The world is facing a humanitarian crisis, resulting from infections caused by antimicrobial resistant (AMR) organisms.  It is estimated that bacterial AMR was directly responsible for 1.27  million global deaths in 2019 and further contributed to 4.95 million deaths.  In addition to death and disability, AMR has significant economic costs. The World Bank estimates that AMR could result in US$ 1 trillion additional healthcare costs by 2050, and US$ 1 trillion to US$ 3.4 trillion gross domestic product (GDP) losses per year by 2030. But thankfully there’s a lifeline to the threat of AMR, and it comes in the form of bacteriophages.

Bacteriophages are viruses that only infect bacteria and are completely inert and harmless in humans, plants and animals.  As part of their natural lifecycle, they kill their target bacteria, releasing sometimes hundreds of progeny phage, each of which can then go on and kill other bacteria.  They are extremely efficient killers of harmful bacteria and it is estimated that every two days, half of the bacteria on the planet are killed by phages.

This is why NexaBiome Life Sciences LTD is developing phages as human pharmaceuticals. The Company has a patented technology that allows them to deliver phages in a commercially viable way is working towards taking a product through clinical trials, with a first target exemplifying the technology being diabetic foot infections (DFIs).

The challenge

Phages are far more specific for their target bacteria than small molecule antibiotics.  This has advantages, as they can kill only the pathogens and leave the healthy microbiome intact, however this does make selecting the best phage or combination of phages (a “phage cocktail”) more challenging.   While phage cocktails can be developed that will target most infections, there will always be patients for whom a generic cocktail is not sufficient, and a more personalised approach is required.   For both generic cocktails and personalised preparations, phage selection is a complex and vital part of developing effective medicines.   In an infection, there are several complicating factors; variations in the target bacteria, the presence other commensal bacteria, the immune response, environmental factors and the actions of any phages naturally present.   This complexity represents the biggest obstacle in designing phage cocktails, even for highly specialised experts like NexaBiome Life Sciences LTD.

The proposal

NexaBiome Life Sciences LTD has a several in vitro methods for selecting phages for use in humans.  Each phage it uses has a wide array of test performed, including various standard operating procedures it has developed for measuring host range, growth and virulence.  While these are necessarily artificial, they give each phage a fingerprint that is unique to that strain.  The Company also have full genome sequences of all its phage, meaning that it has a detailed picture of the genotypic and phenotypic characteristics of each phage.  These measurements allow it to make a predictive determination of the best cocktail for a specific indication.  However, when these phages are introduced to complex environments, with multiple unique bacteria in varying ratios, its predictions for the best phage combination become relatively crude.  In order to allow the rapid and efficient selection of optimal phage for a specific indication, there needs to be the ability to link the genotype to the phenotype and predict how these phages will respond in a given environment; to be able to predict what the best phage(s) will be for a patient or group of patients based only on the sequence of the phage and the sequence of the full microbiome of the environment containing the target bacteria.

Being able to accurately predict a phage combination to use, based on microbiome data would have immense benefits both for the rational design of phage cocktails and the ability to develop new, personalised therapies to treat a number of issues, ranging from acute and chronic infections to microbiome imbalances.

NexaBiome Life Sciences LTD has an ambitious programme of work planned, and bioinformatics represents one of the main areas of focus. This resulted in the funding a PhD place with Dr. Leighton Pritchard and Dr. Suzie Humphrey at Strathclyde University, which will support implementation of the following:

• In-house, high-throughput, whole-genome next-generation sequencing of microorganisms via Oxford nanopore Technology (ONT) which is being implemented to generate large-scale genomic datasets, using ad-hoc procedures.
• A database is currently being designed for accessibility and storage of all data produced in vitro and in silico.
• Genomic analysis of all sequenced microorganisms is completed via a bioinformatic pipeline including tools for genome assembly, annotation, gene ontology, and phylogenetic analysis, ensuring uniformity and reproducibility of feature findings, allowing for a consistent at-scale generation of genomic data.
• Machine Learning (ML) models are trained using phenotypic and genotypic data, to predict phage activity and host range against specific bacterial hosts. Within individual ML models, the most significant variables are selected, statistically evaluated, and cross-validated, to assess their reliability.
• Data obtained from the ML models is implemented in the designing of phage cocktails, which is also tested in vitro against bacterial hosts which are not included within the training database. 

Ultimately, NexaBiome Life Sciences LTD expects to be able to compare/implement this workflow to practical efficacy in patients by working with collaborators at a new National Phage Service, which will uniquely allow it access to microbiome data before and after phage treatment and therefore allow it to further optimise its ML models and prediction based on real-world data. To support this next stage, NexaBiome Life Sciences LTD is expanding its computational capacity to meet the growing bioinformatic demands of this programme.