Estimating drop-out probabilities in forensic DNA samples: a simulation approach to evaluate different models.

Allele drop-out is a well known phenomenon that is primarily caused by the stochastic effects associated with low quantity or low quality DNA samples. Recently, new interpretation models that employ the use of logistic regression have been utilised in order to estimate the probability of drop-out. The model parameters are estimated using profiles from samples of extracted DNA diluted to low template levels in order to induce drop-out. However, we propose that this approach is over-simplistic, because several sources of variability are not taken into account in this generalised model. For example, in real-life, small (discrete) crime-stains are analysed where cells are (or were) intact. The integrity of the paired chromosomes of the diploid cell is preserved. In extracted DNA that is diluted to low template levels, we argue that the paired-chromosome integrity is lost. This directly affects the outcome of the logistic model. To date, current experimentation procedures are more akin to haploid cells and thus, different logistic models are needed for haploid and diploid cells. In order to simplify the methodology to estimate the multiple logistic regressions, we propose the use of a simulation model of the entire process associated with the analysis of STR loci, as a supplement to the purely experimental approach to support the validation of new methods. We illustrate with an evaluation of some features of the logistic model proposed by Gill et al., 2009 [12] and discuss alternative models.

One year variability of peak heights, heterozygous balance and inter-locus balance for the DNA positive control of AmpFℓSTR© Identifiler© STR kit.

Multivariate analyses of 205 positive control experiments in an AmpFℓSTR© Identifiler© STR kit were used to analyze the factors affecting peak heights at 16 loci. Peak heights were found to be highly correlated between loci and there was evidence for a difference in sensitivity of the two genetic analyzers in the blue channel. Heterozygous balance response at 10 loci was found to behave as a random variable following a beta-distribution with typical median values of 90%, without locus or genetic analyzer effect. Inter-locus balance at 16 loci was influenced by the blue channel effect and a temporal switch of unexplained origin. The implications of these results for the choice of minimum threshold values in quality control are discussed.

The predictive value of the maximum likelihood estimator of the number of contributors to a DNA mixture.

We propose to quantify the accuracy of a likelihood-based estimator that was recently proposed for the determination of the number of contributors to a DNA mixture, when genetic data alone is considered [H. Haned, L. Pène, J.R. Lobry, A.B. Dufour, D. Pontier, Estimating the number of contributors to forensic DNA mixtures: does maximum likelihood perform better than maximum allele count? J. Forensic Sci., in press]. Using Bayes' theorem, we derive a formula for the calculation of the predictive value (PV) of the likelihood-based estimator. The PV gives the probability that a DNA stain contains the DNAs of i people given that the maximum likelihood estimator gave an estimate of i contributors for this stain. We illustrate the PV calculations for two different types of DNA evidence: traces and body fluids. The PV varied according to the number of contributors involved in the DNA stain. Setting the maximum number of possible contributors to five, the lowest predictive values were scored for five-person mixtures with a minimum value of 0.26 for traces, but values were always above 0.94 for stains comprising one, two or three contributors, for both traces and body fluids. Values remained relatively high for four-person mixtures with a minimum value of 0.69. These findings confirm that likelihood-maximization is a powerful approach for the determination of the number of contributors to forensic DNA mixtures.