Search

Annotation of top-ranked DMPs using epigenomic reference datasets. The genomic locus on chromosome 14q24.2 (position = 70,261,006–70,349,114; genome build = hg19) harboring the top-ranked DMR is shown using the WashU Epigenome Browser v40.0.0 ( http://epigenomegateway.wustl.edu/browser/ ). The top-ranked DMR containing CpG classifiers of aGVHD severity (Table 2) is located at a CpG island (position = 70,316,847–70,317,240; indicated with an orange arrow). RefSeq and Gencode v17 genes, as well as CpG islands, are shown in the bottom panel of the figure. A total of 50 epigenomic reference tracks provided by the NIH Roadmap Epigenomics Project are displayed. Specifically, we show both the primary and imputed chromatin state maps in 22 primary hematopoietic cell types. The highlighted DMR overlaps with an active transcription start site (red) or poised promoter (pink) in G-CSF-mobilized CD34+ hematopoietic stem cells, and a Polycomb-repressed region in CD3+ T cells of peripheral blood. (PDF 960 kb)

Ontology annotation of genes flanking DMRs using GREAT. We report the ontology of genes flanking the 453 identified DMRs. Specifically, we indicate all enriched terms in the binomial test over genomic regions at an FDR of less than 5Â %. (PDF 107 kb)

Quality assessment of the Illumina Infinium HumanMethylation450 assay. The quality of the 450K array data was assessed after normalization, probe filtering, and batch correction (for details see Methods). (a) Density of DNA methylation β-values. (b) Multidimensional scaling (MDS) indicates the similarities and differences of samples by calculating the Euclidean distances between samples based on all CpG sites, and then projecting these distances into 2D coordinates. We found Dimension 4 to associate with aGVHD severity, accounting for 3 % of the total variance. HSCT donors matched to healthy recipients and those matched to recipients diagnosed with mild aGVHD could not be stratified using MDS. Therefore, these two sample groups were combined for subsequent analysis. (c) Singular value decomposition (SVD) determines the nature of the largest components of variation (Teschendorff AE et al. PLoS One. 2009;4:e8274). We assessed the first six principal components (PCs), and correlated these to phenotypic factors of donors (e.g., sex, age at transplant, and CMV serostatus), phenotypic factors of recipients (e.g., aGVHD status and severity), factors related to the experimental setup (e.g., Sentrix ID, sample plate, and sample well), as well as internal control parameters (e.g., bisulfite conversion efficiency). DNA methylation age (‘DNAm age’) was predicted based on the raw DNA methylation data using the DNA Methylation Age Calculator ( https://dnamage.genetics.ucla.edu/ ), as described by Horvath (Horvath S. Genome Biol. 2013;14:R115). The phenotypic factor ‘transplant date’ denotes the year of the day of the graft transplant (day 0). We found PC4 and PC5 to most strongly correlate with aGVHD severity, achieving a significance level of P

Box-and-whisker plots of DNA methylation values in graft donors in the discovery cohort assessed using MethyLight and 450K arrays. Only HSCT donors of the discovery cohort that were profiled on both assay platforms are shown. We identified a DNA hypomethylation phenotype at the top-ranked DMP cg20475486 in graft donors matched to recipients with severe aGVHD. HSCT donors matched to healthy recipients and those matched to recipients diagnosed with mild aGVHD could not be discriminated. (PDF 256 kb)

PCR primers and probes used in MethyLight replication experiments. A total of three MethyLight reactions were designed, which targeted top-ranked DMPs associated with aGVHD severity: cg10399005, cg20475486, and cg07280807. Of these reactions, cg20475486 achieved the highest PCR efficiency and was subsequently used in validation experiments (highlighted in gray). Start and end positions of primer and probes are noted in relation to the design start coordinates. Chromosomal positions are reported in genome buildâ =â hg19. (PDF 110 kb)

PCR primers and probes used in MethyLight replication experiments. A total of three MethyLight reactions were designed, which targeted top-ranked DMPs associated with aGVHD severity: cg10399005, cg20475486, and cg07280807. Of these reactions, cg20475486 achieved the highest PCR efficiency and was subsequently used in validation experiments (highlighted in gray). Start and end positions of primer and probes are noted in relation to the design start coordinates. Chromosomal positions are reported in genome buildâ =â hg19. (PDF 110 kb)