The team set out to answer a deceptively simple question: which of the EPIC array’s 866,836 CpG probes actually behave as designed? The answer matters because epigenome-wide association studies, or EWAS, depend on the assumption that each probe reliably measures methylation at one specific genomic location. When a probe fails that assumption, the resulting signal can reflect genetic variation, off-target binding, or outright misannotation rather than genuine methylation differences. Such artifacts have plagued earlier Infinium platforms, and previous studies of the 450K and EPIC arrays had already flagged between 6 and 11 percent of all probes as problematic. The Northwestern group extended that work by systematically characterizing the entire EPIC array and, crucially, by asking how the answers change depending on which human population a study examines.

The first stage of the analysis involved mapping every probe sequence back to the human genome reference. This is not as straightforward as it sounds, because the Infinium assay relies on bisulfite conversion, a chemical treatment that converts unmethylated cytosines to uracil, which is read as thymine, while leaving methylated cytosines untouched. To represent all possible post-conversion states, the researchers generated four single-stranded versions of the genome in silico: forward and reverse strands, each in a methylated and unmethylated configuration, with all non-CpG cytosines converted to thymines. A further complication arises with Type II probes, some of which contain ambiguous R nucleotides that can represent either adenine or guanine depending on the methylation state of an internal CpG site. The team replaced every R with all possible A and G combinations, ultimately producing 3,505,864 distinct probe sequences, which they then aligned to the bisulfite-converted reference using the BLAT alignment tool.

The results of this exhaustive mapping exercise were striking. Of the 866,836 probes on the array, 91,034 failed the alignment criteria, which allowed no more than two mismatches and no more than four ambiguous bases, and required a match at the fiftieth nucleotide of each 50-base probe. In other words, roughly one in ten probes did not align reliably to the human genome reference at all. Beyond that, 21,256 probes mapped ambiguously to multiple loci, meaning their signals represent a mixture of genuine methylation at the intended site and spurious signal from elsewhere in the genome. The researchers also compared their mapping results with the original annotations supplied by Illumina and found that 448 probes were assigned to a different gene than the manufacturer’s annotation file claimed, a small but potentially misleading discrepancy for any study interpreting methylation changes in terms of nearby gene regulation.

To understand where the ambiguous probes sit in the functional landscape of the genome, the team performed an enrichment analysis using regulatory element annotations from the ENCODE Analysis Hub at the European Bioinformatics Institute, including DNase I hypersensitivity sites, transcription factor binding sites, and histone modification peaks pooled across cell lines. They compared the observed number of overlaps between ambiguous probes and each regulatory element against a null distribution generated by randomly sampling matched sets of array probes 10,000 times. The ambiguous probes generally followed the genomic distribution of all EPIC probes, but they showed a significant trend of enrichment for H3K9me3, a histone mark associated with constitutive heterochromatin and transcriptionally silent regions, with an empirical p-value below 0.001. This suggests that problematic probes are disproportionately located in compacted, repetitive portions of the genome, precisely where cross-hybridization is most likely to occur.

The second, and arguably more novel, stage of the study examined how common genetic variation within probe sequences varies across human populations. The researchers drew on variant call data from the 1000 Genomes Project, covering 26 global populations organized into five major continental groups: African, European, East Asian, South Asian, and Admixed American. For each bi-allelic SNP in each population, they calculated allele frequencies and then searched the 754,546 unambiguous EPIC probes for common SNPs, defined as those with a minor allele frequency above 0.05, within 20 bases of the interrogated CpG site. This window matters because SNPs near or within the probe sequence can alter hybridization efficiency, causing the measured fluorescence to reflect the genetic variant rather than the methylation state, a phenomenon known as methylation quantitative trait locus confounding when it propagates into association results.

The population-specific findings were pronounced. East Asian populations carried the fewest common SNP-containing probes, 54,810, followed by European populations with 68,155, South Asian with 70,068, Admixed American with 70,576, and African populations with 96,674. That African populations show the highest count is consistent with the well-established observation that genetic diversity is greatest in populations of African ancestry, a legacy of human demographic history. The practical implication is uncomfortable for the field: a probe that performs cleanly in an East Asian cohort may carry a common SNP in an African cohort, and vice versa. Any EWAS conducted in a diverse or non-European population using a single, population-agnostic probe exclusion list risks either discarding usable data or, worse, retaining probes that generate population-specific artifacts that masquerade as epigenetic differences between groups.

To address this, the authors generated a curated list of optimal CpG probes for each major global population, released as supplemental tables alongside the paper. The resource allows investigators to filter their EPIC array results according to the ancestry of their study cohort, excluding probes that are cross-hybridizing, misannotated, or polymorphic in the relevant population. The authors envision a straightforward workflow: researchers who identify differential methylation at any CpG using the EPIC array can cross-reference their hits against this resource to confirm that their findings are not driven by cross-hybridization or by common SNPs specific to the population under study. Given that the supplemental tables enumerate both the problematic probes and the population-specific SNP content of the array, the resource functions as a practical quality-control layer that can be adopted without new experiments.

The broader significance of the study extends beyond a single chip. Methylation arrays remain far more cost-effective than whole-genome bisulfite sequencing for large cohort studies, and they will continue to underpin EWAS in epidemiology, cancer research, aging studies, and environmental epigenetics for years to come. But the field has increasingly recognized that genetic variation and epigenetic variation are entangled, and that interpreting methylation differences between populations without accounting for underlying allele frequency differences can produce spurious conclusions about epigenetic contributions to disease. The Northwestern analysis provides a concrete, quantified demonstration of the scale of that problem on the most widely used current platform, showing that the interplay between probe design and population genetics is not a marginal concern but affects tens of thousands of probes.

For the research community, the message is clear: the EPIC array remains a powerful and efficient tool, but its outputs are only as reliable as the quality control applied to them. Studies of diverse human populations stand to benefit most from the new characterization, since they face the greatest risk of population-specific artifacts. As epigenetic research expands beyond historically overrepresented European cohorts, resources like this probe-level audit will be essential for ensuring that discoveries about the epigenetic roots of complex traits and diseases are genuine, reproducible, and applicable to all of humanity rather than a genetically narrow slice of it.

Subject of Research: Technical characterization of the Illumina EPIC DNA methylation array for population-diverse epigenetic research

Article Title: Characterization of the Illumina EPIC array for optimal applications in epigenetic research targeting diverse human populations

Article References: Zhang, Z., Zeng, C., & Zhang, W. (2022). Characterization of the Illumina EPIC array for optimal applications in epigenetic research targeting diverse human populations. Epigenetics Communications, 2(1), Article 7. https://doi.org/10.1186/s43682-022-00015-9

Image Credits: AI Generated

DOI: 10.1186/s43682-022-00015-9

Keywords: DNA methylation, EPIC array, Illumina, epigenomics, cross-hybridization, single nucleotide polymorphism, 1000 Genomes Project, EWAS, population genetics, probe quality control, bisulfite conversion, genomic annotation

A team led by Anqi Chen, Qi Yang, Suhua Zhang, and Chengtao Li at the Institute of Forensic Science of Fudan University in Shanghai systematically compared the two dominant sequencing strategies for reading the full mitochondrial genome. On one side stood Illumina short-read sequencing, the workhorse of modern genomics, which chops DNA into fragments of a few hundred bases and reassembles them computationally. On the other stood nanopore long-read sequencing performed on a QitanTech platform, which threads much longer DNA molecules through protein pores and reads them in real time. The researchers applied both technologies to samples from 177 individuals spanning 39 maternal pedigrees, a design that allowed them to check whether each platform called the same variants in mothers, children, siblings, and cousins who should, by the rules of maternal inheritance, share essentially identical mitochondrial genomes.

The technical heart of the comparison lies in how each platform prepares DNA for sequencing. The short-read workflow relied on multiplex PCR, in which many primer pairs amplify overlapping fragments of the mitochondrial genome in a single reaction. This approach is efficient but notoriously uneven: some regions amplify enthusiastically while others lag behind, producing a patchwork of coverage depth across the genome. The long-read workflow took a fundamentally different route, using a single-amplicon enrichment strategy that amplifies the entire mitochondrial genome in one long piece. The payoff was dramatic. Long-read sequencing achieved a median coverage uniformity of 100 percent, effectively flattening the coverage variability that plagues multiplex PCR approaches and ensuring that every position of the mitochondrial genome is read a comparable number of times.

Coverage uniformity is not merely an aesthetic virtue. In forensic work, uneven coverage can mask genuine variants or inflate the apparent frequency of sequencing errors, complicating the interpretation of mixed or degraded samples. The long-read approach delivered a second, equally important advantage: the ability to see through one of mitochondrial DNA analysis’s most insidious traps. Scattered throughout the human nuclear genome are fragments of ancient mitochondrial DNA that migrated into our chromosomes over evolutionary time. These nuclear mitochondrial DNA segments, known as NUMTs, are molecular fossils that can be mistakenly amplified and sequenced alongside genuine mitochondrial DNA, contaminating forensic profiles with phantom variants. Short reads, typically a few hundred bases long, often cannot distinguish whether a sequence originated from a true mitochondrion or from a nuclear NUMT that happens to match the primer binding sites. Long reads, by contrast, can span the homologous regions entirely, revealing the surrounding nuclear context and allowing analysts to exclude these impostors with confidence.

The short-read data exposed the scale of this problem in vivid detail. The researchers observed pronounced off-target alignment in their Illumina datasets, with spurious reads piling up on particular human chromosomes. Chromosome 17 and chromosome 2 stood out as hotspots, accumulating misaligned sequences because of spurious sequence homology between NUMTs embedded there and the mitochondrial targets being amplified. In a forensic laboratory, such off-target material is more than a nuisance; it consumes sequencing capacity, complicates bioinformatic filtering, and in the worst case can introduce false variants into a profile that might later be used to include or exclude a suspect or identify human remains.

Yet the story is not a simple tale of long reads winning. When the researchers compared the consensus haplotypes produced by the two platforms, the sequences agreed: both technologies called the same set of mitochondrial variants for each individual. The critical divergence emerged in a more delicate measurement, the detection of heteroplasmy. Heteroplasmy refers to the coexistence of two or more mitochondrial DNA variants within a single person, a natural consequence of the way mitochondria are inherited through a genetic bottleneck and copied independently within cells. In forensic science, heteroplasmy can be a double-edged sword. On one hand, a shared rare heteroplasmic variant can provide powerful additional evidence linking relatives or identifying remains. On the other hand, distinguishing a genuine low-level heteroplasmic variant from a random sequencing error requires exquisite measurement fidelity, because true variants may be present at frequencies of just a few percent.

Here the short-read platform revealed its enduring strength. Illumina sequencing exhibited higher accuracy for detecting low-level heteroplasmy, thanks to the sheer depth and low intrinsic error rate of its short reads, which allow genuine minor variants to be separated from background noise with statistical confidence. The long-read workflow, despite its many advantages, carried a heavier background-noise burden. When the team validated their results against the family pedigrees, they found that the current long-read pipeline produced noise signals that were misclassified as heteroplasmy at a 10 percent allele frequency threshold. In other words, sequencing artifacts occurring at low frequencies could masquerade as genuine mixed mitochondrial populations, a potentially serious problem in casework where a false heteroplasmy call could undermine the interpretation of evidence.

Rather than declaring one technology the victor, the researchers proposed something more pragmatic and arguably more useful for the forensic community: a complementary workflow that assigns each platform the task it performs best. Under this integrated strategy, long-read sequencing would be deployed for structural characterization of the mitochondrial genome and for excluding NUMT interference, exploiting its superior coverage uniformity and its ability to span ambiguous regions. Short-read sequencing would then take over for high-fidelity quantification of heteroplasmy, where its precision at low variant frequencies is unmatched. The two technologies, rather than competing, would function as checks on each other, each compensating for the other’s weaknesses.

The implications for forensic casework are substantial. Laboratories handling complex or low-input samples, such as touch DNA from a weapon, hair shafts without roots, or skeletal remains recovered years after death, often face exactly the conditions where these platform trade-offs matter most. A degraded sample may yield too little DNA for confident short-read analysis of NUMT-contaminated regions, while the same sample might benefit enormously from long reads that clarify structural ambiguities. Conversely, a case hinging on a rare heteroplasmic variant shared between a missing person and a maternal relative demands the quantification accuracy that only deep short-read sequencing can provide. The Fudan team’s pedigree-based validation offers laboratories a data-driven rationale for choosing, or combining, their sequencing strategies rather than relying on convention or convenience.

The study also reflects a broader shift in forensic genetics, as nanopore sequencing matures from an experimental curiosity into a validated platform for legal applications. Chinese-made nanopore devices from QitanTech are emerging as alternatives to the long-dominant Oxford Nanopore systems, and their performance in this head-to-head comparison demonstrates that the technology has reached a level of maturity where systematic forensic validation, not just proof-of-concept demonstrations, is possible. With funding from China’s National Key Research and Development Program and the National Natural Science Foundation of China, and computations performed on Fudan University’s CFFF platform, the work signals that the infrastructure for rigorous platform benchmarking is now in place. For a field in which a single base call can help identify a fallen soldier or exonerate the innocent, knowing precisely when to trust long reads and when to demand the precision of short ones is not a technical footnote. It is the difference between evidence that stands up in court and evidence that collapses under scrutiny.

Subject of Research: Comparison of short-read and long-read sequencing platforms for forensic mitochondrial genome analysis

Article Title: Pedigree-Based comparison of short-read and long-read sequencing for forensic mitochondrial genome analysis

Article References: Chen, A., Yang, Q., Cao, Y., Liu, Y., Yang, F., Zhang, S., & Li, C. (2026). Pedigree-Based comparison of short-read and long-read sequencing for forensic mitochondrial genome analysis. International Journal of Legal Medicine. https://doi.org/10.1007/s00414-026-04031-5

Image Credits: AI Generated

DOI: 10.1007/s00414-026-04031-5

Keywords: mitochondrial DNA, forensic genetics, long-read sequencing, short-read sequencing, nanopore sequencing, heteroplasmy, NUMTs, DNA profiling, pedigree analysis, Illumina, QitanTech, human identification

The study, published in Epigenetics Communications by researchers at Karolinska Institutet and their collaborators in Denmark and the United States, set out to test whether the Illumina Infinium MethylationEPIC v2.0 array could produce high-quality data from archival newborn blood spots using vanishingly small amounts of DNA. This is no trivial challenge. The EPIC v2.0 array, which profiles approximately 937,000 cytosine-phosphate-guanine sites across the genome, officially recommends between 250 and 500 nanograms of DNA as input material. Decades-old dried blood spots, however, typically surrender far less genetic material, often well below 100 nanograms per sample, raising serious doubts about whether the resulting methylation data would be complete, accurate, and free of technical artifacts.

To find out, the team selected seven random samples from individuals born between 1985 and 2003, meaning the blood spots had been stored for periods ranging from 18 to 40 years. The laboratory workflow began with two 3-millimeter punches from each blood spot, which were placed into 96-well plates. Proteins and metabolites were eluted by incubating the punches in phosphate-buffered saline for two hours at room temperature on a rotary shaker, after which DNA was extracted using the Highprep Blood and Tissue DNA kit from MagBio Genomics, processed on an automated Beckman Coulter BioMek I7 liquid handler and eluted in a final volume of 50 microliters. After a portion of the extract was used for genotyping on an Illumina Global Screening Array, the remainder was carried forward to bisulfite conversion and methylation profiling at the National Genomic Infrastructure in Uppsala, part of Science for Life Laboratory.

The DNA volumes recovered from these archival spots ranged from just 19.2 nanograms to 99.2 nanograms, figures dramatically below the manufacturer’s recommended input of 250 nanograms, with the lowest sample representing less than 10 percent of the prescribed amount. Yet when the raw data came back from the array, the results were extraordinary. Probe call rates, the proportion of methylation sites successfully measured in each sample, averaged 99.75 percent at the conventional significance threshold of p less than 0.01, comfortably exceeding the greater than 98 percent benchmark commonly used to define successful probe detection in methylation studies. Coverage remained robust across all seven samples, with between 930,588 and 936,110 probes reliably detected, close to the array’s full complement of sites.

Quality control went well beyond call rates. The researchers examined the balance between methylated and unmethylated signal intensities, a key indicator of whether DNA degradation or processing problems have distorted the underlying signal. All seven samples clustered tightly around expected values, with median methylated intensity spanning a log-two range of 11.34 to 12.07 and median unmethylated intensity between 10.36 and 10.82, and no outliers emerged from the group. The team also calculated pairwise sample-to-sample Pearson correlations across the full set of genome-wide beta values, which ranged from 0.972 to 1.000. Such uniformly high correlations indicate consistent methylation profiles, reproducible processing, and, crucially, the absence of batch effects, a common scourge of microarray experiments that can masquerade as biological signal.

The fundamental structure of the methylation data also held up under scrutiny. Raw beta value distributions displayed the characteristic bimodal pattern expected of high-quality methylation data, with the vast majority of CpG sites falling into clearly methylated or clearly unmethylated classes and relatively few occupying intermediate values. This bimodality was preserved after noob normalization, a preprocessing step implemented in the R Bioconductor package minfi that corrects for background signal and dye bias, and the normalized density curves showed slightly improved alignment across samples. Embedded control probes monitoring bisulfite conversion, staining, hybridization, target removal, extension, and specificity all performed within expected ranges, with negative control signals remaining low and target-removal controls showing appropriately reduced signal compared with hybridization controls, together indicating minimal background fluorescence and efficient chemistry at every step.

The significance of these findings extends far beyond a single national biobank. DNA methylation, the chemical modification of DNA that influences gene activity without altering the underlying sequence, is increasingly recognized as a record of early-life biology and environmental exposure. Because newborn blood spots capture the epigenome at the very beginning of life, the ability to profile them retrospectively gives researchers a unique window into the molecular conditions present at birth, potentially decades before the onset of disease. Previous work had already shown promise: two earlier studies using the older Infinium 450K array achieved correlations exceeding 0.99 between dried blood spot samples and matched frozen blood fractions, and one proof-of-principle study using the HumanMethylation27 BeadChip demonstrated that neonatal blood spots stored for 26 and 28 years produced methylation profiles nearly identical to adult reference samples with only 30 nanograms of DNA. Other evaluations had pushed the EPIC v2.0 array to inputs as low as 1 nanogram, but only with fresh cell line DNA, never with archival clinical material stored at refrigeration temperatures.

What distinguishes the new study is precisely this combination of archival age, real-world storage conditions, and ultra-low input on the current-generation platform. The samples in the analysis had endured 18 to 40 years of refrigerated storage with variable collection dates, and none showed evidence of degradation-related signal imbalance. The authors note that while Kaur and colleagues had previously demonstrated adequate EPIC v2.0 performance with 1-nanogram inputs using fresh DNA, the present results extend those findings to the far more demanding context of decades-old clinical biobank samples, where DNA may be fragmented, chemically modified, or partially degraded by years of storage.

The limitations of the pilot are nonetheless clear and acknowledged by the team. Seven samples cannot establish population-level generalizability, and future work will need to validate these results in larger cohorts with broader demographic representation. Moreover, all samples analyzed had been collected after 1981 and therefore stored under controlled refrigerated conditions. Samples from before 1981, which spent their early years at room temperature, may behave differently and will require separate validation, although encouraging results from dried blood spots stored three to ten years at room temperature using the older 450K platform have been reported by Joo and colleagues and by Walker and colleagues, suggesting that even less favorable storage histories may not be disqualifying.

Even so, the proof of principle is transformative. The Swedish PKU biobank, with its five million-plus samples collected systematically since 1975 under Swedish Biobank Law provisions that permit ethically approved research use, now stands as an invaluable resource for large-scale epigenetic investigation into conditions ranging from neurodevelopmental disorders to cardiovascular and metabolic disease. Neonatal screening biobanks of similar design exist around the world, collectively representing millions of additional archived samples that could become accessible to methylation profiling. The study was approved by the Swedish Ethical Review Authority without individual informed consent, with all data handled in deidentified form in accordance with GDPR and Swedish legislation. For researchers who have long dreamed of tracing the epigenetic origins of disease back to the first days of life, the message of this small but rigorous study is unmistakable: the molecular memory written into a newborn’s blood can survive decades in a cold archive, and modern technology can now read it with remarkable fidelity, even from less than one-tenth of the DNA the platform’s designers thought necessary.

Subject of Research: DNA methylation profiling of archival neonatal dried blood spots using the Illumina MethylationEPIC v2.0 array with ultra-low DNA input

Article Title: Performance of the Illumina Infinium MethylationEPIC v2.0 array with low DNA input from Swedish neonatal dried blood spots

Article References: Brander, G., Karlsson, H., Dalman, C., Bybjerg-Grauholm, J., Crowley, J. J., & Mataix-Cols, D. (2025). Performance of the Illumina Infinium MethylationEPIC v2.0 array with low DNA input from Swedish neonatal dried blood spots. Epigenetics Communications, 5(1), Article 9. https://doi.org/10.1186/s43682-025-00042-2

Image Credits: AI Generated

DOI: 10.1186/s43682-025-00042-2

Keywords: DNA methylation, dried blood spots, MethylationEPIC v2.0, neonatal biobanks, epigenetics, low-input DNA, archival samples, PKU biobank, quality control, Illumina, Swedish biobank, CpG sites