Decision-making in familial database searching: KI alone or not alone?

Area of Science:

• Forensic Science
• Statistical Genetics
• Population Genetics

Background:

• Familial database searching frequently involves comparing parent-child or full sibling hypotheses against an unrelated alternative using DNA profiles.
• A recent report proposed a decision rule combining the kinship index (KI) and identity-by-state (IBS) statistic, claiming superiority over KI alone.
• This proposal contradicts the Neyman-Pearson Lemma, which posits the likelihood ratio (KI) as the optimal criterion for binary hypothesis testing.

Purpose of the Study:

• To rigorously evaluate the efficacy of using kinship index (KI) alone versus a combined KI and identity-by-state (IBS) statistic for familial relationship determination.
• To validate or refute the claim that combining KI and IBS offers a superior decision rule in DNA profile analysis for familial searching.
• To reconcile the proposed decision rule with fundamental statistical principles, specifically the Neyman-Pearson Lemma.

Main Methods:

• A large-scale simulation study was conducted, significantly exceeding the scope of previous research.
• The study compared the performance of a decision rule based solely on the kinship index (KI) against a rule incorporating both KI and the identity-by-state (IBS) statistic.
• Hypotheses tested included parent-child and full sibling relationships against the alternative of unrelated individuals.

Main Results:

• The simulation results strongly corroborated the theoretical expectation derived from the Neyman-Pearson Lemma.
• The kinship index (KI) alone demonstrated superior performance in distinguishing between familial relationships and unrelated individuals.
• The combined use of KI and IBS did not improve, and potentially impaired, the decision-making accuracy compared to KI alone.

Conclusions:

• The kinship index (KI) alone provides a more powerful and statistically sound criterion for familial relationship determination in DNA profile analysis.
• The addition of identity-by-state (IBS) statistics to the kinship index (KI) does not enhance, and may detract from, the accuracy of familial database searching.
• These findings reinforce the foundational principles of statistical hypothesis testing in forensic genetics and DNA analysis.