ABSTRACT Discovery of genetic variants underlying bacterial phenotypes and the prediction of phenotypes such as antibiotic resistance are fundamental tasks in bacterial genomics. Genome-wide association study (GWAS) methods have been applied to study these relations, but the plastic nature of bacterial genomes and the clonal structure of bacterial populations creates challenges. We introduce an alignment-free method which finds sets of loci associated with bacterial phenotypes, quantifies the total effect of genetics on the phenotype, and allows accurate phenotype prediction, all within a single computationally scalable joint modeling framework. Genetic variants covering the entire pangenome are compactly represented by extended DNA sequence words known as unitigs, and model fitting is achieved using elastic net penalization, an extension of standard multiple regression. Using an extensive set of state-of-the-art bacterial population genomic data sets, we demonstrate that our approach performs accurate phenotype prediction, comparable to popular machine learning methods, while retaining both interpretability and computational efficiency. Compared to those of previous approaches, which test each genotype-phenotype association separately for each variant and apply a significance threshold, the variants selected by our joint modeling approach overlap substantially. IMPORTANCE Being able to identify the genetic variants responsible for specific bacterial phenotypes has been the goal of bacterial genetics since its inception and is fundamental to our current level of understanding of bacteria. This identification has been based primarily on painstaking experimentation, but the availability of large data sets of whole genomes with associated phenotype metadata promises to revolutionize this approach, not least for important clinical phenotypes that are not amenable to laboratory analysis. These models of phenotype-genotype association can in the future be used for rapid prediction of clinically important phenotypes such as antibiotic resistance and virulence by rapid-turnaround or point-of-care tests. However, despite much effort being put into adapting genome-wide association study (GWAS) approaches to cope with bacterium-specific problems, such as strong population structure and horizontal gene exchange, current approaches are not yet optimal. We describe a method that advances methodology for both association and generation of portable prediction models.

Adaptive sampling is a method of software-controlled enrichment unique to nanopore sequencing platforms. To test its potential for enrichment of rarer species within metagenomic samples, we create a synthetic mock community and construct sequencing libraries with a range of mean read lengths. Enrichment is up to 13.87-fold for the least abundant species in the longest read length library; factoring in reduced yields from rejecting molecules the calculated efficiency raises this to 4.93-fold. Finally, we introduce a mathematical model of enrichment based on molecule length and relative abundance, whose predictions correlate strongly with mock and complex real-world microbial communities.

Background: Adaptive sampling is a method of software-controlled enrichment unique to nanopore sequencing platforms recently implemented in Oxford Nanopore's own control software. By examining the first few hundred bases of a DNA molecule as it passes through a pore, software can determine if the molecule is sufficiently interesting to sequence in its entirety. If not, the molecule is ejected from the pore by reversing the voltage across it, freeing the pore for a new molecule. User supplied sequences define the targets to be sequenced or ejected. Here we explore the potential of using adaptive sampling for enrichment of rarer species within metagenomic samples. Results: We created a synthetic mock community consisting of seven bacterial species at different proportions ranging from 1.2% to 47% and used this as the basis for a series of enrichment and depletion experiments. To investigate the effect of DNA length on adaptive sampling efficiency, we created sequencing libraries with mean read lengths of 1.7 kbp, 4.7 kbp, 10.6 kbp, and 12.8 kbp and enriched or depleted for individual and multiple species over a series of sequencing runs. Across all experiments enrichment ranged from 1.67-fold for the most abundant species with the shortest read length to 13.87-fold for the least abundant species with the longest read length. Factoring in the reduction to sequence output associated with repeatedly rejecting molecules reduces the calculated efficiency of this enrichment to between 0.96-fold and 4.93-fold. We note that reducing ejections due to false negatives (approximately 36%) would significantly increase efficiency. We used the relationship between abundance, molecule length and enrichment factor to produce a mathematical model of enrichment based on molecule length and relative abundance, whose predictions correlated strongly with experimental data. A web application is provided to allow researchers to explore model predictions in advance of performing their own experiments. Conclusions: Our data clearly demonstrates the benefit for enriching low abundant species in adaptive sampling metagenomic experiments, especially with longer molecules, and our mathematical model can be used to determine whether a given experimental DNA sample is suitable for adaptive sampling. Notably, repeated voltage reversals have no effect on pore stability.