1. DNA Identification: Inclusion Genotype and LR
Mark W Perlin, PhD, MD, PhD
Cybergenetics, Pittsburgh, PA
TrueAllele® Lectures
Fall, 2010
Cybergenetics ©
Cybergenetics ©

2. Mixture Data Some amount of contributor A genotype Mixture data with
genotypes of
contributors A & B +
PCR
Other amount of
contributor B
genotype

3. Quantitative Likelihood

4. Qualitative Likelihood

5. Punnett Square 1 2

6. Prior Probability
0.10
probability
1, , , …
allele pairs

7. Likelihood Function included excluded likelihood 1, 1 1, 2 2, 2 …
1, , , …
allele pairs

8. Genotype Inference

9. Posterior Probability
0.30
0.20
probability
0.10
1, , , …

10. Likelihood Ratio

11. Uncertain Genotypes Bayes theorem updates probability
• prior probability
• likelihood function
• posterior probability
Probability concentration determines likelihood ratio

12. Uniform population prior

13. Realistic population prior

14. Likelihood: all pairs are equal

15. Filter allele pair list

16. Unequal: quantitative data

17. Population prior probability

18. Combine with equal listing

19. Posterior proportional to prior

20. Population genotype prior

21. Likelihood: smaller allele pair list

22. Posterior focuses probability

23. Population genotype prior

24. Quantitative likelihood function

25. Evidence genotype posterior

26. Conclusions • No real inclusion vs. LR controversy: same construction
• All uncertain genotypes are probability distributions
• Efficacy measured by preserving identification information
apply log(LR) information measure to inclusion method
• Inclusion has justification for use in court, since is a LR
• PI statistic understandable via inclusion likelihood function
• Relevance of inclusion? Depends on the data.