Towards the Determination of the Active Set of Elementary Flux Modes using Metabolic Flux Data

Abstract

Vaccine development at lab scale through to large scale produc6on can take 10-15 years. With the outbreak of the SARS-CoV-2 disease, emphasis on fast vaccine produc6on was emphasised. However, the cells that are grown to produce an6gens have complex metabolic networks consis6ng of thousands of reac6ons, metabolites, and genes. There is liJle understanding of why a cell in the same environmental condi6ons may grow via one route over another. If this process was beJer analysed, process op6misa6on to increase biomass growth and reduce inhibi6ng metabolites could be performed. All routes that a cell can use during its life are collec6vely known as elementary flux modes. Genome networks are being constructed over 6me allowing for full reac6on stoichiometry to be known. However, genome networks do not have all the elementary flux modes iden6fied due to the combinatorial explosion that occurs when solving as there can be billions of possible routes. In this thesis, mixed integer linear programming has been presented to enumerate elementary flux modes as a future proof method towards genome scale solving. It is compared to publicly available tools and mixed integer linear programming methods throughout literature. The benefits of this method in the future for finding elementary flux modes are also discussed. Compression techniques and code parallelisa6on are examined and reduced solve 6mes presented. Alongside elementary flux mode enumera6on this thesis also applies flux analysis techniques as a method of finding biologically relevant elementary flux modes. Disadvantages of these techniques are highlighted whilst presen6ng an integrated form of metabolic flux analysis to alleviate some of the issues. The technique presented is proven to be a viable method for enumera6ng elementary flux modes with the integra6on of fluxes. E.coli can be modelled as the full genome network or a reduced set of reac6ons represen6ng the key areas of the network; this is known as the core network. E. coli fermenta6on data from GlaxoSmithKline was provided for this work, allowing for analysis techniques iden6fied and created in this work to be applied. However, this data was found to be underdetermined preven6ng aspects to flux analysis and elementary flux mode enumera6on to be performed. This thesis discusses the process data and es6mates specific growth and uptake rates for all III fermenta6ons in batch and fed batch opera6ons. This key data was missing and helpsin beJer understanding the opera6ons taking place in the fermenters. More importantly however areas where more data is required for flux analysis are presented along with the issues of data limita6on on finding the elementary flux modes even for the core network. Underdetermined flux analysis allowed for es6ma6ons on the number of possible elementary flux modes in batch and fed batch opera6ons, highligh6ng the reduc6on in feasible routes during fed batch due to the cell’s phase.