01 medRxivpreprint · 9 September 2026_
Genetic mechanisms of mitotic error in 840,853 blood genomes
- Human genetics
- Structural variation
Read-depth ploidy calls in 840,853 genomes find autosomal gains in about 70,000 people and more than 800 inherited loci for 12 mosaic aneuploidies, including SVs that change missegregation of specific chromosomes.
Summary
Chromosomal ploidy in leukocytes was quantified from whole-genome read depth in 840,853 individuals. Autosomal gains were found in about 70,000, seven times more than earlier data sets; loss of X in females and loss of Y in males were ubiquitous, each with heritability above 50%. GWAS of 12 mosaic aneuploidies found more than 800 common loci. Spindle assembly checkpoint genes affected most chromosomes; a PMF1 missense variant raised segregation fidelity for some chromosomes and lowered it for others. A hybrid centromere from an ancient 22q-13p recombination, the largest common SV in the human genome, carries four-fold higher risk of chromosome 22 gain. Copy number and sequence variation in the X-linked repeat arrays ICCE and DXZ4 gives a four-fold range in X-loss likelihood.
What to take from it
- Read-depth ploidy from WGS detects far more mosaic gains than array-based calling
- Mosaic X and Y loss are highly heritable
- Some loci act on specific chromosomes, not on segregation in general
- Large SVs and repeat arrays in cis change missegregation of the chromosome that carries them
Review
The scale and the read-depth approach give power that array studies did not have, and the chromosome-specific cis effects of SVs are a new class of result. The weak points are technical. Read-depth ploidy at low cell fractions is sensitive to GC bias, batch and library preparation, and the 22q-13p hybrid centromere and DXZ4 copy number change read depth themselves, so a cis association could partly be a calling artefact on the same chromosome. Short reads cannot resolve those arrays, so their genotypes are imputed or inferred. Before using the loci, ask for replication in long-read or single-cell data, the cis associations repeated with the variant region masked from the ploidy estimate, and batch-stratified results. Preprint, not peer reviewed.
02 Genome Research · 1 October 2026_
Optical genome mapping identifies clinically relevant somatic structural variation in epilepsy-affected brain tissue
- Structural variation
- Human genetics
- Methods
An optical genome mapping protocol for resected brain tissue gives about 450x effective coverage and finds a somatic 13.2 kb DEPDC5 deletion at about 20% VAF as the second hit in a patient with a germline variant.
Summary
A protocol for optical genome mapping of surgically resected brain tissue was optimised to about 450x effective coverage, enough for somatic variants at low allele fraction. In four patients with epilepsy, OGM found large and complex SVs, most missed by short-read exome sequencing of the same specimen. In a patient with a known germline DEPDC5 variant, OGM found a somatic 13.2 kb DEPDC5 deletion at about 20% VAF. PacBio HiFi data resolved its breakpoints to recombination between two flanking Alu elements.
What to take from it
- High-molecular-weight DNA for OGM can be recovered from surgical brain tissue
- 450x effective coverage reaches low-VAF somatic SVs
- The DEPDC5 case fits the two-hit model for lesional epilepsy
- HiFi reads were still needed for breakpoints and mechanism
Review
The technical result is real: getting OGM-grade DNA from resected tissue and reaching low allele fractions. Four patients is a feasibility study, not a yield estimate. Comparing against exome sequencing inflates the gain, because exomes are known to miss SVs; the fair comparison is short-read genomes or long reads at matched depth. The abstract gives no detection limit by VAF, no false positive rate and no orthogonal validation for the SVs other than the DEPDC5 deletion. Before adopting it, ask for a sensitivity curve from dilution series, validation of every reported SV, and comparison with deep short-read and HiFi genomes from the same tissue.
03 Journal of Advanced Research · 2 October 2026_
Integrative genomics decoding DNA methylation-mediated genetic control of complex traits in pigs
- Animal science
- Methods
185 skeletal muscle methylomes map cis-meQTLs at CpG and CpG-block level, link them to eQTL and GWAS for growth and meat quality, and validate one block at SOCS3 by targeted demethylation.
Summary
185 whole-genome bisulfite sequencing data sets from pig skeletal muscle were used to map cis-meQTLs for single CpGs and CpG blocks. These were combined with RNA-seq, Hi-C and GWAS of growth and meat quality traits by colocalisation, Mendelian randomisation and mediation analysis. CpG-block methylation captured more genetic signal than single CpGs. The analyses prioritised 6,894 candidate genes. Genotype-predicted methylation scores showed moderate prediction for carcass and meat-quality traits, including backfat thickness. CRISPR-dCas9-TET1 demethylation of one CpG block raised SOCS3 protein and changed proliferation in muscle satellite cells.
What to take from it
- Cis-genetic control of muscle methylation in pigs is widespread
- CpG blocks capture more regulatory signal than single CpGs
- Methylation-based genetic scores add a layer for prediction of carcass traits
- One locus has functional support from targeted demethylation
Review
A useful resource for pig functional annotation of muscle traits. 185 samples is modest for meQTL mapping, and 6,894 prioritised genes is too many to act on; colocalisation and MR with that sample size give many weak instruments. The methylation scores are reported as moderate without comparison to plain SNP-based genomic prediction, which is the baseline a breeding programme would use. The SOCS3 experiment is in cultured cells and does not show an effect on a trait. Before using it, ask for the gain of methylation scores over GBLUP on the same animals, replication of top meQTLs in an independent population, and the number of loci that pass colocalisation with high posterior.
04 Human Mutation · 1 October 2026_
Novel strategy for structural variant genotyping by short-read genomic sequencing from restriction-circles: experimental and bioinformatics proof-of-concept
- Structural variation
- Methods
- Human genetics
BclI digestion and self-ligation of genomic DNA before short-read WGS give linked read pairs that join restriction-fragment ends and genotype deletions, duplications and repeat-mediated inversions.
Summary
Genomic DNA is digested with BclI, fragment ends are ligated into circles, and the library is sequenced with short reads. Linking paired-end reads join the two ends of each restriction fragment and carry long-range information; each SV class gives a specific pattern. In a Duchenne muscular dystrophy patient, coverage was 8.6x vertical and 83% horizontal, and the known DMD exon 45-54 deletion and exon 38-43 duplication were recovered. Two new Xq28 SVs were found and confirmed: a 1.4 kb deletion and an inversion between 99% identical 8.4 kb inverted repeats, both polymorphic in the population. The method correctly called absence of the recurrent haemophilia F8 inversions, and standard small-variant calling still works on the data.
What to take from it
- A wet-lab step makes short reads carry restriction-fragment-scale linkage
- Detects inversions between near-identical repeats that standard short-read WGS misses
- Small-variant calling runs on the same data
- Validated on one patient and a few loci
Review
A cheap idea in the line of mate-pair and inverse PCR, and it targets the right gap: copy-neutral inversions mediated by long repeats. Evidence is one patient, a handful of X-chromosome loci and 83% horizontal coverage, so 17% of the genome is not assessed and resolution depends on BclI site spacing. There is no genome-wide comparison with long reads or with standard short-read SV callers, so sensitivity and false positive rate are unknown. Before using it, ask for a benchmark on a GIAB sample against the HG002 SV truth set, coverage by BclI fragment size, and a cost comparison with low-pass long-read sequencing.
05 Genome Biology and Evolution · 30 September 2026_
Repeated domestication shaped the evolution of grain amaranth in Mesoamerica
- Population genetics
Whole genomes of 138 accessions split Amaranthus cruentus into three groups, with ongoing gene flow from wild A. hybridus in central Mexico and a secondary population in Guatemala.
Summary
138 accessions of A. cruentus, A. hypochondriacus and the wild relative A. hybridus, mainly from southern Mexico and Guatemala, were whole-genome sequenced. A. cruentus forms three groups: central Mexico, mostly Guatemala, and a group dispersed across Mexico. The dispersed group has the lowest differentiation from A. hybridus and the highest diversity. Gene flow continues between the central Mexico group and A. hybridus. The data suggest first domestication in central Mexico and a secondary population in Guatemala.
What to take from it
- A. cruentus has structured subgroups, not one domestication lineage
- Wild-to-crop gene flow continues in the domestication centre
- Guatemala holds a distinct, secondarily established population
- Domestication history is non-linear, as in several livestock species
Review
Dense sampling in the domestication region is the strength, and the pattern of continued wild gene flow is the same one seen in pigs, goats and cattle. The inference of a central Mexico origin rests on differentiation and diversity, which gene flow from A. hybridus also changes, so the group with the lowest differentiation from the wild relative may be the most admixed rather than the oldest. The abstract reports no demographic model, divergence times or explicit test of single versus multiple origins. Before relying on the history, ask for model-based demography with admixture (for example, fitted with moments or fastsimcoal), dated splits, and sampling of Andean material to place the Mesoamerican groups in the wider genus history.