Am. J. Hum. Genet. 51:549-553, 1992
A Note on Hardy-Weinberg Equilibrium of VNTR Data by Using the Federal Bureau of Investigation's Fixed-Bin Method B. Devlin * and Neil Risch *'t *Departments of Epidemiology and Public Health and tDepartment of Human Genetics, Yale University, New Haven
Summary To fully utilize the information of VNTR data for forensic inference, the probability of observing the matching suspect and evidentiary profile in a reference population is estimated, usually by assuming independence of alleles within and between loci. This assumption has been challenged on the basis of the observation that there is frequently an excess of single-band phenotypes (SBP) in forensic data bases, which could indicate lack of independence. Nevertheless, another explanation is that the excess SBP are artifacts of laboratory methods. In this report we examine the excess of SBP for three VNTR loci studied by the FBI (D17S79 and D2S44, for blacks, and D14S13, for Caucasians). The FBI claims that the excess is due to the effect of null alleles; the null alleles are suspected to be too small to be detected. We estimate the frequency of null alleles for two loci (D17S79 and D14S13) by comparing, for these loci, the data from the FBI data base and the data from the Lifecodes data base. These comparisons yield information on small fragments because Lifecodes uses the restriction enzyme PstM, which yields larger fragments than does HaeIII, which the FBI uses. For D17S79 in blacks, we estimate a null allele frequency of 4.4%, and, for D14S13 in Caucasians, we estimate a frequency of 3.0%. The null-allele frequency for D2S44 in blacks is derived similarly, again being based on analyses of DNA cut with HaeIII and PstI; our estimate of the null-allele frequency for this locus is 1.5%. Using these null-allele frequency estimates and a goodness-of-fit test, we show that there is no evidence for deviations from Hardy-Weinberg expectations of genotype probabilities at these loci.
Introduction One of the many uses of the data from human VNTR loci is forensic inference. Even a two-locus VNTR profile can be very informative, because the profiles will rarely match if the samples are obtained from two different individuals, and the evidence is usually quite persuasive when the profiles are from the same individual (Devlin et al. 1992). For a five-locus VNTR profile, when the FBI's methods are used, the probability of a chance match of unrelated individuals is
A Note on Hardy-Weinberg Equilibrium of VNTR Data by Using the Federal Bureau of Investigation's Fixed-Bin Method B. Devlin * and Neil Risch *'t *Departments of Epidemiology and Public Health and tDepartment of Human Genetics, Yale University, New Haven
Summary To fully utilize the information of VNTR data for forensic inference, the probability of observing the matching suspect and evidentiary profile in a reference population is estimated, usually by assuming independence of alleles within and between loci. This assumption has been challenged on the basis of the observation that there is frequently an excess of single-band phenotypes (SBP) in forensic data bases, which could indicate lack of independence. Nevertheless, another explanation is that the excess SBP are artifacts of laboratory methods. In this report we examine the excess of SBP for three VNTR loci studied by the FBI (D17S79 and D2S44, for blacks, and D14S13, for Caucasians). The FBI claims that the excess is due to the effect of null alleles; the null alleles are suspected to be too small to be detected. We estimate the frequency of null alleles for two loci (D17S79 and D14S13) by comparing, for these loci, the data from the FBI data base and the data from the Lifecodes data base. These comparisons yield information on small fragments because Lifecodes uses the restriction enzyme PstM, which yields larger fragments than does HaeIII, which the FBI uses. For D17S79 in blacks, we estimate a null allele frequency of 4.4%, and, for D14S13 in Caucasians, we estimate a frequency of 3.0%. The null-allele frequency for D2S44 in blacks is derived similarly, again being based on analyses of DNA cut with HaeIII and PstI; our estimate of the null-allele frequency for this locus is 1.5%. Using these null-allele frequency estimates and a goodness-of-fit test, we show that there is no evidence for deviations from Hardy-Weinberg expectations of genotype probabilities at these loci.
Introduction One of the many uses of the data from human VNTR loci is forensic inference. Even a two-locus VNTR profile can be very informative, because the profiles will rarely match if the samples are obtained from two different individuals, and the evidence is usually quite persuasive when the profiles are from the same individual (Devlin et al. 1992). For a five-locus VNTR profile, when the FBI's methods are used, the probability of a chance match of unrelated individuals is
