The problem the study addresses is deceptively simple to state and notoriously difficult to solve. Monozygotic twins arise from a single fertilized egg and therefore share nearly identical genomic sequences. This poses a genuine practical dilemma for investigators: when a DNA sample recovered from a crime scene matches the profile of one member of an identical twin pair, standard autosomal short tandem repeat (STR) genotyping cannot determine which brother or sister deposited it. STR profiling, the workhorse of forensic genetics for decades, reads the lengths of repeated DNA sequences at a set of genomic locations and produces a profile that is effectively unique across the population — except, crucially, for identical twins. In criminal investigations involving twin suspects, in paternity or inheritance disputes, and in disaster victim identification, this blind spot has real consequences. Courts and investigators have therefore sought alternative molecular signatures that diverge between co-twins after the fertilized egg splits.

The new study focused on long non-coding RNAs (lncRNAs), a large class of RNA molecules that do not encode proteins but participate in regulating gene expression. Once dismissed as transcriptional noise, lncRNAs are now recognized as key players in chromatin remodeling, transcriptional control and post-transcriptional regulation, and thousands of them are expressed in human blood. Because lncRNA levels are shaped by both genetic background and postnatal environmental influences — diet, stress, exposure to pathogens, lifestyle and countless other experiences — the authors reasoned that co-twins living through different lives could accumulate measurable differences in which lncRNAs are active in their blood, and that these differences could serve as forensic markers. Unlike DNA sequence, which is essentially frozen at conception, the transcriptome is a dynamic record of biology that continues to diverge as twins age.

The work began with next-generation sequencing. The researchers performed lncRNA sequencing on peripheral blood samples from four pairs of monozygotic twins, cataloguing the transcripts present in each individual. In total they identified 12,547 expressed lncRNAs, a figure that underscores just how much of the human genome’s output consists of non-coding transcripts rather than protein-coding messenger RNA. By comparing expression levels between the members of each twin pair, the team searched for transcripts that were consistently differentially expressed in every pair examined. That screen yielded 31 lncRNAs commonly differentially expressed across the four twin pairs — evidence that transcriptional divergence between genetically identical individuals is not a rare event but a systematic feature of twin biology.

Because sequencing is costly and not well suited to routine forensic laboratories, the team moved to a targeted approach. Quantitative PCR, the standard instrument in molecular biology labs worldwide, requires only short, well-characterized amplicons and modest amounts of template, making it far more practical for casework than whole-transcriptome sequencing. In a preliminary PCR experiment the researchers tested whether candidate transcripts could be reliably amplified from forensic-type samples. Eleven lncRNAs showed stable amplification and high specificity and were carried forward for validation: MSTRG.30665.2, ENST00000414030, MSTRG.73422.1, MSTRG.87582.24, MSTRG.100816.5, MSTRG.54028.15, MSTRG.16169.7, MSTRG.98182.1, MSTRG.68271.29, MSTRG.63897.85 and MSTRG.16294.1.

The researchers then used quantitative PCR (qPCR) to measure these 11 transcripts, validating them first in the original four twin pairs and subsequently in an independent cohort of five additional MZT pairs, for a total of nine pairs. This two-stage design — discovery by sequencing, validation by qPCR in an independent set — mirrors the analytical logic used in biomarker development in clinical research and helps guard against markers that appear promising by chance in a small discovery sample. The markers did not perform equally: their discriminatory power varied from transcript to transcript, as expected given that individual lncRNAs respond differently to the environmental and stochastic forces that drive divergence between co-twins. The standout was MSTRG.73422.1, which showed consistent, significant differential expression between co-twins in all nine twin pairs examined, making it the strongest single discriminator in the panel. A marker that separates every tested twin pair is a notable result, because any single-locus forensic test must perform reliably across unrelated pairs, not just in most of them.

A key concern for any RNA-based forensic method is that RNA degrades readily. Ribonucleases — enzymes that chew up RNA — are ubiquitous in the environment and on human skin, and RNA molecules are chemically less robust than DNA. Forensic samples often sit for days, months or years in variable conditions before analysis. So the team assessed whether their markers could survive the conditions typical of casework. Using samples from five unrelated individuals, they subjected the 11 lncRNAs to stability testing. Five of the markers proved especially hardy: their expression levels did not significantly decrease after storage at room temperature for up to 180 days, and they withstood 10 repeated freeze-thaw cycles. These five also displayed robust longitudinal stability, meaning their expression remained consistent over time in a given individual, an essential property if a marker measured in a bloodstain found at a scene is to be compared with a reference sample collected from a suspect weeks or months later. A marker whose levels drift within a person over time would produce false mismatches; a marker that holds steady can be compared across time points with confidence.

By integrating three criteria — discriminatory power, resistance to degradation and long-term stability — the authors proposed a combinatorial lncRNA assay as a practical molecular tool for identifying which member of a monozygotic twin pair contributed a biological sample. The combinatorial logic is important: no single marker is likely to be decisive in every case, but a panel of markers, each independently informative, can be evaluated jointly to reach a conclusion with quantifiable confidence. Such an assay could, in principle, be run in laboratories already equipped for quantitative PCR, without requiring the whole-transcriptome sequencing used in the discovery phase, which lowers the barrier to adoption considerably.

The study sits within a growing body of research aimed at cracking the identical-twin problem. Previous work has explored postzygotic mutations, which accumulate after the embryo splits and create small genetic differences between co-twins, as well as copy-number variation, mitochondrial DNA heteroplasmy, T-cell receptor repertoires and the microbiome. Each approach captures a different slice of the biology that separates twins: postzygotic mutations reflect the random errors of DNA replication in the developing embryo; T-cell receptor rearrangements reflect the idiosyncratic somatic recombination events that build each individual’s immune repertoire; the microbiome reflects the distinct communities of microbes that colonize each person from birth. Epigenetic approaches have been particularly active: DNA methylation differences between co-twins were documented as early as 2005, and more recent studies have profiled blood DNA methylomes, microRNA expression by droplet digital PCR, circular RNAs, transfer RNA-derived fragments and metabolomic signatures, all with the goal of distinguishing twins from peripheral blood. The lncRNA strategy adds a new layer to this toolbox and complements these existing markers rather than replacing them. Because lncRNAs, microRNAs, methylation and metabolites each respond to different combinations of genetic and environmental influence, a multi-modal panel could in principle be more robust than any single class of marker alone.

The authors were careful to follow established standards for quantitative PCR research, including normalization of expression data using multiple internal control genes and adherence to the MIQE guidelines for reporting qPCR experiments. The MIQE guidelines — Minimum Information for Publication of Quantitative Real-Time PCR Experiments — were developed to combat a widespread problem in the qPCR literature: poorly reported assays whose results could not be reproduced or properly evaluated. Practices such as validating primer efficiency, documenting reference gene stability and disclosing full assay details are exactly what forensic journals have increasingly urged on researchers in this field, since any method offered to courts must withstand intense scrutiny of its analytical foundations.

The study also has limitations that temper immediate application. The discovery phase rested on only four twin pairs, and although validation extended the panel to nine pairs, this remains a small sample relative to the diversity of populations forensic laboratories serve. Twin pairs recruited in one region of China may not represent the range of ages, environments and lifestyles encountered in casework elsewhere, and the frequency with which a given marker discriminates a random twin pair cannot be estimated precisely from so few pairs. The markers were evaluated in peripheral blood, so their performance in other body fluids such as saliva, semen or touch DNA remains untested here — a significant gap, since many forensic samples are not blood. Environmental degradation testing covered specific conditions, and real casework samples may present harsher or more varied challenges, including exposure to heat, humidity, UV light or microbial contamination. Whether expression differences remain stable across illness, medication, age or lifestyle changes — factors known to influence the transcriptome — was not addressed in this study, and any courtroom deployment would require full developmental validation under forensic accreditation standards, including likelihood-ratio frameworks for evaluating multivariate evidence. Likelihood ratios, the standard currency of forensic interpretation, allow a court to weigh how much more probable the evidence is under one proposition (the suspect’s twin deposited the stain) than under another (the suspect did), and building such frameworks for multi-marker RNA panels is a nontrivial statistical task that remains to be done.

Nonetheless, the implications are significant. The findings demonstrate that lncRNA expression differences between co-twins are common, detectable with routine qPCR, and, for a subset of markers, sufficiently stable to survive prolonged storage and repeated temperature cycling. That combination — biological informativeness, technical accessibility and forensic-grade robustness — is exactly what has been missing from earlier proposals that remained confined to research settings. As RNA-based forensic methods mature, panels combining lncRNA markers with methylation, miRNA or metabolomic signatures could offer courts a defensible way to attribute evidence to one twin rather than the other in cases that currently cannot be resolved by DNA profiling alone. The study was supported by the Shanghai Key Laboratory of Forensic Medicine and a China Medical University student innovation program, and the underlying data are available from the corresponding authors on reasonable request. For now, the work stands as a proof of concept that the transcriptome carries a personal signature strong enough to separate even genetically identical individuals — a signature written not in the genome itself, but in the way the genome is read.