The interpretation of single source and mixed DNA profiles.
Citations
171 citations
Cites methods from "The interpretation of single source..."
...Our method differs from STRmix and TrueAllele in that we compute the marginalized likelihood expressions using exact methods without any need for MCMC sampling....
[...]
...STRmix and TrueAllele are based on a Bayesian approach through specifying prior distributions on the unknown model parameters....
[...]
...Commercial continuous software include: STRmix [14], TrueAllele[11] and DNAmixtures[6]....
[...]
...[14], EuroForMix can also accommodate allele drop-in....
[...]
152 citations
119 citations
Cites methods from "The interpretation of single source..."
...Tvedebrink et al. (2010) evaluate the weight of evidence for two person mixtures, using a multivariate normal distribution of peak heights....
[...]
...Recently, Taylor et al. (2013) used a log-normal model for the ratio between observed and expected peak heights....
[...]
117 citations
114 citations
References
609 citations
"The interpretation of single source..." refers background in this paper
...A model that is partially continuous based on allowing a probability for dropout and drop-in (hereafter the ‘‘drop model’’) [7]....
[...]
450 citations
389 citations
"The interpretation of single source..." refers methods in this paper
...A summary of nomenclature used within this paper a the allele a 1 signifies the stutter product for allele a A the mass variable for locus amplification efficiency, Al : l ¼ 1; :::; L n o – locus offset at locus l. c a constant in modelling the variance in peak height D the mass variable for degradation, dn : n ¼ 1; :::; Nf g – degradation in template vs. molecular weight for contributor n E the vector of expected peak heights Elanr ¼ Tlanr=ð1 þ plaÞ the contribution of contributor n to the expected height of the allelic peaks at locus l formed from allele a in replicate r Elða 1Þnr ¼ plaðTlanrÞ=ð1 þ plaÞ the contribution of contributor n to expected height of the stutter peaks at locus l formed from allele a where a 1 signifies the stutter product in replicate r GC the evidence of the crime stain across all R replicates Hm hypotheses, H1 and H2 hypotheses chosen to align with the prosecution and the defence, respectively J the number of contributors with j representing a specific contributor L the number of loci with l representing a specific locus LRC the continuous LR LRB the binary LR LUS the longest uninterrupted sequence within an allele M is the mass variables D, A, R and T collectively mla is the molecular weight of allele a at locus l N number of contributors with n representing a specific contributor O the vector of observed peak heights Olar the observed peak height for allele a at locus l for replicate r P the probability of observed data given mass parameters Q a catch-all allele to cover all possibilities outside a specified set R number of replicates with r representing a specific replicate R the mass variable for replicate amplification, {Rr : r = 1, ..., R} is a multiplier applied to replicate r....
[...]
...Since we have not yet implemented a LUS based model it is inappropriate to apply this variance and a larger variance is required and applied....
[...]
...We currently implement a simplified stutter model that models stutter ratio as linear with respect to the allelic designation rather than longest uninterrupted sequence (LUS) [21,22]....
[...]
...We are uncertain whether currently available LUS data derived from largely African American and Caucasian samples translates simply to Maori, Polynesian and Australian Aboriginal samples because specific sequence data is not available....
[...]
...Empirical data suggests that the variance in a stutter peak in a model based on LUS follows a different pattern to an allelic peak [17]....
[...]
362 citations
Additional excerpts
...stated ‘‘Once reliable continuous methods become available the binary method will have to be viewed as ‘‘second best’’ and will become obsolete’’ [16]....
[...]
321 citations