Developing tools to predict antimicrobial resistance gene and mobile genetic element profiles from metagenomic analysis and demographic data

The goal of my PhD is to create a tool to predict the profiles of antimicrobial resistance (AMR) in humans using insights from metagenomic data, and other demographics, such as lifestyle, location and socio-economic status. The PhD will include multiple investigations into different metrics for the prediction tool that can act as independent projects, making this PhD strategy flexible and robust.
The World Health Organisation describes AMR as "one of the biggest threats to global health, food security, and development today". Common bacterial infections are becoming untreatable by antibiotics due to antibiotic resistance, especially in countries such as India and China, where antibiotic over-use and misuse are common. AMR genes have not only been identified from AMR infections, but also in saliva and stool samples from healthy individuals (Diaz-Torres et al., 2006, Pehrsson et al., 2016). Microbes in healthy individuals may acquire AMR genes through: association with mobile genetic elements (MGE) that undergo horizontal gene transfer (HGT) from microbes in the environment, especially from overcrowded living conditions, poor sanitation and farming where antimicrobial drugs are constantly administered to animals; and misuse of antibiotics from a previous infection i.e. not completing the course. Quantifying AMR status of individuals across the globe could have a significant impact on targeting policies most rapidly to those areas with populations at high risk of AMR, for example, earlier surveillance of potential AMR infections and regulating the use of certain antibiotics.
To be able to treat infections quickly and efficiently, health professionals need to prescribe antibiotics that will be completely effective after a first course of treatment. Since AMR is becoming so prevalent, it is important for healthcare professionals to know whether an individual is likely to be at risk of expressing a certain AMR gene and whether this gene could lead to an antibiotic being ineffective against an infection. If this is known, alternative antibiotics or even bacteriophage 1 treatments could be used instead as a first course of treatment. So far, there is no universal profiling tool for AMR. However, research into AMR gene prediction tools and AMR severity scores for worldwide surveillance is already underway by Prof Andrew McArthur and collaborators in Hamilton, Canada, and draft software will be created within the year (personal correspondence). Because of this, I feel it is necessary to focus on a prediction tool that can benefit future diagnosis as well as surveillance.
I plan to create a tool that will aim to predict an individual's AMR profile and association with MGEs based on 1) parameters of microbial diversity: species abundance, richness and co-occurrence, 2) demographics, such as profession and location, and 3) how this information could be applied to antimicrobial resistance diagnosis in the future.