# The statistical weight of mixed samples with allelic drop out First serious attempt by Gill et al. 2006, Forensic Science International 160:90 An important.

## Presentation on theme: "The statistical weight of mixed samples with allelic drop out First serious attempt by Gill et al. 2006, Forensic Science International 160:90 An important."— Presentation transcript:

The statistical weight of mixed samples with allelic drop out First serious attempt by Gill et al. 2006, Forensic Science International 160:90 An important general paper about mixtures: Curran et al. 1999, J. Forensic Science 44:987

Mixed sample with drop out

Standard Mixture Analysis Assume there are 2 people and 3 alleles: A 1, A 2, A 3 There must be a total of 4 alleles allowing for the following possible combinations: (A 1,A 1,A 2,A 3 ) and (A 1,A 2,A 2,A 3 ) and (A 1,A 2,A 3,A 3 ). Let the frequency of the 3 alleles be p 1 p 2 p 3

Details of (A 1,A 1,A 2,A 3 ) Possible pairs of sampled genotypes are: [A 1 /A 1 and A 2 /A 3 ] or [A 2 /A 3 and A 1 /A 1 ] [A 1 /A 2 and A 1 /A 3 ] or [A 1 /A 3 and A 1 /A 2 ] These pairs are chosen with frequencies 2[p 1 2 2 p 2 p 3 ] 2[2 p 1 p 2 2 p 1 p 3 ] The sum of these is 12p 1 2 p 2 p 3 Repeating this for the other two orderings and adding them all up gives 12p 1 p 2 p 3 (p 1 +p 2 +p 3 )

General formula let c=number of distinct alleles x= number of people in the mixture u i = number of copies of allele i the frequency of any particular combination

Mixtures with drop out Let Q be the dropped out allele The frequency of Q is 1-sum(distinct alleles) Suppose evidence is A 1,A 2,Q Possible orderings are (A 1,A 1,A 2,Q) and (A 1,A 2,A 2,Q) but not (A 1,A 2,Q,Q) since we have assumed only one allele dropped out frequency is 12p 1 p 2 p Q (p 1 +p 2 )

Two people mixtures Number of alleles out evidencefrequency 1A 1,A 2, A 3,Q24p 1 p 2 p 3 p Q 1A 1,A 2,Q12p 1 p 2 p Q (p 1 +p 2 ) 1A 1,Q4p13 pQ4p13 pQ 2A 1,A 2,Q12p 1 p 2 p Q 2 2A 1,Q6p12 pQ26p12 pQ2 3 4p 1 p Q 3

Likelihood Ratios Compare the probability of two hypotheses, the prosecution and the defense Each hypothesis must compute the probability of the observed genetic evidence Let L = Prob[evidence|prosecution] / Prob[evidence|defense]

Example Three person mixture Evidence: 9 Suspect: 11, 14 Two alleles dropped out Let D be the probability that one allele will drop out. In this sample the State assumes at least two alleles dropped out, and four alleles did not: This probability is: (1-D) 4 D 2

Example: state hypothesis (1-D) 4 D 2 {prob[two people with only the 9 allele]} (1-D) 4 D 2 p 9 4

Example: defense hypothesis There are several possibilities No drop out: (1-D) 6 p 9 6 One allele dropped out, five did not: (1-D) 5 D prob[three people with only the 9 allele and one allele dropped out] = (1-D) 5 D 6p 9 5 p Q Two alleles dropped out, four did not: (1- D) 4 D 2 prob[three people with only the 9 allele and two alleles dropped out] =(1-D) 4 D 2 15p 9 4 p Q 2

Example Results 13 loci with a total of 5 alleles dropped out and a minimum of three people in the mixture, 1 known, 2 unknown The lab CPI for Caucasians was 1 in 42 million DL 0.0010.0008 0.018 0.051500 0.12800 0.23000 0.52100 0.92900

Download ppt "The statistical weight of mixed samples with allelic drop out First serious attempt by Gill et al. 2006, Forensic Science International 160:90 An important."

Similar presentations