Thesis
Bayesian verification of stochastic models in systems biology
- Abstract:
- The accurate construction and verification of mathematical models from data in biology are paramount to the success of the fields of systems and synthetic biology. For systems biology, mathematical models are required to accurately model and provide further understanding of biological phenomena of interest, while for synthetic biology, such models are required for the efficient design and development of novel biological devices. In this thesis we integrate techniques from Bayesian methods, formal verification and machine learning to provide a framework that allows one to formally verify a property of interest on an underlying stochastic system with limited observed data. First we introduce this framework, titled Bayesian verification, which integrates likelihood-free inference techniques and probabilistic model checking to construct and verify mathematical models with accurate results, using sparse and noisy experimental data. Secondly, we then extend the generality and thus applicability of this framework by utilising statistical model checking and supervised machine learning, resulting in the Approximate Bayesian Computation - sequential Monte Carlo with Statistical model Checking (ABC-(SMC)2) framework. With this extended Bayesian verification framework, with noisy and sparsely observed data, we can simultaneously learn and verify a model of interest, the only requirement being that we are able to generate simulations from a proposed model. Although this extension greatly extends the generality of the Bayesian verification framework, the number of simulations required to perform Bayesian verification becomes restrictive. Thirdly, because of the aforementioned, we present a multifidelity Bayesian verification framework. This multifidelity Bayesian verification framework reduces the computational cost of the simulations required to perform data-driven verification, and here we also seek to further reduce the number of simulations required by exploring when and how to perform statistical model checking. We conclude the thesis with a summary of our results and discuss directions for future work.
Actions
Access Document
- Files:
-
-
(Preview, Dissemination version, pdf, 3.3MB, Terms of use)
-
Authors
Contributors
+ Abate, A
- Institution:
- University of Oxford
- Division:
- MPLS
- Department:
- Computer Science
- Role:
- Supervisor
+
Engineering and Physical Sciences Research Council
More from this funder
- Grant:
- EP/L016494/1
- Programme:
- EPSRC and BBSRC Centre for Doctoral Training in Synthetic Biology
- DOI:
- Type of award:
- DPhil
- Level of award:
- Doctoral
- Awarding institution:
- University of Oxford
- Language:
-
English
- Keywords:
- Subjects:
- Pubs id:
-
2043472
- Local pid:
-
pubs:2043472
- Deposit date:
-
2022-06-30
- ARK identifier:
Terms of use
- Copyright holder:
- Molyneux, GW
- Copyright date:
- 2020
If you are the owner of this record, you can report an update to it here: Report update to this record