Olin K. Silander

Gene loss is common, yet whether it is merely an evolutionary dead-end or can promote future adaptation to new environments remains debated. Here, we systematically tested whether harmful loss of metabolic genes, followed by compensatory evolution, generates preadaptations to alternative nutrients in Escherichia coli. Our analysis revealed a striking outcome beyond simple growth restoration, where repair of a genetic defect yielded growth exceeding the unperturbed wild type on carbon sources not encountered during evolution. Such latent growth gains occurred across multiple deletion backgrounds and carbon sources, with growth rate increases up to 46% relative to the wild type. In some cases, the corresponding ancestral deletion strain failed to grow on these carbon sources, indicating functional innovations relative to the deletion ancestor. Genetic reconstruction and transcriptomic profiling implicate regulatory reprogramming, particularly in carbon catabolite repression and nutrient uptake, as mechanisms underlying latent growth gains. In conclusion, evolutionary recovery from gene loss can rapidly produce preadaptations to future environments in the absence of direct selection.

Genome-wide significant loci explain only part of the heritable signal for most complex traits, leaving a broad sub-significant polygenic background whose organisation is largely uncharacterised. Whether this background is largely diffuse noise or contains coherent structure shared across traits has been difficult to test, because SNP-level effects are noisy and LD-correlated, and conventional cross-trait analyses have exclusively focused on GWAS significance-filtered, LD-pruned variants. Here, using 403 GWAS summary statistics, we show that coherent cross-trait signal is detectable using signed regional medians of SNP effects. Singular value decomposition of regionally averaged effects recovered latent phenotypic axes. Focusing on the sub-significant polygenic background, we found that the axes persisted after removal of all genome-wide significant loci and flanking regions observed in any of the 403 GWAS analysed, a step which eliminated more than 75% of genomic loci. The dominant axis of trait variation, primarily metabolic, but spanning anthropometric, musculoskeletal, and cognitive traits, was unlike secondary axes in being enriched for regions more than 250 kb from any gene body, both before and after removal of genome-wide significant loci observed in these GWAS. These results show that the sub-significant polygenic background carries structured cross-trait information, and identify gene-distal non-coding variation as a defining feature of the dominant axis of human trait covariation, patterns that are largely inaccessible to approaches that rely on genome-wide significance for variant prioritisation.

RNA modifications are important for RNA structure, stability, and ribosome function, but their identification and localisation remains challenging. Oxford Nanopore direct RNA sequencing (DRS) enables modification-agnostic detection in native RNA, but existing tool benchmarks have focused almost exclusively on m6A in eukaryotic mRNA, leaving multi-modification tool performance in bacterial systems largely untested. Here, we benchmark ten RNA modification detection tools spanning signal-comparison, error-rate, and hybrid approaches on Escherichia coli K-12 MG1655 16S and 23S rRNA, which harbour 11 and 25 known modified sites, respectively, across 17 modification types. Using native RNA and in vitro transcribed (IVT) unmodified RNA, we evaluate performance across 25 coverage levels (5× to 1000×). DiffErr and JACUSA2 showed the strongest discrimination performance (AUROC >0.9 on both 16S and 23S rRNA), with DiffErr achieving the highest F1 score on 16S and JACUSA2 showing the most consistent precision-recall balance across both rRNAs. Both tools achieved full transcript-wide scoring and, along with DRUMMER, exact positional localisation. Several other tools produced no output at many rRNA positions, and restricting evaluation to reported positions inflated apparent performance. Signal-based tools showed a systematic 1-4 nucleotide 5ʹ offset from known modified positions, consistent with the ∼5-mer nucleotide stretch present in the read head of the nanopore; applying tool-specific offset corrections substantially improved per-site recovery and reduced false positives, substantially improving the performance of tools such as EpiNano and nanoDoc. At single-site resolution, no known modified site was recovered by all tools, and several m5C, m5U, and m6A sites were missed by the majority of tools. Tool combination analysis showed that pairing error-rate-based tools with offset-corrected signal-based tools improved site recovery beyond any individual tool, with the best three-tool combination recovering 30 of the 36 known sites while maintaining low false positive rates. These results establish that discrimination metrics (e.g. AUROC) alone are insufficient to evaluate modification detection tools: output completeness, positional precision, and per-modification-type sensitivity should be reported alongside standard benchmarking metrics.

The hourglass dolphin (Lagenorhynchus cruciger) is a small cetacean species of the Southern Ocean, with significance to iwi Māori (Māori tribes) of Aotearoa New Zealand as taonga (treasured/valued). Due to the remoteness and difficulty of surveying Antarctic waters, it remains one of the least-studied dolphin species. A recent stranding of an hourglass dolphin represented a rare opportunity to generate a genome assembly as a resource for future study into the conservation and evolutionary biology of this species. In this study, we present a high-quality genome assembly of an hourglass dolphin individual using a single sequencing platform, Oxford Nanopore Technologies, coupled with computationally efficient assembly methods. Our assembly strategy yielded a genome of high contiguity (N50 of 8.07 Mbp) and quality (98.3% BUSCO completeness). Compared to other Delphinoidea reference genomes, this assembly has fewer missing BUSCOs than any except Orcinus orca, more single-copy complete BUSCOs than any except Phocoena sinus, and 20% fewer duplicated BUSCOs than the average Delphinoidea reference genome. This suggests that it is one of the most complete and accurate marine mammal genomes to date. This study showcases the feasibility of a cost-effective mammalian genome assembly method, allowing for genomic data generation outside the traditional confines of academia and/or resource-rich genome assembly hubs, and facilitating the ability to uphold Indigenous data sovereignty. In the future, the genome assembly presented here will allow valuable insights into the past population size changes, adaptation, vulnerability to future climate change of the hourglass dolphin and related species.

Random Forest models are widely used in genomic data analysis and can offer insights into complex biological mechanisms, particularly when features influence the target in interactive, non-linear, or non-additive ways. Currently, some of the most efficient Random Forest methods in terms of computational speed are implemented in Python. However, many biologists use R for genomic data analysis, as R offers a unified platform for performing additional statistical analysis and visualization. Here we present an R package, pyRforest, which integrates Python scikit-learn `RandomForestClassifier` algorithms into the R environment. pyRforest inherits the efficient memory management and parallelization of Python, and is optimized for classification tasks on large genomic datasets, such as those from RNA-seq. pyRforest offers several additional capabilities, including a novel rank-based permutation method for biomarker identification. This method can be used to estimate and visualize p-values for individual features, allowing the researcher to identify a subset of features for which there is robust statistical evidence of an effect. In addition, pyRforest includes methods for the calculation and visualization of SHapley ADditive Explanations (SHAP) values. Finally, pyRforest includes support for comprehensive downstream analysis for gene ontology and pathway enrichment. pyRforest thus improves the implementation and interpretability of Random Forest models for genomic data analysis by merging the strengths of Python with R. pyRforest can be downloaded at: https://www.github.com/tkolisnik/pyRforest with an associated vignette at https://github.com/tkolisnik/pyRforest/blob/main/vignettes/pyRforest-vignette.pdf.

The hourglass dolphin ( Lagenorhynchus cruciger ) is a small cetacean species of the Southern Ocean, with significance to iwi Māori ( Māori tribes ) of Aotearoa New Zealand as taonga ( treasured/valued ). Due to the remoteness and difficulty of surveying Antarctic waters, it remains one of the least-studied dolphin species. A recent stranding of an hourglass dolphin represented a rare opportunity to generate a genome assembly as a resource for future study into the conservation and evolutionary biology of this species. In this study, we present a high-quality genome assembly of an hourglass dolphin individual using a single sequencing platform, Oxford Nanopore Technologies, coupled with computationally efficient assembly methods. Our assembly strategy yielded a genome of high contiguity (N50 of 8.07 Mbp) and quality (98.3% BUSCO completeness). Compared to other Delphinoidea reference genomes, this assembly has fewer missing BUSCOs than any except Orcinus orca , more single-copy complete BUSCOs than any except Phocoena sinus , and 20% fewer duplicated BUSCOs than the average Delphinoidea reference genome. This suggests that it is one of the most complete and accurate marine mammal genomes to date. This study showcases the feasibility of a cost-effective mammalian genome assembly method, allowing for genomic data generation outside the traditional confines of academia and/or resource-rich genome assembly hubs, and facilitating the ability to uphold Indigenous data sovereignty. In the future the genome assembly presented here will allow valuable insights into the past population size changes, adaptation, vulnerability to future climate change of the hourglass dolphin and related species.

Accurate determination of animal diets is difficult. Methods such as molecular barcoding or metagenomics offer a promising approach allowing quantitative and sensitive detection of different taxa. Here we show that rapid and inexpensive quantification of animal, plant, and fungal content from stomach contents is possible through metagenomic sequencing with the portable Oxford Nanopore Technologies (ONT) MinION. Using an amplification-free approach, we profile the stomach contents from 24 wild-caught rats.

The gold standard treatment for locally advanced rectal cancer is total mesorectal excision after preoperative chemoradiotherapy. Response to chemoradiotherapy varies, with some patients completely responding to the treatment and some failing to respond at all. Identifying biomarkers of response to chemoradiotherapy could allow patients to avoid unnecessary treatment-associated morbidity rate. While previous studies have attempted to identify such biomarkers, none have reached clinical utility, which may be due to heterogeneity of the cancer. In this study, potential human gene and microbial biomarkers were explored in a cohort of rectal cancer patients who underwent chemoradiotherapy. RNA sequencing was carried out on matched tumour and adjacent normal rectum biopsies from patients with rectal cancer with varying chemoradiotherapy responses treated between 2016 and 2019 at two institutions. Enriched genes and microbes from tumours of complete responders were compared with those from tumours of others with lesser response. In 39 patients analysed, enriched gene sets in complete responders indicate involvement of immune responses, including immunoglobulin production, B cell activation and response to bacteria (adjusted P values Ruminococcaceae bacterium and Bacteroides thetaiotaomicron were documented to be abundant in tumours of complete responders compared with all other patients (adjusted P value <0.100). These results identify potential genetic and microbial biomarkers of response to chemoradiotherapy in rectal cancer, as well as suggesting a potential mechanism of complete response to chemoradiotherapy that may benefit further testing in the laboratory.

Quantifying SARS-like coronavirus (SL-CoV) evolution is critical to understanding the origins of SARS-CoV-2 and the molecular processes that could underlie future epidemic viruses. While genomic analyses suggest recombination was a factor in the emergence of SARS-CoV-2, few studies have quantified recombination rates among SL-CoVs. Here, we infer recombination rates of SL-CoVs from correlated substitutions in sequencing data using a coalescent model with recombination. Our computationally-efficient, non-phylogenetic method infers recombination parameters of both sampled sequences and the unsampled gene pools with which they recombine. We apply this approach to infer recombination parameters for a range of positive-sense RNA viruses. We then analyze a set of 191 SL-CoV sequences (including SARS-CoV-2) and find that ORF1ab and S genes frequently undergo recombination. We identify which SL-CoV sequence clusters have recombined with shared gene pools, and show that these pools have distinct structures and high recombination rates, with multiple recombination events occurring per synonymous substitution. We find that individual genes have recombined with different viral reservoirs. By decoupling contributions from mutation and recombination, we recover the phylogeny of non-recombined portions for many of these SL-CoVs, including the position of SARS-CoV-2 in this clonal phylogeny. Lastly, by analyzing >400,000 SARS-CoV-2 whole genome sequences, we show current diversity levels are insufficient to infer the within-population recombination rate of the virus since the pandemic began. Our work offers new methods for inferring recombination rates in RNA viruses with implications for understanding recombination in SARS-CoV-2 evolution and the structure of clonal relationships and gene pools shaping its origins.

DNA methylation in bacteria frequently serves as a simple immune system, allowing recognition of DNA from foreign sources, such as phages or selfish genetic elements. However, DNA methylation also affects other cell phenotypes in a heritable manner (i.e. epigenetically). While there are several examples of methylation affecting transcription in an epigenetic manner in highly localized contexts, it is not well-established how frequently methylation serves a more general epigenetic function over larger genomic scales. To address this question, here we use Oxford Nanopore sequencing to profile DNA modification marks in three natural isolates of Escherichia coli. We first identify the DNA sequence motifs targeted by the methyltransferases in each strain. We then quantify the frequency of methylation at each of these motifs across the entire genome in different growth conditions. We find that motifs in specific regions of the genome consistently exhibit high or low levels of methylation. Furthermore, we show that there are replicable and consistent differences in methylated regions across different growth conditions. This suggests that during growth, E. coli transiently differentiate into distinct methylation states that depend on the growth state, raising the possibility that measuring DNA methylation alone can be used to infer bacterial growth states without additional information such as transcriptome or proteome data. These results show the utility of using Oxford Nanopore sequencing as an economic means to infer DNA methylation status. They also provide new insights into the dynamics of methylation during bacterial growth and provide evidence of differentiated cell states, a transient analog to what is observed in the differentiation of cell types in multicellular organisms.