Single Cells, Big Answers: New Forensic DNA Tools Crack Complex Crime Scene Mixtures

Crime scenes rarely hand investigators a clean DNA sample. A single swab from a sexual assault kit or a touch deposit from a door handle can carry DNA from several people, sometimes in wildly unequal amounts, and sometimes from the very same kind of cell. For decades, forensic laboratories have struggled to pull apart these complex mixtures, and the hardest cases—blood mixed with blood, semen mixed with semen, or a minor contributor buried so deep in the profile that standard software cannot even detect them—have often ended in inconclusive reports. A new review published in the International Journal of Legal Medicine by Ruijia Liu, Zheng Wang, and Jing Liu of the Institute of Forensic Medicine at Sichuan University argues that the answer may lie at an unexpectedly small scale: the single cell.

The logic is disarmingly simple. Every cell in a mixture belongs to one person, so if investigators can physically isolate individual cells before DNA analysis, each isolated cell yields a single-source profile rather than a tangled blend. The idea is not entirely new—researchers first demonstrated DNA fingerprinting from single cells in 1997, when a team at the University of Leeds amplified DNA from a single cell and produced a usable genetic profile. What has changed is the sophistication of the isolation tools, the integration of genotype data across many cells, and the emergence of probabilistic frameworks that can turn a handful of single-cell profiles into courtroom-grade evidence. The Sichuan University review, published in October 2026, systematically evaluates these advances and charts a path toward routine forensic application.

The first family of techniques the authors examine is micromanipulation and laser capture microdissection, or LCM. Micromanipulation is the most direct approach: an operator uses a fine glass capillary under a microscope to pick out target cells one by one, a method first applied to forensic identification in sexual assault cases as early as 2008, when researchers isolated single sperm cells by micromanipulation for human identification. LCM takes a more automated approach, using a laser to cut target cells directly from a stained slide while leaving unwanted cells behind. The technique was originally developed at the National Institutes of Health in the late 1990s to open the microscopic frontier to molecular analysis, and forensic scientists quickly adapted it to separate pure spermatozoa from epithelial cells for short tandem repeat, or STR, analysis. Studies published in 2006 and later confirmed that laser microdissection could recover clean sperm profiles from sexual assault swabs, and more recent work has refined the hardware itself, including systems based on structured light modulation with dual cutting modes and negative pressure adsorption collection. The strength of these approaches is precision: the operator sees exactly which cell is being taken. The weakness is throughput, because isolating enough cells for a robust profile remains slow and labor-intensive.

The second category pairs fluorescence-activated cell sorting, known as FACS, with imaging flow cytometry, or IFC. Flow cytometry separates cells based on measurable properties: sperm cells, for example, are haploid while vaginal epithelial cells are diploid, and cells can also be distinguished by surface markers such as CD45, cytokeratin, and MHC class I expression. Researchers demonstrated sperm and vaginal cell separation by flow cytometry for DNA typing after sexual assault as early as 1999, and subsequent work extended the approach to fluorescence- and magnetic-activated cell sorting strategies for separating spermatozoa from multi-contributor biological mixtures. Imaging flow cytometry adds a crucial forensic capability: it captures high-resolution images of every cell as it flows past the detector, allowing classification based on morphological characteristics. A 2024 study showed that imaging flow cytometry could classify epidermal, buccal, penile, and vaginal epithelial cells using morphological features alone—a breakthrough for touch DNA cases, where the mixture often involves the same cell type from multiple people and where traditional separation methods fail entirely. FACS offers far higher throughput than micromanipulation, but it requires the sample to be in single-cell suspension, which means solid tissues and sticky stains must first be dissociated, a step that can damage fragile forensic material.

The third and arguably most promising category is microfluidic-based systems, represented by the DEPArray platform. This technology uses dielectrophoresis—the movement of cells in non-uniform electric fields—to trap individual cells in electronically controlled cages within a microfluidic cartridge. Operators can then image each trapped cell and selectively release only the cells of interest, achieving a digital precision that earlier methods could not match. Studies published in 2017 demonstrated the isolation and genetic analysis of pure cells from forensic biological mixtures with this approach, and subsequent research showed enhanced DNA mixture deconvolution of sexual offense samples. A 2024 validation study for forensic applications confirmed the technology’s reliability, and other work has applied it to the deconvolution of blood-blood mixtures using separated single-cell STR profiling—one of the most challenging mixture types in forensic practice. The platform’s precision comes at a cost, however: throughput is limited by the cartridge capacity, and the expense of the system places it beyond many routine laboratories. Newer microfluidic innovations, including integrated dielectrophoretic presorting with wireless electrode arrays for high-throughput selective single-cell isolation published in 2026, and forensic lab-on-a-disc devices for enhanced sperm cell purification, suggest the field is actively working to close the throughput gap.

Isolating a cell is only half the battle. Once a single cell is captured, its DNA must be amplified, and this is where the technique encounters its most stubborn technical obstacles. A single cell contains roughly six picograms of DNA, so whole genome amplification is required before STR typing. At this extreme low-template regime, two artifacts dominate: allelic drop-out, in which one allele at a locus fails to amplify and produces a false homozygous result, and allelic drop-in, in which stray DNA from the environment contaminates the amplification and introduces alleles that were never in the original cell. Drop-out rates in single-cell STR profiles can exceed thirty percent at some loci, and stutter—artifactual peaks one repeat unit shorter than the true allele—behaves differently in single cells than in bulk samples, complicating interpretation. Researchers have developed increasingly refined amplification chemistries to mitigate these problems, including abasic-site-mediated semi-linear preamplification for low-template DNA published in 2025, and recombinase polymerase amplification of forensic STR loci, which operates at constant temperature and could enable rapid point-of-care typing.

The real forensic power emerges when investigators analyze not one cell but many, and then integrate the genotype data computationally. Because each cell carries the same genome from one contributor, clustering algorithms can group single-cell profiles by their underlying genotype, and consensus reconstruction can correct for drop-out by combining information across replicates. Research from the group of Kristen Duffy and Catherine Grgicak demonstrated that single-cell data produces genotype distributions concentrated at the true genotype across all mixture complexities, and that evidentiary evaluation of single cells renders highly informative forensic comparisons across multifarious admixtures. Probabilistic genotyping software validated for single-cell STR analysis can compute likelihood ratios from clustered single-cell replicates, and studies have shown that probabilistic genotyping of single-cell replicates from complex DNA mixtures recovers higher contributor likelihood ratios than standard analysis. The approach has even been applied to first-degree relatives, where probabilistic genotyping of single-cell replicates prevents the false inclusion of non-donor relatives—a critical safeguard in cases involving family members. Y-chromosomal haplotyping of single sperm cells isolated from semen mixtures has successfully identified three perpetrators in a multi-suspect sexual assault case, and cluster analysis has been used to group partial autosomal haplotypes derived from single sperm STR profiling.

The review also highlights the comparative performance of the three isolation categories across throughput, sample adaptability, and cost. Micromanipulation and LCM offer the finest precision and work directly on stained slides without requiring single-cell suspension, making them adaptable to degraded or limited samples, but their throughput is low and operator skill is essential. FACS and IFC deliver high throughput and quantitative classification but demand intact single-cell suspensions and expensive instrumentation. DEPArray systems provide digital precision and image-based selection but are constrained by cartridge capacity and cost. The authors note that no single technique dominates across all criteria, and the choice depends on the mixture type, the cell types involved, and the resources available. For same-cell-type mixtures—such as blood from two donors or touch DNA from multiple handlers—only single-cell strategies can physically separate the contributors, because differential extraction and other conventional methods rely on cell-type differences that do not exist in these cases.

Looking forward, the Sichuan University team identifies three converging directions that could transform single-cell forensics from a specialized tool into a routine workflow. Automation and high-throughput systems, including deep neural networks for automated sperm detection on stained slides and fully integrated microfluidic sample-in-answer-out platforms, could dramatically reduce the manual labor that currently limits casework. Artificial intelligence-assisted analysis, from convolutional networks that hunt sperm on microscope slides to machine learning models that classify cell types from morphological features, is already demonstrating feasibility in published studies. And standardized interpretation frameworks—the development of validated pipelines for multi-single-cell clustering, consensus genotype reconstruction, and probabilistic evidence evaluation—will be essential for courts to accept single-cell evidence with the same confidence as conventional profiles. The authors emphasize that the forensic value of single-cell strategies is already clear: they can recover probative DNA profile information from unresolvable minor donors, deconvolve same-cell-type mixtures that defeat all other methods, and provide genotype data that concentrates at the true contributor genotype regardless of mixture complexity. As these techniques mature, the review concludes, the combination of automated isolation, AI-assisted analysis, and standardized probabilistic interpretation could finally give investigators the power to read the full genetic story hidden in even the most tangled crime scene sample.

Subject of Research: Forensic Science

Article Title: Single-cell strategies for the interpretation of complex forensic DNA mixtures: isolation techniques, genotype data integration, and prospects for application

Article References: Liu, R., Wang, Z., & Liu, J. (2026). Single-cell strategies for the interpretation of complex forensic DNA mixtures: isolation techniques, genotype data integration, and prospects for application. International Journal of Legal Medicine. https://doi.org/10.1007/s00414-026-04046-y

Image Credits: AI Generated

DOI: 10.1007/s00414-026-04046-y

Keywords: forensic DNA, single-cell analysis, DNA mixtures, STR profiling, laser capture microdissection, FACS, DEPArray, probabilistic genotyping, forensic science, legal medicine, Single-cell, strategies