Browsing by Subject "optimisation"
Now showing items 1-6 of 6
-
(2021)Adeno-Associated Viruses (AAVs) are quickly becoming one of the most applied vectors for gene therapy applications. In the recent years three new AAV-based gene therapies have been approved by U.S. Food and Drug Administration (FDA) and European Medicines Agency (EMA). The regulatory bodies require accurate and reliable characterisation of the clinical grade viral vectors during and after production. Analytic methods measuring the purity, potency and safety of the product support the up-stream and down-stream processes during the production and are used for final-drug substance characterisation. Median Tissue Culture Infectious Dose (TCID50) is a well-established method for measuring the infectious titer of a virus. Here, an assay for determining the infectious titer of AAVs, which has previously been used to characterise the existing AAV2 Reference Standard Material (AAV2RSM) was set up and optimised for research use at Kuopio Center for Gene and Cell Therapy (KCT). The assay utilizes the HeRC32-cell line, a HeLa clone, that stably expresses AAV Rep and Cap -proteins and in presence of adenovirus, enables the replication of recombinant AAV-vectors. The cells were grown in 96-well plates and infected with 10-fold dilution series of AAV vectors (AAV2 and AAV6) using human adenovirus type 5 as the co-infector. 72 hours post infection the vector genome replication of AAV was detected with quantitative PCR (qPCR). Thresholds for qPCR determined copy number and cycle threshold (Ct) were set and used for the determination of infection-positive wells. The 50-percent endpoint was observed and used to calculate the infectious titer according to the Spearman-Kärber method. The assay was set up and optimised with the AAV2 Reference standard material (AAV2RSM) using two different primer-probe sets (targeted sequences were; SV40 polyadenylation signal and AAV inverted terminal repeats (ITRs)). Plates infected with AAV2RSM were analysed separately with both primers resulting in mean infectious titers of 8.07 ± 3.13 x 108 TCID50 Infectious Units (IU) / mL (n = 9) and 1.27 ± 0.464 x 109 TCID50 IU/mL (n = 9) for SV40 and ITR, respectively. After the assay was set up with the AAV2RSM, an in-house AAV6 product was analysed with the ITR primers yielding 6.09 ± 3.94 x 109 TCID50 (IU) / mL (n = 5). The assay protocol was successfully set up for research use at the KCT laboratory. Improvements were added to the original protocol to increase assay robustness, accuracy and precision as well as to minimize the possibility of over-estimation of the infectious titer. The assay can be further optimised for a particular therapeutic AAV product in the research laboratory or technology transferred to a production facility for optimisation and validation for the analytics needs of a production pipeline.
-
(2017)The aim of this thesis is to demonstrate the importance of selecting feasible and, preferably data-based prior assumptions for the Bayesian time-varying coefficient vector autoregressive model (TVC VAR model for further reference) of Primiceri (2005) and Del Negro and Primiceri (2015). The TVC VAR model would be suitable for quantifying, for example, the impacts of different monetary policy or fiscal policy regimes. The biggest advantage of the TVC VAR model is that it takes into account both changes in economic policy and in the private sector behaviour. The latter feature makes the model very compelling to use, because the private sector plays an important role in facilitating mote stable change in monetary and fiscal policy regimes. In complex mathematical models, such as the TVC VAR model, the objectiveness of the model may be compromised by deliberate selection of parameters. The TVC VAR model uses the Bayesian approach, which means that the researcher’s choice for the prior assumptions for the model plays an important role in the estimation. Unfortunately, Primiceri’s (2005) approach for selecting hyperparameters for the model is poorly explained and difficult to follow. Given that the model depends only for a small number of hyperparameters, it might be possible that the model can be tuned in a predefined way. To investigate whether the TVC VAR model can be tuned according to a researcher’s preferences, I design a proof of concept approach for optimising the hyperparameters of the model according to a set of predefined results. In other words, my research question is: could one tune the TVC VAR model to produce results according to the researcher’s bias? In my proof of concept approach I tune the TVC VAR model for six different targets for the Finnish government consumption multiplier. Given that Finland is a small open economy, Primiceri’s (2005) original hyperparameter values for the United States are not feasible and other values have to be used. The results from my proof of concept analysis show that the TVC VAR model can be tuned for predefined results, which shows that the practical reliability of the model can be easily compromised. My findings highlight the need for a comprehensible, data-based approach for selecting the hyperparameters for the model.
-
(2012)Brains are capable of processing information with remarkable efficiency under constraints set by the limited supply of physical resources such as the amount of space and the availability of metabolic energy. Natural selection has optimised the structure and function of brain networks using simple design rules similar to those found in man-made electronic and information systems. This study presents findings concerning a number of general principles of brain design governing the evolution and organisation of neural information processing. The rule of minimising wiring in neuronal networks is one such principle operating on multiple levels of brain organisation. Both individual components and larger brain architectural units are seen to feature characteristics of near-optimal wiring. Miniaturisation of neuronal components conserves space but raises problems about noise in signalling. Small-world organisation of anatomical and functional networks is widely employed in the brain, contributing to high global efficiency at low cost. Metabolic costs severely constrain signal traffic in the human brain, necessitating the use of energy-efficient sparse neural representations. Extensive evidence is presented of anatomical and physiological optimisations facilitating efficient information processing in brain networks. Limitations of current experimental techniques are discussed, with a view on possible future avenues of research.
-
(2023)The plastic pollution has become a massive problem in the Arctic, affecting aquatic, and terrestrial ecosystems, the cryosphere, and the atmosphere. One of the solutions proposed by the Arctic Council is to improve waste management by using renewable and sustainable materials. This is where bioplastics reveal their importance. They can be bio-produced by microorganisms from organic waste, they are biodegradable and can be reused. Their production relies on a circular economy system making it sustainable. Here lies the relevance of developing the bioplastic bioproduction and technology. The present research focused on the development of a specific production of polyhydroxyalkanoates (PHAs) from organic waste, in collaboration with the start-up Dionymer (Bordeaux, France). First, the purpose of the study was to up scale the process from the fermentation of chemical volatile fatty acids in flasks (400 mL culture medium) to 2 L bioreactors (BR) by characterizing the main differences in the two processes. Secondly, the research consisted in implementing and testing different set-up for the BR to enhance and improve bioplastic and biomass yields, including aeration and agitation. The characterization of the culture parameters differences between BR and flask pointed out; a higher viscosity of the medium at the end of the process, a darker PHA product and a lower final optical density (OD) (8 versus 12) respectively. Secondly, the focus was on the increase of the OD in BR and finding the origin of the stress, to do so, the following parameters were tested; - three aerations strategies; pO2<10%, <20% and >20%; - two agitations blades; marine and Rushton with baffles; - two aerations spargers; circular and micro. The results revealed that; the pO2 needs to be higher to 20% and it may be linked with the reduction of stress induced to the cells; the marine blades increased the OD and reduced the medium viscosity; the impact of the micro sparger seemed to improve aeration and tent to be very sensitive to antifoam agent that reduced the aeration of the medium. So far, the optimum BR set-up seemed to include the use of marine blades and a pO2 above 20%. More experiments of optimization still need to be performed to unsure a stable and higher production performance.
-
(2023)The scholarship allotment problem describes the goal of strategically offering scholarships to prospective students of a university in a way that optimises the expected return for that investment. The goal of this thesis is to formulate the scholarship allotment problem in multiple variations of increasing complexity while also introducing algorithms to solve those variations optimally as efficiently as possible. The thesis also offers some insight into the way more complex variations and generalisations heighten the difficulty of finding an optimal solution in a reasonable amount of time. The main focus and the main tool used to tackle these problems is the classic knapsack algorithm and different variations of it, like multiple-choice knapsack and multidimensional knapsack. In addition to the theoretical side, the thesis contains an empirical study into the performance and feasibility of the algorithms introduced. Concrete implementations of the algorithms discussed are all available on a public GitHub repository online: https://github.com/SirVeggie/scholarship-allotment.
-
(2023)The aim of this work is to develop and optimise an atmospheric inverse modelling system to estimate local methane (CH4) emissions in peatlands. Peatlands are a major source of CH4 regionally in boreal areas and they have significance on a global scale as a soil carbon storage. Data assimilation in the inverse modelling system is based on an ensemble Kalman filter (EnKF) which is widely used in global and regional atmospheric inverse models. The EnKF in this study is an implementation of the EnKF used in the global atmospheric inversion model CarbonTracker Europe-CH4 (CTE-CH4) applied to local setting in the peatland. Consistency of the methodology with regional and global models means that it is possible to expand the system in scale. Siikaneva fen in Southern Finland is used as a testbed for the optimisation of the system. Prior natural CH4 fluxes in Siikaneva are acquired from the HelsinkI Model of MEthane buiLd-up and emIssion for peatland (HIMMELI) which simulates exchange of gases in peatlands. In addition to the peatland fluxes, anthropogenic fluxes at the site are estimated as well in the inversion. For the assimilation of atmospheric CH4 concentration observations, the CH4 fluxes are transformed into atmospheric concentration with a simple one-dimensional box model. The optimisation of the system was done by changing parameters in the model which affect the data assimilation. In model optimisation tests it was discovered that the performance of the modelling system is unstable. There was large variability in the produced estimates between consecutive model runs. Model evaluation statistics did not indicate improvement of the estimates after the inversion. No exact reason for the unstability was able to be determined. Posterior estimates of CH4 fluxes for years 2012–2015 did not differ much from prior estimates and they had large uncertainty. However, evaluation against flux measurements showed reasonable agreement and posterior concentration estimates were within the uncertainty range of the observed concentration.
Now showing items 1-6 of 6