 |
| Διαθεσιμότητα: | https://scholar.sun.ac.za/handle/10019.1/132433 |
| Rights: | Stellenbosch University |
| Αριθμός Καταχώρησης: | edsbas.73DC2083 |
| Βάση Δεδομένων: | BASE |
καταχωρήστε σχόλιο πρώτοι!