Infinium DNA Methylation BeadChip Annotation

Coordinates, masks, gene & SNP annotation, and KnowYourCG features for every Illumina Infinium methylation array — human, mouse, and mammalian.
Coherent — current, versioned together (release v8.1)  ·  Legacy — frozen, not updated  ·  download  ·  n/a
Coherent · release (InfiniumAnnotation) Coherent · tables (InfiniumAnnotationData) Legacy
PlatformGenome order<plat>.ordering.tsv.gzThe probe index — every other coherent file is row-aligned to it. Genome-independent, LC_ALL=C sorted, built from the manufacturer manifest.Probe_ID · M · U · colM/U are the bead (tango) addresses; for Infinium-I allele A is U and allele B is M. col is the color channel (G/R, or 2 for Infinium-II).coord<plat>.<genome>.coord.tsv.gzPer-probe genomic position.CpG_chrm · CpG_beg · strand · mapQCpG_beg is 0-based. The interval spans 2 nt for cg probes and 1 nt for ch/rs/nv (end = beg+2 or +1). strand is the interrogated cytosine's: + = the C of the CpG, − = the C complementing the G.mask.cm<plat>.<genome>.mask.cmDesign/quality masks as a YAME bitset — one record per tag, plus M_general (the recommended masking).one bit per probe per tagM_mapping = unmapped or alignment score < 35; M_nonuniq = mapQ < 10; plus common-SNP and population/strain masks. Full glossary below the table.snp<plat>.<genome>.snp.tsv.gzGenotyping table — sesame's formatVCF input.chrm · beg · end · strand · rs · designType · U · REF · ALT · Probe_IDU says how the unmethylated allele reads: REF or ALT for rs probes, REF_InfI for Infinium-I colour-channel switching (which supersedes the old channel-switch table).typeI_ext<plat>.typeI_ext.tsv.gzInfinium-I extension base, one value per ordering row — drives the GCT bisulfite-conversion control.extC = extension C, so Grn signal there means unconverted; T = a genuine T, the background reference; . = everything else. M_general-masked probes are excluded. manifest<plat>.<genome>.manifest.tsv.gzThe full 28-column table: coordinates, addresses, sequences and per-allele mapping.CpG_chrm · CpG_beg · CpG_end · address_A/B · target · nextBase · channel · Probe_ID · mapFlag/Chrm/Pos/Q/Cigar/NM/AS/YD (A and B) · AlleleA/B_ProbeSeq · typetarget is CG for methylation probes, else the reference allele. nextBase is the extension base on the probe strand. NM = mismatches, AS = alignment score, YD (f/r/n) = bisulfite strand.gene v41<plat>.<genome>.manifest.gencode.v41.tsv.gzGene and promoter association (all isoforms considered).CpG_chrm · CpG_beg · CpG_end · probe_strand · Probe_ID · genesUniq · geneNames · transcriptTypes · transcriptIDs · distToTSSCovers 1.5 kb upstream of the TSS through the termination site; a probe within ±1.5 kb of a TSS is promoter-associated, otherwise gene body. design<plat>.design.tsv.gzProbe design annotation (frozen).Probe_ID · design / Trait_AssociationsMSA: EWAS trait associations as trait:PMID:q-value. MM285: design groups, including syntenic EPIC probe mapping.morelegacy tablesFrozen, no longer updated.Old mask.tsv with the per-probe SNP detail, SNP→VCF, 3′-subsequence copy number (probeID plus copy_10 … copy_50 = bisulfite-genome copies of that 3′ length), mouse ChromHMM, conservation, hg19 and per-species manifests.
MSAhg38browse
EPICv2hg38browse
EPIChg38browse
HM450hg38browse
HM27hg38browse
MM285mm10browse
MM285mm39
Mammal40hg38species
Coherent files are row-aligned to <plat>.ordering.tsv.gz and carry no probe IDs — attach them with sesame-cli: sesame attach-probe --platform EPIC EPIC.hg38.coord.tsv.gz (works on a coord table, a .cm mask/feature, or a beta .cg; or a positional paste with the ordering's first column). Read and query the .cm files with YAME, and test feature enrichment with KnowYourCG. The snp table is sesame's formatVCF input (rs-probe genotyping + Infinium-I channel-switch, folded into its U column) and typeI_ext drives the GCT bisulfite-conversion control. See the README. Downloads track the latest release; for a fixed version replace main with a tag.
Functional annotation. Don't download the feature sets by hand — KnowYourCG fetches them and runs the enrichment. It pins this release, so the sets already line up with the ordering: kycg fetch to pull a knowledgebase, then kycg test -m hg38:CGI,ChromHMM query.cg to test a CpG set against it — one row per (query, record) with 2×2 counts, a hypergeometric p in log10, effect sizes and an FDR. Use -m hg38 for everything cached, or kycg info to describe a .cg/.cm. What each set is — biology, source, citation — lives in the knowledgebase registry that ships with kycg.

Formats, previous versions & legacy resources

Mask tags — what each M_* flag means

The mask table is Probe_ID · mask · maskUniq · M_general: mask is the comma-delimited detail, maskUniq its short form, and M_general the recommended TRUE/FALSE merge — human: M_mapping + M_nonuniq + M_SNPcommon_5pt + M_1baseSwitchSNPcommon_5pt + M_2extBase_SNPcommon_5pt; mouse: M_mapping + M_nonuniq.

M_mapping
unmapped, or alignment score under 35 (either probe for Infinium-I), or Infinium-I alleles A and B mapping to different locations
M_nonuniq
mapped but mapping quality under 10 (either probe for Infinium-I)
M_uncorr_titration
titration correlation under 0.9 — a working probe should track titrated methylation closely
M_commonSNP5_5pt
a common SNP (MAF ≥ 5%) within 5 bp of the 3′ extension
M_commonSNP5_1pt
a common SNP (MAF ≥ 1%) within 5 bp of the 3′ extension
M_1baseSwitchSNPcommon_1pt
Infinium-I probe whose extension base carries a SNP (MAF ≥ 1%) that switches the colour channel (CCS probes)
M_2extBase_SNPcommon_1pt
Infinium-II probe whose extension base carries a SNP (MAF ≥ 1%)
M_SNP_<POP>_1pt
population-specific SNP mask (MAF ≥ 1%) — EAS, EUR, AFR, AMR, SAS; also 1baseSwitch and 2extBase variants
M_<STRAIN>
mouse strain-specific SNP within 5 bp of the 3′ extension — PWK_PhJ, AKR_J, A_J, NOD_ShiLtJ, MOLF_EiJ, 129P2_OlaHsd … also 1baseSwitch and 2extBase variants

Previous versions

Coherent: any release by git tag — swap mainv8/v7/… in a download URL.
Legacy / manifests: under each platform's archive/ (e.g. 202209, 20260718); older GENCODE (v22/v36), LEGX & Illumina-A2 mouse.

Mouse MM285: sesame default preprocessing uses the N=296,070 design-paper set (vs Illumina A2, N=287,692); manifest comparison. Working with the mouse array: sesame nonhuman vignette.

Release notes

References

MSA — Goldberg et al., Scalable Screening of Ternary-Code DNA Methylation Dynamics Associated with Human Traits, Cell Genomics 2025.

EPICv2 — Kaur & Lee et al., Comprehensive evaluation of the Infinium human MethylationEPIC v2 BeadChip, Epigenetics Communications 2024.

EPIC / HM450 / HM27 — Zhou, Laird & Shen, Comprehensive characterization, annotation and innovative use of Infinium DNA methylation BeadChip probes, Nucleic Acids Research 2017.

MM285 — Zhou et al., DNA methylation dynamics and dysregulation delineated by high-throughput profiling in the mouse, Cell Genomics 2022.

Mammal40 — Arneson et al., A mammalian methylation array for profiling methylation levels at conserved sequences, Nature Communications 2022.

Nonhuman species — Ding et al., Comparative epigenome analysis using Infinium DNA methylation BeadChips, Briefings in Bioinformatics 2023.

mLiftOver — Chen & Zhou, mLiftOver: Harmonizing Data Across Infinium Platforms, Bioinformatics 2024.

Questions: wanding.zhou@pennmedicine.upenn.edu