The projected depletion of fossil oil resources, and the threat of global warming as a direct consequence of their continued use, has led to the challenge of providing alternative and renewable routes to oil derived fuels and commodity chemicals. However, any such process faces severe demands in terms of yield and costs before it can be considered industrially relevant. In addition, the production of oil derived compounds (ie hydrocarbons) or their functionalization does not belong to what can be considered the "standard" biochemical repertoire. Hence, comparably few biocatalysts, pathways or organisms have been reported to might support such processes. Therefore, while the challenge is clearly defined, the route to a suitable solution if far from clear, and there is an unmet industrial need for new biocatalysts that can produce commodity chemicals such as butadiene (precursor to synthetic rubber) and aromatic dicarboxylic acids (precursors to PET and other polymers). We have recently been able to provide some detailed insights into the mechanism of the UbiD family of enzymes. These act as reversible decarboxylates (ie adding or removing a CO2 group to/from an organic molecule). We and others have shown they are able to interconvert unsaturated hydrocarbons (ie containing a double bond) with the corresponding carboxylic acids (ie version containing the additional CO2 group). A wide range of substrates has been reported, and the reaction catalysed appears readily reversible depending on [CO2] levels. Our recent work on these enzyme has established the achieve this unusual chemistry by making use of a previously unknown enzyme cofactor. In this project, we seek to complete our understanding of how this cofactor is made, and also how it is inactivated, to guide future application. Building on our fundamental understanding, we seek to derive novel and green production routes to commodity chemicals using renewable feedstocks as UbiD substrates. More specifically, in this grant we will seek to establish to what extend UbiD enzyme can be evolved to produce compounds such as butadiene and aromatic dicarboxylic acids. We will make use of the renewable feedstocks muconic and benzoic acid as UbiD substrates respectively. Our results will provide proof of concept and outline the scope for future biotechnological application.

Carboxylase enzymes are able to link CO2 to other molecules, thereby using it as a readily accessible and cheap one-carbon building block. Furthermore, carboxylase activity contributes to the reduction of global CO2 levels. Indeed, most life on the planet directly or indirectly depends on the photosynthesis driven action of a key carboxylase able to generate biomass from CO2. Thus, carboxylases could serve to meet the challenge of reducing atmospheric CO2 levels and creating a more sustainable economy. However, engineering and application of these enzymes has often met with slow progress due to the complexity and/or strict requirements of these proteins for activity. In contrast, decarboxylase enzymes (that normally achieve the opposite reaction, i.e. cleaving a CO2 from a molecule) in the reverse direction has met with some success. Achieving carboxylation traditionally requires either large amounts of CO2 (pushing the reaction in the right direction) or additional enzymes that are highly efficient in rapidly converting the carboxylate product (pulling the reaction in the right direction). A third option is to couple the decarboxylase reaction directly to a favourable reaction such that the decarboxylase effectively runs in reverse under ambient CO2 and yields the carboxylated product at the expense of the reagents from the coupled reaction. Nature has achieved this feat in the form of the phenol carboxylase enzyme system, which is involved in bacterial phenol degradation. This enzyme catalyses the biological equivalent of the industrial Kolbe-Schmitt reaction, which uses high pressure CO2 and temperatures exceeding 100 degrees Celsius to carboxylate phenol. In contrast, this intriguing enzyme system operates and ambient conditions, and uses ATP (the natural "energy currency") to drive phenol + CO2 yielding the product para-hydroxybenzoic acid. We seek to determine how the phenol carboxylase achieves the coupling of both reactions (i.e. carboxylation and ATP consumption) to determine the natural engineering principles and apply these to development of new pathways that incorporate ATP-dependent carboxylase enzymes such as the phenol carboxylase. We will make use of protein crystallography to determine the structure of the various components of this enzyme, and combine that with detailed computational and solution studies to present a complete picture of the molecular choreography that underpins activity. We will use the insights generated as a blueprint for the laboratory guided evolution of this system to expand the substrate repertoire to non-phenolic aromatic alcohols. Evolved carboxylase enzymes will be used to generate new and renewable pathways for the production of key chemical commodities such as FDCA (furan dicarboxylic acid) or terephthalate from biomass.

In this Turing Artificial Intelligence Acceleration Fellowship, I will focus on artificial intelligence for medical treatments and therapies. I take the view that AI is a question on how to realise artificial systems that solve practical problems currently requiring human intelligence to solve, such as those solved by clinicians, nurses and therapists. Critical care is high risk and highly invasive environment caring for the sickest patients at greatest risk of death. Patients within this environment are highly monitored, enabling sudden changes in physiology to be attended to immediately. In addition, this monitoring requires a heavier staffing ratio (often 1:1 nursing; 1:8 medical) and variances in human factors and non-technical pressures (e.g. staffing, skill-mix, finances) leads to critical care delivery being disparate. AI in healthcare is a hard problem as, due to the diversity and variability of human nature, systems have to cope with unexpected circumstances when solving perceptual, reasoning or planning problems. Crucially, AI has two facets: Understanding from data, and Agency. While rapid strides have been made on learning from data, e.g. how to make medical diagnosis more precise and faster than human experts, there is little work on how to carry on after the diagnosis, e.g. which therapy and treatment to conduct. The latter requires agency and has seen fewer applications as it is a harder problem to solve. My clinical partners and I want to develop the required AI algorithms that can learn and distil the best plan of action to treat a specific patient, from the expert knowledge of clinicians. We will focus on an area of AI called RL that has been successful in enabling robots and self-driving cars to learn a form of autonomous agency. We want to transform these methods into the healthcare domain. This will require the development of new RL algorithms, able to efficiently understand the state of a patient from noisy and ambiguous hospital data. The system will not only learn to recommend interventions such as prescribing drugs and changing dosages as needed per patient but to make these recommendations in a manner that is meaningful to the clinical decision-makers and helps them make the best final decision on a course of action. The methods developed as part of this project can be used in different applications beyond healthcare. Many sectors within industry, such as aerospace, or energy, deal with similar bottlenecks. These are highly regulated environments, with great need for decisions making support, but a scarcity of highly skilled human experts. With sufficient data, our methods can be applied to these sectors as well, to distil the required human expertise and best practices from top experts, and use them to drive decision making all over the sector, for increased efficiency and safety.

Decarbonising the transport sector is a top priority worldwide. The difficult-to-decarbonise transport applications (including mainly shipping, road freight and aviation) emit more than 50% CO2 of the entire transport sector. Among efforts on developing low-emission fuels, liquid synthetic fuels that can massively reduce pollutant emissions are drawing increasing attention, as they can be integrated into the current transportation system using existing infrastructure and combusted in existing engines (such as diesel engines for optimal fuel economy) with minor adjustments as drop-in fuels. Liquid synthetic fuels such as oxymethylene ethers (OMEx, which possess liquid properties similar to diesel when x=3-5) can be produced from a range of waste feedstocks and biomass, thereby avoiding new fossil carbon from entering the supply chain. OMEx can also be produced as an electrofuel (or e-fuel), thereby used as a sustainable energy carrier. However, due to the lack of complete knowledge of the physicochemical properties associated with the fuel composition variability, i.e. variation in the oligomer length (the x value of OMEx) and the composition variation of OMEx-diesel blends in real engine environment, there are challenges in utilising OMEx in practical engines, mainly in engine and its operation adjustments for optimal performance and minimal pollutant emissions. To address the technical issues of OMEx utilisation, accurate information on physicochemical properties and pollutant emissions of the synthetic fuels over the engine operational ranges is mandatory, but this is not readily available. This project is intended to obtain a thorough understanding on liquid synthetic fuel utilisation. The project will address the fundamental challenges in utilising renewable synthetic fuels, in particular OMEx and the associated OMEx-diesel fuel blends. The study will follow a combined modelling / simulation - experimentation approach, predicting the physicochemical properties including emission characteristics of the alternative fuels using molecular dynamics simulations, tailor-made experimentation for first-hand information on fuel utilisation, and establishing a database / mapping to guide the synthetic fuel utilisation in real engines over a wide range of conditions using machine learning.

Ammonia (NH3) is a promising zero-carbon fuel for future transportation. Today transportation emits around 8.9 billion tonnes of CO2 annually. Whilst some sectors (e.g. cars) can be decarbonised using batteries, heavier transport (marine or freight) are less likely to use batteries due to their cost and energy density. Ammonia is a hydrogen carrier, and (by volume) contains 50% more hydrogen than liquid hydrogen (which alone is extremely energy intensive to liquefy and store). Ammonia has among the highest energy densities of any non-hydrocarbon (traditionally fossil) fuel. Ammonia is particularly attractive because it can be made using the well-established Haber-Bosch process, which today is used to make 230 million tonnes of ammonia per year. Ammonia production can be 100% renewable when powered by solar and wind. This means that ammonia production can be scalable and can be undertaken repurposing a large amount of existing infrastructure. A number of pilot projects are underway worldwide with Ammonia, including for energy storage, shipping and freight transportation. Many of these are in the UK, including at the Rutherford Appleton Laboratory, Cardiff University and the University of Nottingham. However, these projects typically adapt existing technology, which is designed for a different fuel (fossil fuels usually). There is a significant lack of fundamental data to enable the design of energy conversion systems specific to ammonia. This project, AmmoSpray, aims to fill this gap. AmmoSpray will provide, for the first time, fundamental data on ammonia sprays into air. Ammonia can be sprayed into air either as a liquid or as a gas, and both will be investigated in this project. The fundamental data obtained will include spray break-up (how liquid ammonia breaks up and evaporates upon injection) and how ammonia and air mix under realistic conditions. These studies will be undertaken on three different pieces of test equipment: 1. An ambient conditions spray rig 2. A Cold Driven Shock Tube (CDST) 3. An optical access thermal propulsion system (TPS) The spray rig is fast and cheap to run, and will enable the development of the experimental systems required for this project, the testing of large numbers of spray test conditions, and will be used to undertake a scoping exercise to identify project boundaries. The CDST is a unique facility in the UK, able to replicate conditions found during combustion (150 bar pressure, 1500 K temperature) without turbulence, and with space for test equipment. This will enable for the first time imaging and break-up studies of ammonia sprays at conditions that will be seen in-use - key fundamental data. The optical TPS tests are the logical next step, adding turbulence, and replicating as closely as possible 'real' conditions, whilst still allowing access for imaging and test equipment. The key tests here will be on mixing, using a laser-based technique (PLIF) to obtain ammonia:air ratio measurements throughout the combustion volume. This will link the sprays information developed earlier to their combustion characteristics. The tests on the optical access TPS will also enable studies of how these different spray and mixing methodologies influence emissions formation for ammonia combustion, with NH3 and NOx the key emissions which will be measured. This step-by-step nature is perfectly suited for improving existing models. The data obtained will be coded into commercial modelling software (computational fluid dynamics (CFD)) provided by project partner, Convergent Science. Its CONVERGE CFD software is used by companies globally. The data obtained will be used to develop models for ammonia spray break-up, mixing, and emissions formation upon combustion. This will all happen in parallel with the experimental program and will ensure that the project's utility well beyond the project itself, with the models developed being available to be used by any of the global users of the software.