The increased sensitivities of recently developed polymerase chain reaction (PCR) and separation techniques have afforded forensic deoxyribonucleic acid (DNA) analysts an opportunity to detect low template deoxyribonucleic acid (LT-DNA) samples. However, with LT-DNA samples stochastic effects become more prevalent, compromising the reliability and robustness of these techniques. In addition, these innovations have presented analysts with an increased incidence of higher-order mixtures. These types of mixtures, confounded by LT-DNA effects, continue to test the interpretation step of the DNA analysis pipeline. The combination of allele drop-out, allele drop-in and allele sharing create such complex samples that it necessitates the transition from traditional, threshold-based, interpretational methods to a probabilistic approach. Therefore, this study has two objectives: 1) to optimize the computational tool NOCIt, designed to provide a probability distribution on the number of contributors (NOC) to a sample and 2) to investigate the source of the variability present in quantitative real-time polymerase chain reaction (qPCR) in hopes of minimizing variations observed when determining the quantity of an unknown DNA sample.
The NOCIt graphical user interface (GUI) was validated during its developmental phase. With no current forensic guidelines for the validation of computational tools, the protocol was designed based on a guide developed by the Center for Devices and Radiological Health for the Food and Health Administration (FDA). The protocol required using a variety of test types and detailed documentation of the methods used, the inputs, outputs and results. A total of 325 tests were completed across 11 different software distributions.
As a critical software system, NOCIt’s settings also had to be optimized because of its ability to substantially influence the interpretation, the statistical conclusions and the accuracy of the results. Two different settings (Condition 3 and Condition 4) were tested on AmpFlstr® Identifiler® Plus and PowerPlex® 16 HS PCR amplified samples. The differences between the two conditions were the parameter Number of Samples in Batch and the Multiplicative Factor. All other settings were kept the same. The reproducibility and accuracy were examined to determine which condition was more reliable. Both settings had similar results, with both performing better in different categories.
As with interpretation, quantification is an integral part of the human identification pipeline. Previous studies have shown that the Quantifiler® recommended protocol of generating a standard curve for every quantification (referred to in this study as the Recommended Method) introduces additional sources of variability compared to protocols that utilize one, external, validated curve (referred to in this study as the Experimental Method). To identify the major source of variability inherent in the quantitative polymerase chain reaction (qPCR) process, these two methods of determining the quantity of a DNA sample were investigated using the Quantifiler® Duo and Quantifiler® Trio DNA Quantification assays. Four tenfold serial dilutions were quantified in five independent runs using the Quantifiler® Duo DNA Quantification kit and five independent runs using the Quantifiler® Trio DNA Quantification kit. Quantification with Quantifiler® Duo reported less variability using the Experimental Method than the Recommended Method. Conversely, quantification with Quantifiler® Trio exhibited approximately equal variability between both methods.
To assess whether the errors associated with generating a calibration played a substantive role in introducing additional variability, the test samples were also quantified using digital polymerase chain reaction (dPCR). The data for the more dilute samples were indistinguishable from the noise associated with the instrument. The more concentrated samples showed less variability than the samples quantified with Quantifiler® Duo and approximately the same as those amplified with Quantifiler® Trio. This suggests that both qPCR and dPCR processes can be used to quantify DNA amounts, however, fundamental differences in the ways each determines the values suggests that noise is an inherent and measurable part of dPCR. Thus, for purposes of DNA quantification signal thresholds will need to be determined prior to implementation.
Identifer | oai:union.ndltd.org:bu.edu/oai:open.bu.edu:2144/26951 |
Date | 02 November 2017 |
Creators | Yearwood-Garcia, Xia Marie |
Source Sets | Boston University |
Language | en_US |
Detected Language | English |
Type | Thesis/Dissertation |
Page generated in 0.0016 seconds