PaReBrick: PArallel REarrangements and BReaks identification toolkit

Machine learning and phylogenetic analysis allow for predicting antibiotic resistance in M. tuberculosis

BACKGROUND

Antimicrobial resistance (AMR) poses a significant global health threat, and an accurate prediction of bacterial resistance patterns is critical for effective treatment and control strategies. In recent years, machine learning (ML) approaches have emerged as powerful tools for analyzing large-scale bacterial AMR data. However, ML methods often ignore evolutionary relationships among bacterial strains, which can greatly impact performance of the ML methods, especially if resistance-associated features are attempted to be detected. Genome-wide association studies (GWAS) methods like linear mixed models accounts for the evolutionary relationships in bacteria, but they uncover only highly significant variants which have already been reported in literature.

RESULTS

In this work, we introduce a novel phylogeny-related parallelism score (PRPS), which measures whether a certain feature is correlated with the population structure of a set of samples. We demonstrate that PRPS can be used, in combination with SVM- and random forest-based models, to reduce the number of features in the analysis, while simultaneously increasing models' performance. We applied our pipeline to publicly available AMR data from PATRIC database for Mycobacterium tuberculosis against six common antibiotics.

CONCLUSIONS

Using our pipeline, we re-discovered known resistance-associated mutations as well as new candidate mutations which can be related to resistance and not previously reported in the literature. We demonstrated that taking into account phylogenetic relationships not only improves the model performance, but also yields more biologically relevant predicted most contributing resistance markers.

Population genomic analyses suggest recent dispersal events of the pathogen Cercospora zeina into East and Southern African maize cropping systems

A serious factor hampering global maize production is gray leaf spot disease. Cercospora zeina is one of the causative pathogens, but population genomics analysis of C. zeina is lacking. We conducted whole-genome Illumina sequencing of a representative set of 30 C. zeina isolates from Kenya and Uganda (East Africa) and Zambia, Zimbabwe, and South Africa (Southern Africa). Selection of the diverse set was based on microsatellite data from a larger collection of the pathogen. Pangenome analysis of the C. zeina isolates was done by (1) de novo assembly of the reads with SPAdes, (2) annotation with BRAKER, and (3) protein clustering with OrthoFinder. A published long-read assembly of C. zeina (CMW25467) from Zambia was included and annotated using the same pipeline. This analysis revealed 790 non-shared accessory and 10,677 shared core orthogroups (genes) between the 31 isolates. Accessory gene content was largely shared between isolates from all countries, with a few genes unique to populations from Southern Africa (32) or East Africa (6). There was a significantly higher proportion of effector genes in the accessory secretome (44%) compared to the core secretome (24%). PCA, ADMIXTURE, and phylogenetic analysis using a neighbor-net network indicated a population structure with a geographical subdivision between the East African isolates and the Southern African isolates, although gene flow was also evident. The small pangenome and partial population differentiation indicated recent dispersal of C. zeina into Africa, possibly from 2 regional founder populations, followed by recurrent gene flow owing to widespread maize production across sub-Saharan Africa.

Systematic identification of gene-altering programmed inversions across the bacterial domain

Programmed chromosomal inversions allow bacteria to generate intra-population genotypic and functional heterogeneity, a bet-hedging strategy important in changing environments. Some programmed inversions modify coding sequences, producing different alleles in several gene families, most notably in specificity-determining genes such as Type I restriction-modification systems, where systematic searches revealed cross phylum abundance. Yet, a broad, gene-independent, systematic search for gene-altering programmed inversions has been absent, and little is known about their genomic sequence attributes and prevalence across gene families. Here, identifying intra-species variation in genomes of over 35 000 species, we develop a predictive model of gene-altering inversions, revealing key attributes of their genomic sequence attributes, including gene-pseudogene size asymmetry and orientation bias. The model predicted over 11,000 gene-altering loci covering known targeted gene families, as well as novel targeted families including Type II restriction-modification systems, a protein of unknown function, and a fusion-protein containing conjugative-pilus and phage tail domains. Publicly available long-read sequencing datasets validated representatives of these newly predicted inversion-targeted gene families, confirming intra-population genetic heterogeneity. Together, these results reveal gene-altering programmed inversions as a key strategy adopted across the bacterial domain, and highlight programmed inversions that modify Type II restriction-modification systems as a possible new mechanism for maintaining intra-population heterogeneity.