Journal: bioRxiv

Article Title: MethylBench: A comprehensive benchmark of DNA methylation profiling methods across diverse sequencing platforms

doi: 10.64898/2026.04.28.721268

Figure Lengend Snippet: This figure summarizes key quality control and performance metrics across five DNA methylation profiling techniques (ONT, PacBio, RRBS, TWIST, WGEC) applied to blood, fibroblast and GIAB samples. (A): Barplots showing the mean CpG coverage per sample and method; dashed lines indicate multiples of a 10× coverage threshold & numbers represent the rounded mean CpG coverage. (B): Mean methylation levels per sample and method, shown both unfiltered and with a 10× coverage cutoff, revealing reduced variance across platforms at higher coverage. In the GIAB cohort, the deviation observed for one PacBio sample is likely attributable to its markedly lower overall sequencing coverage, reflecting technology-specific sensitivity of PacBio methylation calling to reduced molecule sampling rather than insufficient CpG-level filtering. (C): Log-scale line plot of the number of overlapping CpG sites retained at increasing coverage thresholds, indicating a rapid decline in shared CpGs with stricter filters. (D): Dot plot comparing the number of unique reads obtained per sample-method combination, with long-read methods showing fewer but larger alignments.

Article Snippet: We compared six widely used technologies — Illumina EPIC array, TWIST, Whole-Genome Enzymatic Conversion (WGEC), Reduced Representation Bisulfite Sequencing (RRBS), long-read genome sequencing (LR-GS) with Pacific Biosciences (PacBio) and Oxford Nanopore Technologies (ONT) — using Genome in a Bottle (GIAB) reference samples and ten samples derived of blood and fibroblast cultures of 5 individuals.

Techniques: Control, DNA Methylation Assay, Methylation, Sequencing, Sampling

Journal: bioRxiv

Article Title: MethylBench: A comprehensive benchmark of DNA methylation profiling methods across diverse sequencing platforms

doi: 10.64898/2026.04.28.721268

Figure Lengend Snippet: Barplot showing the mean genomic coverage per method over all samples; dashed lines indicate multiples of a 10× coverage threshold & numbers represent the rounded mean genomic coverage. For ONT, RRBS, WGEC & TWIST, n Samples = 12 and PacBio n Samples = 2.

Article Snippet: We compared six widely used technologies — Illumina EPIC array, TWIST, Whole-Genome Enzymatic Conversion (WGEC), Reduced Representation Bisulfite Sequencing (RRBS), long-read genome sequencing (LR-GS) with Pacific Biosciences (PacBio) and Oxford Nanopore Technologies (ONT) — using Genome in a Bottle (GIAB) reference samples and ten samples derived of blood and fibroblast cultures of 5 individuals.

Techniques:

Journal: bioRxiv

Article Title: MethylBench: A comprehensive benchmark of DNA methylation profiling methods across diverse sequencing platforms

doi: 10.64898/2026.04.28.721268

Figure Lengend Snippet: (A): Pearson correlation coefficients of CpG methylation levels between all pairwise combinations of five methylation profiling methods (ONT, RRBS, TWIST, WGEC and PacBio) across different samples. PacBio data was only available for the GIAB sample. Correlations were computed at increasing minimum coverage thresholds (x-axis) and are shown separately for each sample. A consistent increase in inter-method correlation is observed with rising coverage cutoffs, particularly in comparisons involving ONT and RRBS, indicating that higher coverage thresholds reduce noise and improve measurement agreement between technologies. In the GIAB2 sample, correlations involving PacBio plateau despite increasing coverage thresholds, indicating that residual inter-platform differences persist even at high PacBio coverage and are not primarily driven by low-coverage CpGs. (B): Direct comparison between ONT and TWIST on both GIAB samples. Applying progressively stricter coverage filters revealed a rapid increase in correlation up to around 15–20x, after which values plateaued near r = 0.98 − 0.99.

Article Snippet: We compared six widely used technologies — Illumina EPIC array, TWIST, Whole-Genome Enzymatic Conversion (WGEC), Reduced Representation Bisulfite Sequencing (RRBS), long-read genome sequencing (LR-GS) with Pacific Biosciences (PacBio) and Oxford Nanopore Technologies (ONT) — using Genome in a Bottle (GIAB) reference samples and ten samples derived of blood and fibroblast cultures of 5 individuals.

Techniques: CpG Methylation Assay, Methylation, Comparison

Journal: bioRxiv

Article Title: MethylBench: A comprehensive benchmark of DNA methylation profiling methods across diverse sequencing platforms

doi: 10.64898/2026.04.28.721268

Figure Lengend Snippet: (A): The figure shows methylation density distributions in the ranges from 0-30% and 70-100% methylation from one fibroblast, one blood and one GIAB sample profiled with four different methylation detection methods (ONT, RRBS, TWIST, WGEC and PacBio(only GIAB)). Methylation values without coverage filtering are shown as filled ridgelines with black outlines, while red contours indicate methylation distributions after applying a 10× coverage threshold. After filtering, methylation values cluster more distinctly near 0 and 1, supporting the biological expectation of largely unmethylated or fully methylated CpGs. (B): Panel B employs the same analytical framework as subfigure A, focusing exclusively on ONT, PacBio and TWIST methodologies to facilitate more rigorous coverage filtering. The bimodal distribution pattern, with peaks at 0% and 94-98% methylation, indicates cellular heterogeneity in differentiation states, accounting for the absence of perfect 0% or 100% methylation levels across the population. As representative samples, the samples with the highest mean coverage where selected. Fibro := Fibroblast sample 4, Blood := Blood sample 3, GIAB := GIAB sample 2.

Article Snippet: We compared six widely used technologies — Illumina EPIC array, TWIST, Whole-Genome Enzymatic Conversion (WGEC), Reduced Representation Bisulfite Sequencing (RRBS), long-read genome sequencing (LR-GS) with Pacific Biosciences (PacBio) and Oxford Nanopore Technologies (ONT) — using Genome in a Bottle (GIAB) reference samples and ten samples derived of blood and fibroblast cultures of 5 individuals.

Techniques: Methylation

Journal: bioRxiv

Article Title: MethylBench: A comprehensive benchmark of DNA methylation profiling methods across diverse sequencing platforms

doi: 10.64898/2026.04.28.721268

Figure Lengend Snippet: Principal Component Analysis (PCA) plots showing methylation profiles of five blood (subfigures A, C & E) and five fibroblast (subfigures B,D & F) samples, measured with five different technologies (ONT, RRBS, TWIST, WGEC and EPIC). Panels are organized by tissue type (columns) and inclusion of EPIC data (rows). Each point represents a sample-method combination, colored by platform and shaped by sample ID.

Article Snippet: We compared six widely used technologies — Illumina EPIC array, TWIST, Whole-Genome Enzymatic Conversion (WGEC), Reduced Representation Bisulfite Sequencing (RRBS), long-read genome sequencing (LR-GS) with Pacific Biosciences (PacBio) and Oxford Nanopore Technologies (ONT) — using Genome in a Bottle (GIAB) reference samples and ten samples derived of blood and fibroblast cultures of 5 individuals.

Techniques: Methylation

Journal: bioRxiv

Article Title: MethylBench: A comprehensive benchmark of DNA methylation profiling methods across diverse sequencing platforms

doi: 10.64898/2026.04.28.721268

Figure Lengend Snippet: UpSet plot showing the overlap of significantly differentially methylated CpG sites (DMCs) between blood and fibroblast samples across EPIC, TWIST, WGEC, RRBS and ONT datasets, as identified using the limma model.

Article Snippet: We compared six widely used technologies — Illumina EPIC array, TWIST, Whole-Genome Enzymatic Conversion (WGEC), Reduced Representation Bisulfite Sequencing (RRBS), long-read genome sequencing (LR-GS) with Pacific Biosciences (PacBio) and Oxford Nanopore Technologies (ONT) — using Genome in a Bottle (GIAB) reference samples and ten samples derived of blood and fibroblast cultures of 5 individuals.

Techniques: Methylation

Journal: bioRxiv

Article Title: MethylBench: A comprehensive benchmark of DNA methylation profiling methods across diverse sequencing platforms

doi: 10.64898/2026.04.28.721268

Figure Lengend Snippet: (A): UpSet plot showing the overlap of significantly DMCs between blood and fibroblast samples, as identified using a wilcoxon test. Only EPIC and TWIST identify significant DMCs, demonstrating the reduced statistical power of ONT, RRBS and WGEC under non-parametric testing. (B): Cross-platform scatter plot comparing Δ β values between EPIC and TWIST. CpGs are colored by concordance: green indicates CpGs significant and directionally consistent in both platforms, red and blue mark EPIC- or TWIST-specific DMCs, respectively. The strong correlation (r = 0.96) indicates high cross-platform reproducibility under a variance-robust test. (C): Boxplots showing the mean coverage of CpGs included in each platform’s DMC set. Sequencing-based assays display broader coverage variability compared to EPIC, reflecting differences in sequencing depth and genomic sampling. (D): Variance of CpG methylation in blood and fibroblast samples across platforms. EPIC exhibits the lowest within-group variance due to array normalization, whereas sequencing-based platforms show higher dispersion linked to read depth and CpG representation.

Article Snippet: We compared six widely used technologies — Illumina EPIC array, TWIST, Whole-Genome Enzymatic Conversion (WGEC), Reduced Representation Bisulfite Sequencing (RRBS), long-read genome sequencing (LR-GS) with Pacific Biosciences (PacBio) and Oxford Nanopore Technologies (ONT) — using Genome in a Bottle (GIAB) reference samples and ten samples derived of blood and fibroblast cultures of 5 individuals.

Techniques: Sequencing, Sampling, CpG Methylation Assay, Dispersion

Journal: bioRxiv

Article Title: Centromeres are hotspots of cytosine methylation epimutations in a filamentous fungus

doi: 10.64898/2026.03.03.709258

Figure Lengend Snippet: Samples taken from the two MA pedigrees for different experiments. A) Ten MA lines were sampled from the two pedigrees. Samples for sequencing were taken after 5, 20, and 40 transfers and DNA methylation was detected using bisulfite sequencing (red circles). B) Three additional MA lines were sampled from both pedigrees and samples were taken after 1, 5, 7, 8, 10, and 15 transfers for Nanopore sequencing and after 5, 8, 10, 20, and 40 transfers for H3K9me3 ChIP-seq.

Article Snippet: For whole genome bisulfite sequencing, DNA samples were sent to Novogene (Cambridge, UK).

Techniques: Sequencing, DNA Methylation Assay, Methylation Sequencing, Nanopore Sequencing, ChIP-sequencing

Journal: bioRxiv

Article Title: Centromeres are hotspots of cytosine methylation epimutations in a filamentous fungus

doi: 10.64898/2026.03.03.709258

Figure Lengend Snippet: Pairwise divergence in cytosine methylation patterns among the MA lines and their ancestors. X-axis shows the number of mitoses separating the two lines that are being compared, and the y-axis shows cytosine methylation divergence. Red line shows a neutral epimutation accumulation model fitted to all data, orange line shows a neutral epimutation accumulation model fitted to data with outlier samples removed, and solid blue line shows a logistic model fitted to all data, and the dotted line shows null model of no increase in divergence. A) The divergence calculated from single cytosines, for CG, CHG, and CHH sequence motifs. B) The divergence calculated for DMRs in different sequence contexts, as in panel A.

Article Snippet: For whole genome bisulfite sequencing, DNA samples were sent to Novogene (Cambridge, UK).

Techniques: Methylation, Sequencing

Journal: BMC Plant Biology

Article Title: Insights into epigenetics suggest a role of DNA methylation in regulating Chinese yam tuber shape

doi: 10.1186/s12870-026-08438-5

Figure Lengend Snippet: Whole-genome DNA methylation analysis of Chinese yam variant F2000 and F60. A Principal component 1 and principal component 2 analyses were performed for each sample. B Methylation level distribution across genetic features including 4 kb upstream of TSS, gene body, and 4 kb downstream of TTS in the three DNA methylation contexts CG, CHG, and CHH. Analyses were performed in biological triplicates for each variant. C Number of DMRs with higher methylation levels in F2000 (hypermethylated) and in F60 (hypomethylated). DMRs were investigated in the three DNA methylation contexts CG, CHG and CHH. Distribution of DMRs upstream in the putative promoter (4 kb of TSS), within the gene body, and down-stream in the putative terminator (4 kb of TTS) were analyzed in DMR-associated genes (q-value ≤ 0.01, methylation difference >|10%|). D Venn analysis of DMR-associated genes split in hyper- and hypomethylated promoters, gene bodies, and terminators. DMR, differentially methylated region; TSS, transcription start site; TTS, transcription termination site

Article Snippet: Prepared DNA samples were sent to Novogene Company (Beijing, China) for low-input whole-genome bisulfite sequencing including the steps of bisulfite treatment and library construction.

Techniques: DNA Methylation Assay, Variant Assay, Methylation