What a peer-reviewed study shows about allele sharing, non-donor relatives, contributor number, and defense strategy when DNA is used against you
Written by Virginia “Ginny” Barron, Minnesota criminal defense attorney and forensic DNA litigator.
A DNA mixture is recovered from a firearm, piece of clothing, vehicle, or other evidence item. The laboratory compares the mixture to you and reports a likelihood ratio that appears overwhelmingly strong. The number is presented as if it settles the question of whose DNA is in the mixture.
But there is an important fact in the background: your brother, sister, parent, child, or another close relative may also be a realistic potential contributor. That matters because close relatives share more alleles than unrelated people. In a complex DNA mixture, that shared genetic information can affect the apparent number of contributors, the genotypes the software considers, and the likelihood ratio assigned to a person who did not actually contribute DNA.
A 2022 peer-reviewed study by Tim Kalafut, Jo-Anne Bright, Duncan Taylor, and John Buckleton examined exactly this problem using STRmix™. The findings do not mean that STRmix is unreliable whenever relatives are involved. They do show why family relationships cannot be treated as an afterthought when the prosecution relies on a mixed DNA profile, and must be closely scrutinized.
A large likelihood ratio is only as meaningful as the propositions, contributor assumptions, and genetics behind it.
For the defense, the practical lesson is simple: if close relatives are realistically connected to the evidence, the laboratory file must be reviewed for more than the final number. The assumptions that produced that number may be as important as the number itself.
The Study: What Kalafut and Colleagues Tested
The paper, published in Forensic Science International: Genetics in 2022, investigated mixtures containing first-order relatives – including parents, children, and full siblings. The researchers created both physical DNA mixtures in the laboratory and simulated mixtures, interpreted them with STRmix™, and compared likelihood ratios for true donors, unrelated non-donors, and related non-donors.
The study focused on several related problems: whether allele sharing can make the number of contributors difficult to determine, whether a non-donating relative can receive a likelihood ratio that appears to support inclusion, and what analytical approaches may reduce misleading interpretations.
The paper is particularly important because three of its authors – Jo-Anne Bright, Duncan Taylor, and John Buckleton – disclose that they are developers of STRmix™. That does not remove the disclosed conflict of interest, but it makes the paper especially relevant to understanding how the STRmix research community itself describes the problem of related contributors.
1. Relatives Share Alleles – and That Can Hide Contributors
DNA mixture interpretation begins with a deceptively basic question: how many people contributed DNA to the sample? Analysts commonly refer to this as the number of contributors, or NoC. That determination matters because the mathematical model is being asked to explain the DNA data using a particular number of contributor positions.
Kalafut and colleagues found that mixtures of close relatives can make this task difficult because relatives share alleles. In their three-person balanced mixtures of relatives, the profiles could look like two-person mixtures by allele count. The authors concluded that many such mixtures would likely be assigned as two contributors rather than the true three when evaluated without knowledge of the ground truth.
That is not a trivial classification issue. If the number of contributors is wrong, the model is trying to solve the wrong problem. In the study, analysis of three-person relative mixtures as two-person mixtures produced many false exclusions of true donors. Other analytical choices produced alternate genotype explanations that did not reflect the known mixture design.
2. A Non-Donating Relative Can Receive Support as a Contributor
The study uses the term “adventitious support” for a particularly important phenomenon: a likelihood ratio appears to support the proposition that the person of interest contributed DNA even though that person is actually a non-donor.
The risk is greater for relatives because a sibling, parent, or child can share a substantial portion of the alleles found in a mixture created by other family members. The researchers found that related non-donors with high allele sharing could receive inclusionary support more often than unrelated non-donors in the scenarios they tested.
This is precisely why a defense attorney should not accept the State’s propositions without examining whether alternative propositions were actually tested. A likelihood ratio compares propositions. If the prosecution proposition includes the defendant while the alternative assumes only unrelated unknown people, that comparison may not address the real defense theory when a close relative is a plausible contributor.
3. The Choice of Propositions Can Change the Meaning of the Likelihood Ratio
Probabilistic genotyping does not produce a likelihood ratio in a vacuum. The laboratory must define competing propositions. A common source-level comparison asks whether the person of interest contributed DNA versus whether the DNA came from unknown, unrelated people. But if the real alternative is that a sibling or other relative contributed, the unrelated-person proposition may answer a different question from the one the case actually presents.
The Kalafut study involved simulated mixtures of three full siblings. In that defined experiment, when related non-donor siblings were evaluated using propositions that did not account for relatedness, the researchers observed inclusionary support for a non-donating sibling in 40 percent of comparisons, with an average likelihood ratio of approximately 1.66 x 10^17 among those comparisons. (This would be 166 quadrillion or 166,000,000,000,000,000).
When the propositions are changed to account for sibling relationships, two separate things can move, and it matters to keep them apart: how often a non-donating sibling receives false inclusionary support, and how large that support is on average.
The standard sibling comparison, often called the brother’s likelihood ratio and the version more readily available in conventional casework, barely changed how often a non-donating sibling was falsely supported. That rate fell only slightly, from 40 percent to 38.5 percent. What changed was the size of the support, not its frequency. The average adventitious likelihood ratio dropped from approximately 1.66 x 10^17 to about 2.5 x 10^4. (This would be a fall to about 25,000). The standard sibling comparison therefore shrinks the misleading number by roughly thirteen orders of magnitude while leaving the frequency of the misleading result largely intact.
Reducing the frequency itself required fuller sibling propositions. Full sibling propositions without knowledge of the parents’ genotypes lowered the rate of adventitious support to 7.4 percent, with an average likelihood ratio near 1.7. Adding full knowledge of the parents’ genotypes lowered the rate further, to 5.1 percent, with an average near 2. The authors themselves describe full knowledge of the parental genotypes as something that “can only be produced with unreasonable knowledge,” and they state that having information this complete is “very unlikely.”
That deepest reduction depends on complete parental-genotype information that the authors acknowledge is very unlikely to be available in an actual case. Importantly, the 7.4 percent result also came from a fuller sibling proposition rather than the standard sibling comparison. The practical question for the defense is therefore not simply whether relatedness can be considered, but what related-person propositions can actually be evaluated with the available reference information, software, and case circumstances.
Important limitation: those percentages are results from a specific simulated three-sibling experiment. They are not a general STRmix false-inclusion rate.
The point is not that every likelihood ratio involving relatives will collapse when a different proposition is used. The point is that the relationship structure in the case can materially affect the answer. If the defense theory involves a related alternative contributor, that relationship deserves explicit scientific evaluation rather than an assumption that every unknown person is unrelated.
4. Correct Contributor Number and Informed Mixture Proportions Matter
The study also examined how STRmix™ handled three-person mixtures of relatives under different analytical approaches. Balanced three-person relative mixtures were difficult to recognize as three-person mixtures by blind review, and interpretations that omitted useful mixture-proportion information could favor an alternate genotype solution.
When the researchers used the correct three-contributor model together with informed mixture-proportion priors, the known mixture proportions were recovered and the study reported no false exclusions in that analysis. Adventitious support still occurred for some related non-donors with high allele sharing, which reinforces the point that contributor number and relatedness are separate issues that both require attention.
For defense review, this means the final STRmix report is not enough. Counsel may need the underlying interpretation records showing the assumed number of contributors, conditioning decisions, mixture-proportion settings, propositions, run diagnostics, and the data supporting those choices.
What the Study Does – and Does Not – Prove
Kalafut and colleagues do not conclude that STRmix™ is defective. Their two-person relative mixtures were generally interpreted straightforwardly when the correct two-contributor model was used. The more difficult scenarios involved higher-order mixtures of close relatives, substantial allele sharing, contributor-number uncertainty, and propositions that did not adequately reflect related alternatives.
The authors also make an important point that cuts both ways: adventitious support for a related non-donor is an expected consequence of genetics, not something created by STRmix™. Close relatives share DNA. Any forensic interpretation system must account for that biological reality.
That is exactly why this research matters to the defense. A challenge grounded in this paper is narrow and specific to the facts of the case: did the interpretation account for the family relationships, contributor structure, and competing propositions that actually fit the evidence? A validated tool can still be aimed at the wrong question, and that is the question worth litigating.
A Composite Case Illustration
Consider a defendant – call him Daniel – who regularly spends time at a house with two brothers. Police recover a handgun from the house and obtain a three-person DNA mixture from it. Daniel denies ever handling the firearm. One or both brothers are realistic potential contributors because they live at or frequent the residence.
Suppose the laboratory reports a very large likelihood ratio comparing a proposition that Daniel contributed DNA to an alternative involving only unknown, unrelated individuals. That number may be mathematically correct for the propositions the laboratory chose. But it still may not answer the defense question: what if the true contributors include Daniel’s brothers or other relatives? The science can be right and still legally misleading.
The Kalafut study provides peer-reviewed support for investigating that possibility. It shows that mixtures of relatives can complicate contributor-number assessment, that non-donating relatives may receive adventitious support because of allele sharing, and that likelihood ratios can change substantially when relatedness is incorporated into the competing propositions.
None of that proves Daniel is a non-donor. It does, however, provide a scientifically supported basis to examine whether the likelihood ratio presented by the State answers the correct forensic question and whether a realistic related-person alternative was adequately evaluated.
How This Research Can Help the Defense
When DNA mixture evidence is being used against a defendant and relatives may realistically be involved, the study identifies several concrete areas for investigation.
1. Build the Family and Access History Before Accepting the DNA Theory
The defense should identify whether siblings, parents, children, or other close relatives had access to the location or item. Who lived there? Who visited? Who used the vehicle? Who owned or handled the object? Were relatives present during the relevant time period?
2. Examine the Number-of-Contributors Decision
Do not treat the reported number of contributors as an uncontested fact. Ask how NoC was determined, what features supported that assignment, whether another contributor number was considered, and whether close-relative allele sharing could have masked an additional donor. The Kalafut study specifically demonstrates why three-person mixtures of close relatives may be difficult to recognize as three-person mixtures.
3. Read the Actual Propositions Used for the Likelihood Ratio
A likelihood ratio is a comparison between specified propositions. Defense counsel should identify exactly who is assumed to be a contributor under each proposition and whether unknown contributors are modeled as unrelated. If a sibling, parent, or child is a realistic alternative source, the proposition may be unhelpful to answering the question in your case.
4. Obtain the Underlying STRmix Records – Not Just the Summary Report
The defense should seek the laboratory file necessary to understand the interpretation: the electropherogram data, analyst notes, contributor-number assessment, STRmix outputs, propositions, conditioning decisions, mixture-proportion information, run diagnostics, and relevant validation materials. A single likelihood ratio on a final report cannot show whether a different scientifically reasonable setup would materially change the interpretation.
For a broader framework for reviewing this type of evidence, see 10 Questions to Ask Your Criminal Defense Attorney if DNA Evidence Is Being Used in Your Case.
5. Consider Reference Samples From Relevant Relatives
Kalafut and colleagues recommend obtaining reference samples from relatives suspected of involvement when the circumstances make that appropriate. Such samples may allow a person to be eliminated, used as a conditioning profile, or evaluated under propositions that actually reflect the family structure in the case. Assess whether reference samples were obtained and compared.
6. Ask Whether Relatedness Was Included in Validation and Case Interpretation
A laboratory may have extensive validation for probabilistic genotyping generally while still facing a case with unusual levels of allele sharing among relatives. Defense review should ask whether the laboratory’s validation work and interpretation procedures address related contributors, high allele sharing, number-of-contributor uncertainty, and the type of mixture actually presented in the case.
7. Keep the Jury From Confusing a Large Number With a Complete Answer
A likelihood ratio can be extremely large and still answer only the propositions that were entered into the calculation. The defense must separate the numerical output from the broader inference the prosecution wants the jury to draw. If the alternative proposition excludes a realistic relative, the jury should understand that the reported number may not quantify the comparison that matters most to the defense theory.
The Minnesota Legal Question Is Case-Specific Reliability
In Minnesota, Rule of Evidence 702 requires expert opinion testimony to have foundational reliability. In a DNA-mixture case involving relatives, the meaningful issue may not be whether probabilistic genotyping is generally accepted. The sharper question is whether the case-specific interpretation – including contributor number, relatedness assumptions, conditioning, mixture-proportion choices, and competing propositions – has a reliable foundation for the opinion being offered.
Rule 403 can present a separate concern when an enormous likelihood ratio is offered in a way that risks misleading the jury about what the statistic actually proves. A source-level likelihood ratio does not automatically establish who handled an object, when DNA was deposited, or whether the propositions used by the laboratory fairly represent a realistic related-person alternative.
The defense does not win simply by pointing to a peer-reviewed paper. The study is useful because it identifies scientifically recognized questions that can be tested against the actual laboratory file, facts, family relationships, and interpretation choices in a particular case.
The Barron Forensic Integrity Protocol™: Applying It to Related-Person Mixtures
Cases involving relatives activate several stages of the Barron Forensic Integrity Protocol™ at once. The most important are often Foundational Reliability, Lab Data Auditing, Interpretation Challenge, SOP and Disclosure Review, and Validation Study Review.
The protocol asks whether the laboratory did more than run validated software. Did the analyst identify the correct forensic question? Was the number of contributors supportable? Were realistic family relationships known and evaluated? Did the propositions actually reflect the defense alternative? Were conditioning and mixture-proportion decisions scientifically justified? And does the laboratory’s validation support the way the method was applied in this particular mixture?
Those questions move the case away from “the computer produced a number” and toward the issue the jury actually needs to understand: what does that number reliably establish in this case?
For Defendants: What You Need to Tell Your Lawyer
If DNA mixture evidence is being used against you, tell your lawyer about family relationships that could matter even if they seem obvious or unimportant. Identify relatives who lived at, visited, worked at, rode in, owned, borrowed, or handled items connected to the scene. Explain shared clothing, vehicles, residences, tools, firearms, personal property, or other circumstances that could place relatives in the relevant environment.
Do not assume a huge likelihood ratio means there is nothing to investigate. It may be powerful evidence, but its meaning depends on the propositions and assumptions that generated it. A qualified defense review should determine whether the result answers the actual question presented by your case.
For Defense Attorneys: Questions I Ask Immediately
- Are any plausible contributors biologically related to the client or to one another?
- What number of contributors did the analyst assign, and what data supported that decision?
- Were alternate contributor numbers considered or run?
- What exact prosecution and alternative propositions produced the reported likelihood ratio?
- Were unknown contributors assumed to be unrelated, and is that assumption realistic in this case?
- Were any relatives used as conditioning profiles or evaluated as alternative contributors?
- Does the laboratory validation address related contributors and substantial allele sharing?
- Would obtaining reference samples from relevant relatives materially test the competing explanations?
- Is the analyst communicating a source-level statistic as though it resolves an activity-level question?
I have more than a decade of experience in forensic DNA litigation and have trained Minnesota defense attorneys on examining and challenging DNA evidence throughout the State. I also undertake STRmix training so that I can evaluate the same probabilistic-genotyping framework used in forensic laboratories and identify the assumptions, limitations, and case-specific questions that matter to the defense.
I offer attorney-to-attorney DNA consulting services for Minnesota defense counsel, including laboratory-file review, STRmix and mixture-interpretation analysis, expert-witness coordination, discovery strategy, cross-examination preparation, and foundational-reliability litigation.
Scientific Source & Further Reading
Kalafut, T., Bright, J.-A., Taylor, D., & Buckleton, J. (2022). Investigation into the effect of mixtures comprising related people on non-donor likelihood ratios, and potential practises to mitigate providing misleading opinions. Forensic Science International: Genetics, 59, 102691. DOI: 10.1016/j.fsigen.2022.102691
Frequently Asked Questions
Can my brother or sister’s DNA make it look like I contributed to a DNA mixture?
It can complicate the interpretation. Close relatives share more alleles than unrelated people, and the Kalafut study found that a non-donating relative could receive likelihood-ratio support in some mixtures involving related contributors. Whether that matters in a particular case depends on the mixture, contributor number, amount of allele sharing, propositions, and other interpretation choices.
Can STRmix give a likelihood ratio that supports someone who did not contribute DNA?
Yes. The study describes “adventitious support,” where a non-donor receives a likelihood ratio that appears to support contribution. The risk examined in this paper was especially important for related non-donors with substantial allele sharing. That does not mean every positive likelihood ratio is wrong; it means the underlying assumptions and alternatives must be evaluated.
Why do relatives make DNA mixtures harder to interpret?
Relatives share alleles. In a mixture, that sharing can make multiple contributors look genetically less distinct from one another, complicate the estimate of how many people contributed DNA, and increase the chance that a non-donating relative fits some of the genetic information in the mixture.
What does “number of contributors” mean in a DNA case?
The number of contributors, often abbreviated NoC, is the number of people the analyst concludes contributed DNA to the mixture. It is a critical interpretation decision because probabilistic-genotyping software evaluates the DNA under a model containing that number of contributor positions. The Kalafut study showed why close-relative mixtures can make that decision difficult.
Does a huge STRmix likelihood ratio prove I contributed the DNA?
No statistic should be interpreted without its propositions. A large likelihood ratio can strongly favor one specified proposition over another, but it does not automatically show that the alternative proposition reflects every realistic explanation in the case. If a close relative is a plausible contributor but the denominator assumes unrelated unknown people, the defense should examine whether the reported comparison addresses the actual disputed issue.
Can the defense ask the laboratory to consider a relative as an alternative contributor?
Yes, relatedness can be scientifically relevant to the competing propositions. The Kalafut study specifically evaluates likelihood ratios that account for sibling relationships and discusses reference samples and conditioning as strategies when relatives are plausibly involved. What additional testing or interpretation is appropriate depends on the facts and procedural posture of the case.
What should a defense lawyer request in a STRmix case involving relatives?
At minimum, counsel should seek the standard operating procedures relied upon in the specific case, the underlying laboratory file which includes the STRmix materials necessary to evaluate contributor number, propositions, conditioning, mixture proportions, run results, analyst decisions, and validation. The final report alone usually does not contain enough information to determine whether relatedness was recognized and appropriately addressed.
Can DNA mixture evidence involving relatives be challenged in Minnesota court?
Yes. Minnesota Rule of Evidence 702 requires expert opinions to have foundational reliability. A case involving related contributors may raise questions about the foundation for the specific mixture interpretation. Rule 403 may also matter when a statistic is presented in a way that risks misleading the jury about what the number does and does not establish.
Contact Barron Law Office
When relatives may be involved, the family relationships, allele sharing, contributor assumptions, and competing propositions can be central to understanding what the result actually means. I review the underlying laboratory data and interpretation decisions so the defense can evaluate the science rather than simply accept the final number.
Call or text directly: 507-822-5735
Serving defendants and defense counsel throughout Minnesota. Southwest Minnesota criminal defense, including Worthington, Windom, Luverne, St. James, Pipestone, Jackson, and Slayton – and DNA forensic consulting statewide.
This is not legal advice. Every case is fact-specific. You need a case-specific consultation.
