Comprehensive Evaluation of The Infinium Human MethylationEPIC v2 BeadChip - PubMed (original) (raw)
doi: 10.1186/s43682-023-00021-5. Epub 2023 Sep 27.
Sol Moe Lee 1 2, David Goldberg 1, Nathan J Spix 3, Toshinori Hinoue 3, Hong-Tao Li 4, Varun B Dwaraka 5, Ryan Smith 5, Hui Shen 3, Gangning Liang 4 6, Nicole Renke 7, Peter W Laird 3, Wanding Zhou 1 8
Affiliations
- PMID: 38455390
- PMCID: PMC10919401
- DOI: 10.1186/s43682-023-00021-5
Comprehensive Evaluation of The Infinium Human MethylationEPIC v2 BeadChip
Diljeet Kaur et al. Epigenetics Commun. 2023.
Abstract
Infinium Methylation BeadChips are widely used to profile DNA cytosine modifications in large cohort studies for reasons of cost-effectiveness, accurate quantification, and user-friendly data analysis in characterizing these canonical epigenetic marks. In this work, we conducted a comprehensive evaluation of the updated Infinium MethylationEPIC v2 BeadChip (EPICv2). Our evaluation revealed that EPICv2 offers significant improvements over its predecessors, including expanded enhancer coverage, applicability to diverse ancestry groups, support for low-input DNA down to one nanogram, coverage of existing epigenetic clocks, cell type deconvolution panels, and human trait associations, while maintaining accuracy and reproducibility. Using EPICv2, we were able to identify epigenome and sequence signatures in cell line models of DNMT and SETD2 loss and/or hypomorphism. Furthermore, we provided probe-wise evaluation and annotation to facilitate the use of new features on this array for studying the interplay between somatic mutations and epigenetic landscape in cancer genomics. In conclusion, EPICv2 provides researchers with a valuable tool for studying epigenetic modifications and their role in development and disease.
Conflict of interest statement
Competing interests WZ received BeadChips from Illumina Inc. to conduct this analysis. NR is an employee of Illumina Inc.
Figures
Figure 1:
Enhanced probe mapping and applicability of EPICv2 in diverse human populations. (A) Total number of Infinium DNA methylation BeadChip studies and deposited datasets in GEO. (B) Probe counts for HM450, EPICv1, EPICv2, and unique prefix counts for EPICv2. Abbreviations: “cg”: CpG cytosine methylation probes; “ch”: non-CG cytosine methylation probes; “rs”: common SNP probes; “nv”: probes for somatic mutations found in cancer; and “ct”: quality control probes. (C) Venn diagram illustrating the percentage of EPICv2 probes retained from predecessor arrays. (D) Infinium-I and Infinium-II chemistry ratios for EPICv1 and EPICv2 probes. EPICv2 data is from all probes, same as panel B. (E) Infinium-I and Infinium-II chemistry ratios for shared EPICv1-v2 probes and exclusive EPICv1/EPICv2 probes. (F) Mapping quality of EPICv1 and EPICv2 probes, differentiated by allele A and allele B. (G) Proportion of probes masked due to ancestry-specific SNP overlaps. Abbreviations: AFR, African population; AMR, Admixed American; EAS, East Asian; EUR, European; SAS, South Asian. (H) Percentage of probes with cross-reactivity and sequence polymorphism influence issues, comparing shared EPICv1-EPICv2 and EPICv1-only probes.
Figure 2:
Assessment of EPICv2 reproducibility between sample replicates and replicate probes. (A) Methylation measurement correlations between technical replicates of GM12878, LNCaP, K562, and HCT116 cell lines. (B) Comparison of Spearman’s rank correlation coefficients (rho) between technical and non-technical replicates. (C) Lower inter-cell line correlation for newly added EPICv2 probes, indicating increased discriminatory power. Arrows represent DNA input from high to low. (D) Methylation measurement correlation using EPICv1 and EPICv2 on four human cell lines. (E) Comparison of EPICv1-EPICv2 design switches and probes with identical sequences in both platforms. (F) Number of loci with multiple probe replication coverage. (G) Correlations among replicate probes compared to non-replicate probes, emphasizing probe design robustness.
Figure 3:
EPICv2 reveals DNA methylation variation in wild type cell lines and cell line models of epigenetic modifiers. (A) EPICv2-WGBS correlation between the same and different cell lines. (B) Comparison of EPICv2-measured DNA methylation level with whole genome bisulfite sequencing (WGBS) on GM12878, LNCaP and K562. (C) Comparison of EPICv2 accuracy on cell line DNA of known titrated methylation fractions. (D) Correlation of EPICv2 probes with known DNA methylation level in titration experiment. (E) Distribution of probes by correlation with titrated fraction. (F) Functional analysis of poorly correlated probes. (G) Drop of global DNA methylation levels in HCT116-derived cell lines with mutated or deleted DNMTs and/or SETD2. (H) Enrichment of CpGs that retain DNA methylation in DKO1 cells. (I) Sequence context of loss of DNA methylation in DNMT1KO cells, stratified by common partially methylated domains (PMDs) vs. common highly methylated domains (HMDs), and CpGs flanked by A/T (W) or G/C (S).
Figure 4:
EPICv2 performance at low input ranges. Scatter plot (A) and heatmap (B) illustrating probe success rates for various input amounts and cell lines. (C) Correlation coefficient between replicates comparing 500 sorted cells (left) and 5000 sorted cells (right). (D) Correlation between low input (from 100ng to 1ng) and 250ng DNA input samples. (E) Correlation between low input (5000 and 500 sorted cells) and 250ng DNA input samples. (F) Distribution of probe detection success rate across genomic regions for different input amounts. (G) tSNE analysis of beta values for low and high input samples, using all probes or only probes added in EPICv2 (subpanel). Labeled the number corresponds to the input amount (ng). Input amounts from sorted cells are estimated assuming 6pg DNA per cell.
Figure 5:
Annotation of EPICv2 probes for epigenetic clocks and cell type deconvolution panels. (A) Odds ratios of EPICv2-vs-EPICv1 across different ChromHMM features (ENCODE v2, Methods). (B) Probe coverage comparison between EPICv2 and previous human methylation BeadChips for various applications including methylation clocks (B), epigenome-wide association studies (EWAS) (C), and cell-type deconvolution panels (D). Dashed line represent expected overlap from random probe selection for deletion. (E) Number of distinct pairwise contrasts covered by EPICv2 probes based on WGBS-derived human cell type panels. (F) Percentage of contrasts covered by varying numbers of probes, showcasing the EPICv2’s ability to capture signature of diverse cell types.
Figure 6:
EPICv2 facilitates somatic mutation analysis in cancer. (A) Distribution of nv probes among different Infinium design types. (B) Pie chart displaying genes targeted by nv probes. (C) Location of TP53 mutations targeted by nv probes. (D) EPICv2 reading of probes targeting KRAS G13 mutations in HCT116 cells. The following probes query the displayed mutations: nv-GRCh38-chr12–25245347-25245347-C-A_BC11 (G13V), GRCh38-chr12–25245347-25245347-C-T_BC11 (G13D), GRCh38-chr12–25245348-25245348-C-T_BC11 (G13S), GRCh38-chr12–25245347-25245347-C-G_TC11 (G13A); (E) Effect of the number of CpGs within 10bp of the 3’-end on total intensities of nv probes. (F) Detection failure rate comparison between nv probes and other probe types. (G) Copy number profile of K562 cells, showing chromosome 9 deletion and chromosome 22 amplification. (H) Copy number profile of LNCaP cells, showing chromosome 2 and 13 deletions.
References
- Greenberg MVC, Bourc’his D. The diverse roles of DNA methylation in mammalian development and disease. Nat Rev Mol Cell Biol. 2019;20:590–607. - PubMed
LinkOut - more resources
Full Text Sources
Other Literature Sources
Molecular Biology Databases