Oxford Nanopore Technologies (ONT) long-read sequencing (LRS) has emerged as a promising genomic analysis tool, yet comprehensive benchmarks with established platforms across diverse datasets remain limited. This study aimed to benchmark LRS performance against Illumina short-read sequencing (SRS) and microarrays for variant detection across different genomic contexts and to evaluate the impact of experimental factors. We sequenced 14 human genomes using the three platforms and evaluated single nucleotide variants (SNVs), insertions/deletions (indels), and structural variants (SVs) detection, stratifying by high-complexity, low-complexity, and dark genome regions while assessing effects of multiplexing, depth, and read length. LRS SNV accuracy was slightly lower than that of SRS in high-complexity regions (F-measure: 0.954 vs. 0.967) but showed comparable sensitivity in low-complexity regions. LRS showed robust performance for small (1–5 bp) indels in high-complexity regions (F-measure: 0.869), but SRS agreement decreased significantly in low-complexity regions and for larger indel sizes. Within dark regions, LRS identified more indels than SRS, but showed lower base-level accuracy. LRS identified 2.86 times more SVs than SRS, excelling at detecting large variants (>6 kb), with SV detection improving with sequencing depth. Sequencing depth strongly influenced variant calling performance, whereas multiplexing effects were minimal. Our findings provide valuable insights for optimising LRS applications in genomic research and diagnostics.
Despite the advances in genetic marker identification associated with severe COVID-19, the full genetic characterisation of the disease remains elusive. This study explores imputation in low-coverage whole genome sequencing for a severe COVID-19 patient cohort. We generated a dataset of 79 imputed variant call format files using the GLIMPSE1 tool, each containing an average of 9.5 million single nucleotide variants. Validation revealed a high imputation accuracy (squared Pearson correlation ≈0.97) across sequencing platforms, showcasing GLIMPSE1’s ability to confidently impute variants with minor allele frequencies as low as 2% in individuals with Spanish ancestry. We carried out a comprehensive analysis of the patient cohort, examining hospitalisation and intensive care utilisation, sex and age-based differences, and clinical phenotypes using a standardised set of medical terms developed to characterise severe COVID-19 symptoms. The methods and findings presented here can be leveraged for future genomic projects to gain vital insights into health challenges like COVID-19.
Oxford Nanopore Technologies (ONT) long-read sequencing (LRS) has emerged as a promising tool for genomic analysis, but comprehensive comparisons with established platforms across diverse datasets remain limited. We present a multi-platform benchmark using 14 human genomes sequenced with ONT LRS, Illumina short-read sequencing (SRS), and Illumina microarrays. Our study evaluates LRS performance for various genetic variants across genomic contexts, while also examining the impact of experimental factors such as multiplexing, depth, and read length. In high-complexity regions, LRS demonstrated competitive yet slightly lower accuracy than SRS for SNV detection (F-measure: 0.954 vs. 0.968), with performance gaps narrowing in low-complexity regions. For indel detection, LRS showed robust performance for small indels (1-5bp) in high-complexity regions (F-measure: 0.869), but accuracy decreased significantly in low-complexity regions and for larger indels. LRS identified 2.86 times more structural variants than SRS, with superior detection of large-scale variations. Sequencing depth strongly influenced variant calling performance across all variant types, while multiplexing effects were minimal after controlling for depth. Our findings provide valuable insights for optimising ONT LRS applications in genomic research and clinical diagnostics.
Despite advances in identifying genetic markers associated to severe COVID-19, the full genetic characterisation of the disease remains elusive. This study explores the use of imputation in low-coverage whole genome sequencing for a severe COVID-19 patient cohort. We generated a dataset of 79 imputed variant call format files using the GLIMPSE1 tool, each containing an average of 9.5 million single nucleotide variants. Validation revealed a high imputation accuracy (squared Pearson correlation ≈0.97) across sequencing platforms, showing GLIMPSE1’s ability to confidently impute variants with minor allele frequencies as low as 2% in Spanish ancestry individuals. We conducted a comprehensive analysis of the patient cohort, examining hospitalisation and intensive care utilisation, sex and age-based differences, and clinical phenotypes using a standardised set of medical terms developed to characterise severe COVID-19 symptoms. The methods and findings presented here may be leveraged in future genomic projects, providing vital insights for health challenges like COVID-19.
Background: Mobile element insertions, particularly transposable elements (TEs) such as Alu, LINE-1 (L1), SVA, and endogenous retroviruses (ERVs), represent a major source of human genetic variation and have been implicated in evolution, genomic instability, and disease. Although long-read sequencing generally outperforms short-read sequencing for the characterisation of such elements, their accurate detection with long-reads remains challenging, with different computational tools adopting varying approaches and producing divergent call sets. As gold standards currently do not exist for TE detection, benchmarking these methods is essential to understand their strengths, limitations, and biases. Here, we systematically evaluate the performance of available state-of-the-art TE detection tools on both simulated and real human genome data using highly characterised samples from the Genome in a Bottle consortium, population level reference databases and an in-house collection for which matching short-read sequencing data are available. Results: Our results show significant differences in calling strategies, leading to substantial variation in precision, recall, and the spectrum of TE families detected across tools. Our benchmark also displays the differences between short-read and long-read calls, highlighting the importance of appropriate method selection. Conclusions: The benchmarking results presented here will aid TE researchers make better informed decisions on which tool to use in their long-read TE analyses. Strengths and limitations of different tools have been highlighted in depth as well as their computational requirements, which will result in less time spent finding the best tool for the job and promote faster TE research.