Biology subjects

Pierce-Ward, N. T.

Publications and source records attributed to Pierce-Ward, N. T..

5 recordsLinked to original sources

bioRxiv2022

Meta-analysis of metagenomes via machine learning and assembly graphs reveals strain switches in Crohn's disease

Microbial strains have closely related genomes but may have different phenotypes in the same environment. Shotgun metagenomic sequencing can capture the genomes of all strains present in a community but strain-resolved analysis from shotgun sequencing alone remains difficult. We developed an approach to identify and interrogate strain-level differences in groups of metagenomes. We use this approach to perform a meta-analysis of stool microbiomes from individuals with and without inflammatory bowel disease (IBD; Crohns disease, ulcerative colitis; n = 605), a disease for which there are not specific microbial biomarkers but some evidence that microbial strain variation may stratify by disease state. We first developed a machine learning classifier based on compressed representations of complete metagenomes (FracMinHash sketches) and identified genomes that correlate with IBD subtype. To rescue variation that may not have been present in the genomes, we then used assembly graph genome queries to recover strain variation for correlated genomes. Lastly, we developed a novel differential abundance framework that works directly on the assembly graph to uncover all sequence variants correlated with IBD. We refer to this approach as dominating set differential abundance analysis and have implemented it in the spacegraphcats software package. Using this approach, we identified five bacterial strains that are associated with Crohns disease. Our method captures variation within the entire sequencing data set, allowing for discovery of previously hidden disease associations.

bioinformatics↗

bioRxiv2022

Protein k-mers enable assembly-free microbial metapangenomics

An estimated 2 billion species of microbes exist on Earth with orders of magnitude more strains. Microbial pangenomes are created by aggregating all genomes of a single clade and reflect the metabolic diversity of groups of organisms. As de novo metagenome analysis techniques have matured and reference genome databases have expanded, metapangenome analysis has risen in popularity as a tool to organize the functional potential of organisms in relation to the environment from which those organisms were sampled. However, the reliance on assembly and binning or on reference databases often leaves substantial portions of metagenomes unanalyzed, thereby underestimating the functional potential of a community. To address this challenge, we present a method for metapangenomics that relies on amino acid k-mers (kaa-mers) and metagenome assembly graph queries. To enable this method, we first show that kaa-mers estimate pangenome characteristics and that open reading frames can be accurately predicted from short shotgun sequencing reads using the previously developed tool orpheum. These techniques enable pangenomics to be performed directly on short sequencing reads. To enable metapangenome analysis, we combine these approaches with compact de Bruijn assembly graph queries to directly generate sets of sequencing reads for a specific species from a metagenome. When applied to stool metagenomes from an individual receiving antibiotics over time, we show that these approaches identify strain fluctuations that coincide with antibiotic exposure.

bioinformatics↗

bioRxiv2022

Evaluation of taxonomic profiling methods for long-read shotgun metagenomic sequencing datasets

BackgroundLong-read shotgun metagenomic sequencing is gaining in popularity and offers many advantages over short-read sequencing. The higher information content in long reads is useful for a variety of metagenomics analyses, including taxonomic classification and profiling. The development of long-read specific tools for taxonomic classification is accelerating, yet there is a lack of information regarding their relative performance. Here, we perform a critical benchmarking study using 11 methods, including five methods designed specifically for long reads. We applied these tools to several mock community datasets generated using Pacific Biosciences (PacBio) HiFi or Oxford Nanopore Technology (ONT) sequencing, and evaluated their performance based on read utilization, detection metrics, and relative abundance estimates. ResultsOur results show that long-read classifiers generally performed best. Several short-read classification and profiling methods produced many false positives (particularly at lower abundances), required heavy filtering to achieve acceptable precision (at the cost of reduced recall), and produced inaccurate abundance estimates. By contrast, two long-read methods (BugSeq, MEGAN-LR & DIAMOND) and one generalized method (sourmash) displayed high precision and recall without any filtering required. Furthermore, in the PacBio HiFi datasets these methods detected all species down to the 0.1% abundance level with high precision. Some long-read methods, such as MetaMaps and MMseqs2, required moderate filtering to reduce false positives to resemble the precision and recall of the top-performing methods. We found read quality affected performance for methods relying on protein prediction or exact k-mer matching, and these methods performed better with PacBio HiFi datasets. We also found that long-read datasets with a large proportion of shorter reads (<2kb length) resulted in lower precision and worse abundance estimates, relative to length-filtered datasets. Finally, for classification methods, we found that the long-read datasets produced significantly better results than short-read datasets, demonstrating clear advantages for long-read metagenomic sequencing. ConclusionsOur critical assessment of available methods provides best-practice recommendations for current research using long reads and establishes a baseline for future benchmarking studies.

bioinformatics↗

bioRxiv2022

Lightweight compositional analysis of metagenomes with FracMinHash and minimum metagenome covers

The identification of reference genomes and taxonomic labels from metagenome data underlies many microbiome studies. Here we describe two algorithms for compositional analysis of metagenome sequencing data. We first investigate the FracMinHash sketching technique, a derivative of modulo hash that supports Jaccard containment estimation between sets of different sizes. We implement FracMinHash in the sourmash software, evaluate its accuracy, and demonstrate large-scale containment searches of metagenomes using 700,000 microbial reference genomes. We next frame shotgun metagenome compositional analysis as the problem of finding a minimum collection of reference genomes that "cover" the known k-mers in a metagenome, a minimum set cover problem. We implement a greedy approximate solution using FracMinHash sketches, and evaluate its accuracy for taxonomic assignment using a CAMI community benchmark. Finally, we show that the minimum metagenome cover can be used to guide the selection of reference genomes for read mapping. sourmash is available as open source software under the BSD 3-Clause license at github.com/dib-lab/sourmash/.

bioinformatics↗

bioRxiv2022

Debiasing FracMinHash and deriving confidence intervals for mutation rates across a wide range of evolutionary distances

Sketching methods offer computational biologists scalable techniques to analyze data sets that continue to grow in size. MinHash is one such technique to estimate set similarity that has enjoyed recent broad application. However, traditional MinHash has previously been shown to perform poorly when applied to sets of very dissimilar sizes. FracMinHash was recently introduced as a modification of MinHash to compensate for this lack of performance when set sizes differ. This approach has been successfully applied to metagenomic taxonomic profiling in the widely used tool sourmash gather. While experimental evidence has been encouraging, FracMinHash has not yet been analyzed from a theoretical perspective. In this paper, we perform such an analysis to derive various statistics of FracMinHash, and prove that while FracMinHash is not unbiased (in the sense that its expected value is not equal to the quantity it attempts to estimate), this bias is easily corrected for both the containment and Jaccard index versions. Next, we show how FracMinHash can be used to compute point estimates as well as confidence intervals for evolutionary mutation distance between a pair of sequences by assuming a simple mutation model. We also investigate edge cases where these analyses may fail, to effectively warn the users of FracMinHash indicating the likelihood of such cases. Our analyses show that FracMinHash estimates the containment of a genome in a large metagenome more accurately and more precisely when compared to traditional MinHash, and the point estimates and confidence intervals perform significantly better in estimating mutation distances.

bioinformatics↗