NCJ Number
227241
Journal
Forensic Science International: Genetics Volume: 3 Issue: 3 Dated: June 2009 Pages: 193-199
Date Published
June 2009
Length
7 pages
Annotation
This paper examines the probabilistic model developed and applied by Cowell, Lauritzen, and Mortera in a series of recent papers, with attention to checking the assumption of using gamma distributions to model the probabilistic variation in peak area values obtained from the PCR process.
Abstract
The analysis shows that the gamma assumption works well when allelic dropout is not present, but performs increasingly less well as dropout becomes an increasing issue, such as occurs in Low Copy Template amplifications. Real PCR amplifications do lead to dropout. This is due to three main causes. First, the preselection of alleles leads to some allele types not being selected from the PCR amplification state, which is particularly acute for low copy number amplifications. Second, it is due to silent alleles that do not amplify because the primer does not bind to the flanking regions of the DNA. Third, the alleles are amplified, but below a threshold for detection. Under these circumstances of dropout, extensions of the gamma model to incorporate dropout for the first two reasons are being developed and will be reported. The problem of modeling the detection threshold must still be addressed within the context of the extended gamma model. Summaries of the gamma model and the PCR simulation model are presented, and the application of the simulation model to testing the gamma model is discussed. 4 figures and 11 references