John A. Lees

ALLCAPS is a deep-learning tool for in silico pneumococcal serotyping that outperforms existing methods and identifies novel serotypes.

This paper introduces a Bayesian ordered logistic regression GWAS model for bacterial antimicrobial resistance phenotypes, showing improved variant discovery and MIC prediction by accounting for threshold-dependent effects.

Proteome-wide variant calling builds amino acid phylogenomic alignments orders of magnitude faster than a BUSCO marker pipeline, with similar tree signal in three of four test taxa.

The paper develops a mathematical model to predict vaccine replacement dynamics in Streptococcus pneumoniae and identifies optimal genomic surveillance strategies.

The paper introduces CABBAGE, a large-scale AMR genotype-phenotype database linking bacterial genomes to antibiotic susceptibility data and metadata, enabling analysis of resistance trends and predictive method benchmarks.

VIDRA models genetic dose-response relationships to predict drug mechanisms, therapeutic potential, and adverse events using phenotypically associated variants.

The paper introduces a fast clustering tool for bacterial genomes, enabling efficient analysis of millions of genomes and revealing taxonomic insights.

This paper trains species-specific transformer models for bacterial genomic surveillance, showing they can cluster genomes, identify emerging lineages, and predict antibiotic resistance gene uptake.

This study characterizes the global population structure and genetic diversity of serotype 12F Streptococcus pneumoniae, identifying strand-slippage mutations as a mechanism for capsule variation and vaccine evasion.

The paper develops a model to analyze pangenome dynamics and finds that differences in bacterial pangenomes are mainly due to variation in the number of rapidly exchanged genes.

This study tracks the 28-year evolution of a Burkholderia pseudomallei strain in sterile water, identifying genetic changes and metabolic adaptations to nutrient-depletion.

This study shows that pneumococcal growth kinetics are influenced by capsular serotype, lineage, and genetic background rather than specific loci.

LexicMap efficiently aligns sequences against millions of prokaryotic genomes, achieving comparable accuracy to existing tools with faster speed and lower memory use.

Trained on over 1.3 million bacterial genomes represented as ordered protein sequences, the transformer foundation model Bacformer predicts protein interactions, operons, essentiality, and phenotypes while generating trait-conditioned synthetic proteomes.

Deacon efficiently filters and depletes contaminant sequences from large genomic datasets using a human pangenome index, achieving high accuracy and speed with minimal computational resources.

PopPIPE clusters bacterial populations and detects transmission events using reference-free methods, outperforming traditional approaches in nosocomial outbreak analysis.

This paper introduces ska lo, a reference-free variant calling tool that uses a coloured De Bruijn graph to detect SNPs, insertions, deletions, and recombination events in pathogen genomes.

This paper introduces a selection-free CRISPR-Cas9 editing protocol effective across multiple Dictyostelid species, enabling efficient knock-out and knock-in mutations without antibiotic selection.

This paper compares four clustering methods for S. pneumoniae genomes to assess their accuracy in reflecting population structure and highlights the need for standardized nomenclature.

PAN-GWES identifies co-selected and epistatic genomic variations in bacterial genomes using de Bruijn graphs to analyze drug resistance and adaptation traits.

CELEBRIMBOR improves core gene identification in metagenome-assembled genomes by adjusting frequency thresholds based on genome completeness.

This paper optimizes machine learning prediction of MICs in Klebsiella pneumoniae by evaluating regression vs classification approaches based on antibiotic concentration levels.

This paper introduces SKA2, a method for rapid and accurate genotyping of bacterial populations, showing superior variant recall and no false positives in outbreak simulations.

A uniformly processed set of 2,440,377 bacterial and archaeal genomes with search tools and annotations, plus an AI-mined peptide set yielding ATB20, active in a mouse skin infection model.

GNASTy improves pneumococcal serotyping by using a pangenome graph algorithm with Nanopore Adaptive Sampling, enhancing sensitivity for novel serotypes in complex samples.

fastlin rapidly and accurately types Mycobacterium tuberculosis complex lineages from sequencing data, detecting mixed-lineage strains and estimating proportions.

ggCaller provides a fast and accurate method for pangenome annotation and clustering by integrating gene prediction, annotation, and clustering using de Bruijn Graphs, outperforming existing tools in speed and accuracy.

The paper introduces iterative-PopPUNK, an extension of PopPUNK that improves flexibility and interpretability in bacterial genome clustering and genealogical inference.

Mandrake reduces the dimensionality of large microbial genome datasets to visualize population structure, using both a pipeline and an interactive web application.

The paper reconstructs the evolutionary histories of antibiotic-resistant pneumococcal lineages PMEN3 and PMEN9, showing how interspecies recombination contributes to antibiotic resistance and serotype switching.