Statistical methods for DNA fingerprinting
Date
1995
Authors
Chhatre, Varsha
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
DNA fingerprinting plays an important role in forensic sciences. Hypervariable regions of the human genome provide valuable evidence which can be used for identifying an individual based on a DNA sample. Though each individual can't be uniquely determined from a DNA fingerprint, it is believed that the probability that the DNA patterns of two randomly chosen individuals match is highly unlikely. If two samples match, it can be concluded that they may have the same genotype by chance. It is important to know how likely it is that the two individual genotypes match by chance. This depends upon how frequent the observed genotype is in the general population. If the genotype is not commonly observed then it can be concluded that the samples came from the same individual. To reach any conclusion, it is necessary to get estimates of the distribution of allele sizes for the general population or a reference population.
Frequency estimates are obtained for five Variable Number of Tandem Repeats (VNTR) loci of Orange County's Sheriff-Coroner Department's (OCSD) reference population using Fixed Bin, the mixture model of Devlin et al. (1991) and a modification of their mixture model. Estimates obtained by our modification are compared with the estimates obtained by Devlin et al.(1991) for Life Codes (LC) VNTR loci D17S79. The fit of the suggested model is tested and variances of the estimates are obtained. There are some controversies about the statistical assumptions, computations and interpretation of the results. The thesis includes a discussion on the use of the estimated allele distribution in forensic cases and on the controversy surrounding its use.