-Part of the Imprinting track collection, these tracks show an atlas of allele-specific +Part of the Imprinting track collection, these tracks show an atlas of allele-specific methylation built from deep whole-genome bisulfite sequencing of 202 purified samples covering about 40 human cell types from 135 donors. Because the sequencing was deep and the reads long enough to cover several CpG sites at once, the authors could look at one DNA fragment at a time and ask whether a region carried two distinct methylation states rather than one intermediate average. Regions that do are called bimodal here. Where a nearby heterozygous SNP is available, the two states can be assigned to the two copies of the chromosome, which separates the cases where methylation follows the DNA sequence from the much rarer cases where it follows the parent the copy came from.
The collection has three annotation sets:
The authors published these annotations for hg19 only. Everything here was lifted to hg38, so positions should be treated with the usual care given to lifted data. See Methods for what was lost or changed in the lift.
In the Parental ASM and ICRs Revised tracks, colour shows the gamete in which the methylation mark is laid down:
| Oocyte germline DMR — the mark is placed in the egg, so the maternal copy is the methylated one | |
| Sperm germline DMR — the mark is placed in sperm, so the paternal copy is the methylated one | |
| Secondary DMR — the difference between the two copies appears after fertilisation rather than in the gametes | |
| Black — the gamete of origin has not been established, which is the case for every region reported for the first time in this study |
The Bimodal Meth track is shaded by how many cell types share a region, from light for a region seen in a single cell type to dark for one seen in more than twenty.
Each track can be filtered from its configuration page. The Parental ASM track filters on whether a region is a known control region, on the gamete of origin, and on how many cell types support it. The Bimodal Meth track filters on the individual cell types a region was called in. Clicking a region opens a details page with the supporting SNPs, the cell types, the best adjusted p-value and, for the parent-of-origin regions, the mean methylation of the region in published oocyte, sperm, inner cell mass and blastocyst methylomes. The oocyte and sperm values are the direct evidence for which parent contributes the methylated copy, and they are missing for some regions.
Two fields record problems with the underlying data rather than biology. The Note field on the Bimodal Meth track marks the 125 regions whose cell type list was cut off in the file the authors published. The Lifting note field, present on all four tracks, marks regions that hg38 stretched because sequence was inserted inside them; their boundaries no longer match the published hg19 ones.
Whole-genome bisulfite sequencing was done on 202 healthy samples covering around 40 primary cell types, sorted or dissected from 135 donors, at 150 bp paired-end reads and an average of 984 million read pairs per sample, mapped to hg19 with wgbstools. Rather than averaging methylation over a region, each sequenced fragment covering at least three CpG sites was scored on its own and called hyper-methylated when at least 65% of its CpGs were methylated, hypo-methylated when at most 35% were, and mixed in between. A region of at least five consecutive CpGs was called bimodal when at least 20% of its fragments were hyper-methylated and at least 20% hypo-methylated, tested with a log-likelihood ratio between a one-epiallele and a two-epiallele model fitted by expectation-maximisation, with Benjamini-Hochberg correction. Bimodal regions were then intersected with 1,360,985 gnomAD SNPs at minor allele frequency of 1% or more. In donors heterozygous at such a SNP and with at least five fragments per allele, a Fisher exact test asked whether the genotype of the fragment predicted its methylation state, again with FDR correction, at a threshold of 0.1. A region where the same allele carried the methylation in every donor was taken as sequence-dependent. A region that was bimodal regardless of genotype, and where the allele carrying the methylation switched between unrelated donors, was taken as parent-of-origin, requiring at least three donors with allele-specific methylation. Parent-of-origin calls were checked against 33 mother-father-child trios by bisulfite treatment and multiplexed PCR of tongue swab DNA followed by NextSeq sequencing, and against published gamete and early-embryo methylomes. See the reference below for the full protocol.
The four region sets were taken from the hg19 custom tracks the authors published, at parental_ASM.bigbed, ICRs-new.bigbed, bimodal+ASM.bigbed and bimodal.bigbed, reachable through the authors' saved session. Those files hold coordinates and little else, so the SNPs, cell types, p-values, gene links, control region names and gamete methylation shown on the details pages were read out of the supplementary data of the paper and joined onto the regions by coordinate. Region sets were not otherwise changed. Coordinates were lifted from hg19 to hg38 with liftOver. The Parental ASM track lost 2 of 460 regions in the lift, Bimodal Meth lost 329 of 385,564, and no control region was lost; in every case liftOver reported the region as deleted, split or only partly present in hg38. A further 33 regions across the three tracks changed length by more than 10% because hg38 added sequence inside them. These were kept rather than dropped, since one of them is TCEB3C, the only control region on chr18, but each carries a note saying its boundaries are no longer reliable. One count differs from the paper by design: the paper reports 324,759 bimodal regions while the file the authors published for browser display holds 385,564. The build steps are in kaplanImprint.txt and the scripts in the imprinting script directory.
The data can be explored interactively in table format with the Table Browser or the Data Integrator and exported from there to spreadsheet or tab-sep tables. From scripts, the data can be accessed through our API, track=kaplanParentalAsm.
For automated download and analysis, the genome annotation is stored in bigBed files that can be downloaded from our download server. The files for this track are called kaplanParentalAsm.bb, kaplanIcr.bb and kaplanBimodal.bb. Individual regions or the whole genome annotation can be obtained using our tool bigBedToBed, which can be compiled from the source code or downloaded as a precompiled binary for your system. Instructions for downloading source code and binaries can be found here. The tool can also be used to obtain features within a given range, e.g. bigBedToBed http://hgdownload.soe.ucsc.edu/gbdb/hg38/imprinting/kaplanImprint/kaplanParentalAsm.bb -chrom=chr21 -start=0 -end=100000000 stdout
The original annotation source data can be downloaded from the authors' server in hg19 coordinates, and the underlying sequencing data from GEO under accession GSE186458. The analysis code is at github.com/yonniejon/imprint_atlas.
Thanks to Jonathan Rosenski, Tommy Kaplan and colleagues at the Hebrew University of Jerusalem for producing this atlas and for making the region sets available as browser tracks.
Rosenski J, Peretz A, Magenheim J, Loyfer N, Shemer R, Glaser B, Dor Y, Kaplan T. Atlas of imprinted and allele-specific DNA methylation in the human body. Nat Commun. 2025 Mar 11;16(1):2141. PMID: 40069157; PMC: PMC11897249