Exploring STR signal in the single‐ and multicopy number regimes: Deductions from an in silico model of the entire DNA laboratory process

Abstract

Short tandem repeat (STR) profiling from DNA samples has long been the bedrock of human identification. The laboratory process is composed of multiple procedures that include quantification, sample dilution, PCR, electrophoresis, and fragment analysis. The end product is a short tandem repeat electropherogram comprised of signal from allele, artifacts, and instrument noise. In order to optimize or alter laboratory protocols, a large number of validation samples must be created at significant expense. As a tool to support that process and to enable the exploration of complex scenarios without costly sample creation, a mechanistic stochastic model that incorporates each of the aforementioned processing features is described herein. The model allows rapid in silico simulation of electropherograms from multicontributor samples and enables detailed investigations of involved scenarios. An implementation of the model that is parameterized by extensive laboratory data is publically available. To illustrate its utility, the model was employed in order to evaluate the effects of sample dilutions, injection time, and cycle number on peak height, and the nature of stutter ratios at low template. We verify the model's findings by comparison with experimentally generated data.

Production of high-fidelity electropherograms results in improved and consistent DNA interpretation: Standardizing the forensic validation process

“…The computational tool RESOLVIt carries out the simulation of a large number of artificial EPGs by in silico execution of a laboratory's amplification and electrophoresis protocols according to the SEEIt model specified in Ref. [31]. Each simulation begins with the random sampling of two different alleles at every locus.…”

“…Since the objective of this work is to investigate the ability of a laboratory protocol to distinguish signal 1copy from noise, dropout that stems from sampling effects is ignored and the number of copies of an allele that undergo amplification is forced to be 1; thus, T c=0 = 1 for every allele. The PCR process is simulated for the chosen number of cycles via a multitype Galton-Watson process, as detailed in [31]. That is, at each cycle, each amplicon of an allele is successfully copied with a probability corresponding to the amplification efficiency, and if copied it gives rise to stutter with a given stutter slippage probability, whereupon the stutter amplicons are then amplified at later rounds of PCR.…”

“…That paper also reports that given a noise peak is observed, the conditional distribution of the peak height in RFU is well described by a lognormally distributed random variable. The EPG simulation model we employ here, and in the model described in [31], describes the post-peak detection processed signal. Thus following those studies, the occurrence of each noise peak is modeled as an independent random variable that with probability 1-P N is zero and with probability P N it is lognormally distributed with mean μ noise and variance σ noise 2 .…”

“…Recently a stochastic model of EPG generation encompassing all aspects of the entire forensic DNA laboratory process, from quantification to peak detection, was described [31], where it was shown that good signal to noise resolution was readily achieved by the application of slight modifications to the laboratory conditions under which the DNA samples are processed. We extend that work here by developing a validation strategy to determine an optimized AT for the forensic DNA process.…”

“…This AT is established by careful evaluation of the false detection rates only after good noise to signal 1copy (i.e., RFU signal obtained from 1 copy of amplifiable DNA) resolution is acquired. To determine the laboratory parameters that result in a sensitivity, α, that produces adequate resolution, we generate simulated capillary electrophoresis (CE) signal from T c=0 = 1 and T c=0 = 0 copies of DNA via the stochastic model described in [31]. For each laboratory condition, we compute the false positive and false negative detection rates for various S T values for the large synthesized dataset to acquire a good approximation of the detection error rates.…”