DOI

https://doi.org/10.25772/WH9T-YB14

Defense Date

2020

Document Type

Directed Research Project

First Advisor

Dr. Tracey Dawson Cruz

Second Advisor

Dr. Sarah Seashols-Williams

Third Advisor

Dr. Edward Boone

Fourth Advisor

Ms. Hannah Wines, MS

Abstract

In the conventional forensic DNA workflow, the number of contributors in a sample is unknown until the final step of STR analysis. We propose a high-resolution melt curve (HRM) mixture screening assay, which uses support vector machine (SVM) modeling of melt morphologies of D5S818 and D18S51 amplicons integrated into a common qPCR-based human DNA quantification kit, to differentiate between single-source samples (and their genotypes) and mixtures at an earlier stage in the DNA workflow (quantification). Previously, using data generated from whole melt curves, 87.5% of single-source samples and 100% of 1:1 mixture samples (2 contributors) classified accurately.

In this study, the HRM assay above was evaluated using Quantifiler Trio™ and Investigator Quantiplex® on a frequently used qPCR platform. When integrated into Quantifiler Trio™, the assay produced inaccurate quantification values, and undesired melt products were formed, necessitating the use of Investigator Quantiplex® on the new qPCR platform for subsequent studies. Genotype prediction accuracies were not significantly altered by this integrated assay on the more frequently used qPCR platform. As a result, using data generated from whole melt curves, 42.1% of single-source samples and 60% of the 1:1 mixture samples (2 contributors) classified accurately. However, these rates dropped when various mixture ratios were tested, demonstrating that as the minor contributor was reduced, the assay was unable to accurately distinguish between mixtures and single-source samples. Moving forward, it may be necessary to incorporate other mixture ratios into the training set as a way to increase prediction accuracies across a range of mixture ratios.

Rights

© The Author(s)

Is Part Of

VCU Master of Science in Forensic Science Directed Research Projects

Date of Submission

4-29-2020

Share

COinS