What I do:
I run the Pathogen Informatics and Modelling group at EMBL-EBI. My previous work has attempted to understand the pathogenesis and transmission of infectious diseases. I primarily works on statistical genetics, genomic epidemiology and genome evolution of pathogenic bacteria. I have also worked on infectious disease modelling, GPU algorithms, and visualisation. I write methods and software to scale these analyses and make them more accessible to others in attempts to democratise bioinformatics.
- Mathematical modeling:
We are creating some stochastic, mechanistic models of competition and transmission. We’re using tools developed during the COVID-19 pandemic to better understand the transmission of bacterial pathogens, and eventually aim to combine models of within-host pathogen evolution with models of between-host transmission (bacterial phylodynamics).
Examples: odin.dust framework
- Real-time genomic epidemiology
Public databases of pathogen genome variation have grown rapidly, with the largest having surpassed one million sequences. As these sequence databases grow, they are becoming more difficult for many researchers to take full advantage of. We are developing methods which help local surveillance labs integrate their data into large sequence databases, using a ‘one-by-one’ analysis approach.
Examples: PopPUNK for genomic epidemiology, see the paper here; PopPIPE for transmission analysis; ggCaller for annotation.
- Pathogen evolution and statistical genetics
We are designing tools to find evolutionary signatures in the masses of genomic data available, and link these findings to function.
We are also developing automated tools to mark and track concerning lineages as they emerge, predict antimicrobial resistance status, and observed the effects of vaccination on local populations. We’ve got a long standing interest in genome-wide association studies, and continue to develop new methods in this area.
Examples: pyseer for GWAS and phenotype prediction; applying GWAS to improve vaccine design.
- Sequencing within-host diversity
Pathogen populations also evolve within a single host, sometimes developing mutations with consequences for the whole population. We will develop tools which combine population genetic knowledge, fast informatics approaches, and flexible sequence to streamline the process of sequencing diversity directly from complex samples.
Examples: within-host diversity in meningitis patients.
- GPU algorithms
The rate of genomic data growth has outpaced the rate of computational capacity for a number of years. GPUs, with tens of thousands of processing cores, offer a promising solution. Faster algorithms will allow rapid analysis suitable for real-time surveillance of pathogens, and more ambitious analyses of larger datasets, yielding greater discovery power. We aim to address scalability of bioinformatics and mathematical modelling by programming efficient algorithms to run on GPUs, hundreds of times faster than their traditional counterparts.
Examples: in modelling; in microbial genomics; in visualisation.
- Democratising bioinformatics
We aim to keep all of our research useful, reusable and accessible – this guides many of our design decisions in the above themes. On top of this, we have specific projects which aim to advance open science, making our research accessible to as many people as possible.
Projects include: searching and indexing genome analysis and metadata; developing WebAssembly versions of tools which both keep user data private, and are easily run in a web browser; computational biology outreach and teaching.
Examples: Bacquerya, AMR prediction tool for S. pneumoniae
