Funding
Funded
AI-Driven Microbiome Profiling for Predicting Ocular Infection Severity
School of Medicine, Dentistry and Biomedical Sciences | PHD
Applications are now CLOSED
Reference Number
SMED-2251-1030
Application Deadline
14 April 2025
Start Date
1 October 2025
Overview
This project is a 4-year PhD project with enhanced training and 3+ month placement, which is funded by UKRI BBSRC through the NI Landscape Partnership in AI for Bioscience (NILAB) Programme, delivered by Queen’s University Belfast and Ulster University. Details of the enhanced training will be available later at qub.ac.uk/nilab/. NILAB aims to bridge the gap between biology and artificial intelligence to accelerate bioscience discovery and foster effective collaboration between academia, industrial partners, and government bodies. NILAB’s mission is to train the next generation of researchers to develop and use AI to uncover the rules of life, addressing challenges in human health, animal welfare, and sustainable food systems.
This NILab project offers a unique opportunity for a fully funded PhD studentship in AI-powered bioscience, focused on improving the diagnosis and management of microbial keratitis—a leading cause of blindness. Our project will apply 16S rRNA sequencing and machine learning techniques to predict infection severity, addressing limitations in current diagnostic methods and ultimately improving patient outcomes.
As a PhD student, you will have the chance to work with cutting-edge bioinformatics software tools like KRAKEN2 and programming machine learning models such as neural networks and XGBoost. You will process genomic sequencing data, integrate clinical metadata, and validate your models with external datasets to ensure their clinical relevance and robustness. Your work will contribute to the development of an AI model capable of predicting infection severity based on microbial community profiles.
This project places a strong emphasis on training and development. You will receive hands-on experience with state-of-the-art sequencing platforms, as well as training in bioinformatics, machine learning, and diagnostic protocol development. In addition to these technical skills, you will learn to think critically, solve problems independently, and collaborate within a multidisciplinary research team.
We are seeking a highly motivated student who is eager to learn and contribute to impactful research. This PhD offers a supportive environment for developing the skills necessary to succeed in both academic and industry settings, with ample opportunities for professional growth and collaboration.
Scientific background
Ocular infections, such as keratitis, can lead to significant vision loss if not treated effectively. Microbial keratitis is a leading cause of unilateral blindness and eye-loss worldwide. However, current diagnostic methods, such as bacterial culture, are time-consuming and often inconclusive, with up to half of cases yielding no diagnostic information. In contrast, 16S rRNA sequencing offers a more comprehensive microbial profile and can be processed more rapidly, potentially within hours using advanced PCR and sequencing techniques. The challenge lies in interpreting these complex genomic profiles to predict the risk of severe cases (i.e., those that are difficult to treat and likely to lead to long-term vision loss). By identifying these cases early, clinicians can prioritize more active management to improve outcomes. Preliminary analyses suggest a correlation between microbial abundances and clinical outcomes, and research in other infection areas has shown that AI models can be used to successfully interpret these genomic data, making it a promising approach for improving diagnostic accuracy in ocular infections.
Aim and Objectives:
The core aim of this project is to predict the severity of ocular infections using 16S rRNA sequencing data, utilising artificial intelligence (AI) to uncover patterns that have the potential to transform diagnostic approaches in healthcare. The research will follow these key objectives:
1. Data Curation and Preprocessing You'll start by working with 16S rRNA sequencing data, utilising bioinformatics tools like KRAKEN2 for taxonomic classification and microbial abundance quantification. By integrating this sequencing data with clinical metadata (such as infection severity, treatment, and recovery), you’ll help create an extensive dataset that will form the basis for machine learning models.
2. Feature Selection and Model Development The next step will involve the evaluation of several AI models, including neural networks, random forests, and XGBoost to pinpoint effective architectures that best predict which microbial taxa correlate with infection severity. You’ll also implement advanced techniques like Weighted Gene Correlation Network Analysis (WGCNA) to uncover co-expressed microbial features. Through this process, you’ll refine feature selection and optimize AI model performance, developing a deeper understanding of how specific microbial patterns drive infection outcomes.
3. Model Validation To ensure the AI model's reliability and applicability, you’ll validate its predictions using cross-validation and external datasets. You’ll assess performance using metrics such as ROC analysis, AUC, and precision-recall curves to confirm the model’s ability to make accurate, real-world predictions.
4. Development of Rapid 16S Sequencing Protocols You’ll also have the opportunity to integrate rapid PCR and real-time 16S rRNA sequencing protocols into the workflow. This aims to accelerate clinical turnaround times, enabling faster AI-driven predictions and more efficient decision-making.
Methodology
Building on an existing dataset of 61 keratitis cases, you’ll further develop the model using an additional existing 400 keratitis cases from our collaborator, Professor of Opthalmology Stephen Kaye at the University of Liverpool; As part of this project, >1,000 corneal samples are to be sequenced using the QUB Genomics CTU Oxford Nanopore Technologies long-read sequencing platform. These novel data will contribute to model development, training, and testing. Independently generated public data from sources like the Gene Expression Omnibus (GEO) will also be used to confirm model robustness. Should additional genomic data types—such as metagenomics for antimicrobial resistance or direct RNA sequencing for modification profiling—become available, they will be incorporated into the project to enhance the AI model’s predictive capabilities.
Upon completion of this project, you will be equipped with valuable skills and expertise that will enable you to tackle complex challenges and drive continued innovation in AI and genomics for health research in academic and industrial settings.
Funding Information
This project is open to both home and international applicants on a competitive basis. Top ranked students will be offered their choice of projects to which they apply. Funding is available for both domestic and international students with 30% of the funding allocated for international students.
Fees: Queen’s standard
Stipend: £20,780 (2025/2026)
Project Summary
Supervisor
Dr David Simpson
Mode of Study
Full-time: 4 Years
Funding Body
UKRI BBSRC through NILAB Programme