# Reporting Statistics

When 'failure to exclude' is the interpretation for reference and evidence samples, then a statistical estimate of the significance of a match is needed. Mitochondrial DNA is inherited in its entirety from one's mother without recombination (discussed later in the chapter). Therefore individual nucleotide positions are inherited in a block and must be treated as a single locus haplotype the same as with Y chromosome information discussed in Chapter 9. The product rule applied to independently segregating STR loci found on separate chromosomes cannot be used with mtDNA polymorphisms.

The current practice of conveying the rarity of an mtDNA type among unrelated individuals involves counting the number of times a particular haplotype (sequence) is seen in a database (Wilson et al. 1993, Budowle et al. 1999). This approach is commonly referred to as the 'counting method' and depends entirely on the number of samples present in the database that is searched. Thus, the larger the number of unrelated individuals in the database, the better the statistics will be for a random match frequency estimate.

Population frequencies for most DNA types (around 60%) are not known presently because they occur only a single time in a database (Isenberg 2004). Based on available population information, confidence intervals can be used to estimate the upper and lower bounds of a frequency calculation (Holland and Parsons 1999, Tully et al. 2001) (see D.N.A. Box 10.3).

Other methods are sometimes used for calculating a statistical weight for an mtDNA match. For example, likelihood ratios have been proposed

In cases where an mtDNA profile is observed a particular number of times (X) in a database containing N profiles, its frequency (p) can be calculated as follows:

An upper bound confidence interval can be placed on the profile's frequency using:

In cases where the profile has not been observed in a database, the upper bound on the confidence interval is

where a is the confidence coefficient (0.05 for a 95% confidence interval) and N is the number of individuals in the database.

For example, the mtDNA type 16129A, 263G, 309d, 315.1C occurs twice in 1148 African-American profiles, twice in 1655 Caucasian profiles, and not at all in 686 Hispanic profiles when searched against the mtDNA Population Database (Monson et al. 2002). Using the equations above, calculations for the rarity of this profile in the respective sample sets are as follows:

For African-Americans: p = 2/1148 +1.96 [(2/1148)(1 - (2/1148))/1148]"2 = 0.0017 + 0.002 = 0.004 = 0.4%

For Caucasians: p = 2/1655 +1.96 [(2/1655)(1 - (2/1655))/1655]1/2 = 0.0012 + 0.0017 = 0.0029 = 0.29%

For Hispanics: 1 - (0.05)1/686 = 1 - 0.9956 = 0.0044 = 0.44%

These calculations demonstrate that the statistical weight can be similar whether or not a match is found to a few previously observed samples in a database.

Sources:

Evett, I.W. and Weir, B.S. (1998) Interpreting DNA Evidence. Sunderland,

Massachusetts: Sinauer Associates, Inc., p. 142. Tully, G., et al. (2001) Forensic Science International, 124, 83-91.

Calculation of mtDNA profile frequency estimates

(Holland and Parsons 1999, Tully et al. 2001, Isenberg 2004). Regardless of the method used for calculating the rarity of an mtDNA profile, it is important to keep in mind that mtDNA can never have the power of discrimination that an autosomal STR marker can since its inheritance is uniparental.

LABORATORIES PERFORMING mtDNA TESTING IN THE UNITED STATES

The first efforts in mtDNA sequence analysis with a forensic applications focus were performed by the Forensic Science Service in England (Sullivan et al. 1991, Hopgood et al. 1992, Sullivan et al. 1992). Within the United States, the Armed Forces DNA Identification Laboratory and the FBI Laboratory have led the efforts in mtDNA analysis but in slightly different arenas. 