Machine Learning versus Statistical Inference in Microbial Genomics

My talk given today at the 2025 Microbiology Society Conference in Liverpool:

Abstract

The advent of vast genomic datasets has transformed microbiology, presenting opportunities and challenges for data analysis. The distinct philosophies of machine learning (ML) and statistical inference gives them complementarity strengths and weaknesses in tackling big data problems in pathogen research. While statistical inference prioritizes understanding underlying relationships, ML focuses on optimizing predictive performance. In this talk I will contrast the approaches and offer a view on their relative utility for three problems: source attribution, bacterial genome-wide association studies, and predicting antimicrobial resistance phenotypes from whole genome sequences.

Postdoctoral Scientist in Statistical Genomics

We are recruiting for a Postdoctoral Scientist in Statistical Genomics working on Antimicrobial Resistance (AMR) gene discovery and focused on Tuberculosis. This will be a joint position at the University of Oxford between Derrick Crook's group and mine, and part of the large international CRyPTIC consortium.

The role is for a population geneticist or statistical geneticist to develop and apply statistical methods, including genome-wide association studies, for discovering rare and common genetic variants underlying antimicrobial resistance in Mycobacterium tuberculosis.

One third of the world's population - 2.5 billion people - are thought to be infected with tuberculosis (TB). This post offers an opportunity to work with global TB experts from five continents, statistical geneticists, clinicians, medical statisticians and software engineers; integrating statistical genetics, bioinformatics and machine learning methods with the aim of uncovering all genomic variants causing at least 1% resistance to first line anti-TB drugs.

We're looking for candidates with a PhD in genomics, evolutionary biology, statistics or a related subject. The post is full-time and fixed-term for up to 3 years initially.

The deadline for applications is noon on Friday 6th May 2016.

Snow Monkeys in Japan

Recently got back from the SMBE Satellite meeting on Pathogen Genomics in Japan. The organizers did a fantastic job and the talks were great. There was also time to visit the Japanese macaques at Snow Monkey Park, where one of the little guys climbed on to my shoulders

Thanks Ashlee Earl for the video and Koji Yahara, Alan McNally and Nick Croucher for additional commentary!

Cheltenham Science Festival

Earlier this week members of the group represented the Nuffield Department of Medicine at the Cheltenham Science Festival with our Modernising Medical Microbiology stall, featuring the Antibiotic Resistance Coconut Shy and the Genome Evolution Dance Mat.

Antibiotic Resistance Coconut Shy

Antibiotic Resistance Coconut Shy: The children (and adults) visiting the stall were given five bean bags (antibiotics) to throw at the coconuts (bacterial pathogens) to try to knock them off. The front row of coconuts, representing bacteria more susceptible to antibiotics, were easier to knock off than the back row, which represented more resistant bacteria. The aim was to show the children that an unwanted side effect of using antibiotics is to increase the frequency of resistant bacteria, because they were usually the ones left standing.

The game was more difficult than it looks, and just one visitor knocked off all five coconuts. We gave out NDM pens to the sixty visitors who managed to knock off three or more.

Microscope and Top Trumps

Digital Microscope: We brought along a light microscope to show the children what bacteria really look like, which helps emphasize how small they are since they are difficult to see even under the highest magnification. We prepared slides for several Gram positive and Gram negative species, and provided a key to help identify them. We also brought along a number of games that have been used in previous departmental outreach activities, including Pathogen Top Trumps and Fact or Fiction.

Genome Evolution Dance Mat

Genome Evolution Dance Mat: In this game, the children had to copy a bacterial DNA sequence by replicating a sequence of dance moves (up=A, left=C, right=G, down=T) without introducing new errors (mutations). Any mutations that were introduced were passed on to the next template sequence. In this way we aimed to show how mutations occur by errors in DNA replication, and that they are inherited. This generates unique DNA fingerprints for bacteria, which we can use to track the spread of outbreaks.

Outbreak Map

The game, which was kindly programmed by Gareth Jenkin-Jones, included a form of natural selection, so that if too many errors were introduced at once, the sequence was considered inviable and did not survive to be passed on. There was also a speed control, which was handy since some people appear to have spent a lot more of their youth playing dance mats than others.

Outbreak Map: We made an Outbreak Map to show the reach of our stall over the day, with visitors that scored highly on the coconut shy pushing in pins to show where they had travelled from. Had we been handing out germs instead of pens, we could have started outbreaks as far afield as Edinburgh, France and Spain, as well as a large cluster in Cheltenham and the surrounding counties.

Other research groups are representing the department throughout the week.

NDM Microbiology Stall at the Cheltenham Science Festival (L-R): Sarah Earle, Louise Pankhurst, Danny Wilson, Liz Batty, Dilrini De Silva, Jess Hedge, Catrin Moore. Amy Mason, Gareth Jenkin-Jones and Jane Charlesworth also helped with the preparations, and Jen Bardsley co-ordinated all the NDM Stalls.

PLoS Pathogens Review Published!

Published today in PLoS Pathogens:

SMBE 2012: Microbial Genome Evolution Symposium

Along with my colleagues Xavier Didelot, Ed Feil, Eduardo Rocha and Howard Ochman, I will be organizing a symposium on Microbial Genome Evolution at the 2012 meeting of the Society for Molecular Biology and Evolution in Dublin, Ireland. The deadline for abstract submission is 27th January 2012. This is the synopsis for our symposium:

High-throughput sequencing makes it possible for the first time to sequence hundreds of microbial genomes rapidly at low cost. These methods have huge potential to significantly improve our understanding of microbial evolution, so that many research projects have recently been set up to generate and analyze such data. This symposium will provide an overview of the progress made by such projects, as well as the many challenges they pose. It is now possible to identify the vast majority of SNPs within large population samples of microbial isolates. These datasets are illuminating the molecular, ecological and population-level dynamic processes occurring over short time scales in natural populations inhabiting a range of habitats from the clinic to the environment. We aim to  explore these recent advances and the development of new methods of analyses required to fully exploit these extremely large sequence datasets. Relevant topics include quantifying the variation in the rates of recombination and mutation between closely related lineages, the evolution of base composition, the relative power of drift and selection, examining the acquisition of adaptive traits (e.g. antibiotic resistance, host adaptation, metabolic flexibility, regulatory changes) within a phylogenetic framework, and the distribution of variation over time and space (phylogeography). The role of phage and conjugative elements in structuring populations as both vehicles for gene flow and parasitic elements will also be considered. The symposium will focus on variation within natural populations rather than experimental evolution.

Campylobacter source attribution in New Zealand

What is the source of the common food poisoning pathogen Campylobacter jejuni was the subject of a paper published in September last year in PLoS Genetics by my colleagues and I, in which we traced the origin of bacterial isolates collected from patients in Lancashire, England. In that study, and a subsequent investigation into campylobacteriosis across Scotland, we found that the majority of cases could be attributed to populations of C. jejuni typically found in poultry.

Now Petra Mullner, Nigel French and colleagues have genetically characterized the C. jejuni populations found in human patients, cattle, sheep, poultry and environmental samples from New Zealand covering the period March 2005 - February 2008. What is special about their study is that the New Zealand poultry industry is a closed system, with no foreign imports, making it possible to directly sample the putative source populations and disease-causing isolates concurrently.

Like the studies in England and Scotland, poultry was the inferred source of the majority of disease in New Zealand. Uniquely however, it was possible to attribute cases separately to the three major poultry suppliers on the islands. One supplier in particular was attributed a disproportionate number of cases using 3 assignment methods, including my method (iSource, soon to be available on this website). Supported in part by this evidence, the New Zealand Food Safety Authority introduced mandatory targets for limiting Campylobacter contamination of poultry products in 2007. Remarkably, the number of cases fell from 15,873 in 2006 before the control measures were introduced to 6,689 in 2008. The next chapter of this intriguing story will be a follow-up study to establish whether the fall in the number of cases corresponded to a reduction in the proportion of campylobacteriosis attributable to poultry sources.