DOI
https://doi.org/10.25772/SXXQ-XK60
Author ORCID Identifier
https://orcid.org/0000-0003-0140-515X
Defense Date
2022
Document Type
Dissertation
Degree Name
Doctor of Philosophy
Department
Integrative Life Sciences
First Advisor
Baneshwar Singh, Ph.D.
Second Advisor
Sarah Seashols-Williams, Ph.D.
Third Advisor
David Edwards, Ph.D.
Fourth Advisor
Tomasz Arodz, Ph.D.
Fifth Advisor
Cydne Holt, Ph.D.
Abstract
The confirmatory identification of body fluids is one of the most critical steps in criminal investigations, as it provides not only information on the source of the DNA, but it can corroborate evidence, aid in the reconstruction of a crime, and associate evidence with a person of interest. Current serological methods utilize biochemistry and immunology to identify biological evidence; however, these serological tests are mostly presumptive while only testing for one fluid at a time, potentially wasting precious evidence. More recently, forensic investigators have suggested the use of molecular methods such as mRNA, miRNA, or DNA methylation. While each of these methods is promising, each of these comes with their own limitations. RNA based methods have been shown with poor stability, require high amounts of starting material, and require additional steps in the workflow. Additionally, while DNA methylation has shown promise where other methods have failed, this methodology also requires additional workflow steps, can result in conversion errors, and has many factors impacting methylation levels, making the validation and integration difficult.
To address these factors, this study developed a non-human DNA (bacteria) based body fluid identification method. For this, 812 body fluid samples from over 200 donors were collected and preserved using methods common for evidence collection to mimic the current forensic workflow. For the developmental validation studies, we collected 80, 728, and 200 body fluid samples for impact of manual versus robotic (man v cube) DNA extraction methods, sample storage time and temperature, and various swab types objectives, respectively. Except for urine samples, all body fluids were extracted using the QIAamp DNA Investigator Kit. DNA from urine was extracted using the QIAamp DNA Micro Kit. DNA extracts were quantified for bacterial DNA using a quantitative PCR method as described in Seashols-Williams et al. (2018). Variable region four (V4) of 16S ribosomal DNA (16S rDNA) was amplified using a dual-indexing strategy and then sequenced on the MiSeq FGx sequencing platform using the MiSeq Reagent Kit v2 (500 cycles) and following the manufacturer’s protocol. Machine learning prediction models were used to assess the classification accuracy of the newly developed method. As there was no significant difference in bacterial communities between vaginal fluid, menstrual blood, and female urine, these were combined as female intimate samples. Except in urine, the bacterial structures associated with male and female body fluid samples were not significantly different from one another. The developed XG Boost prediction model accurately identified human body fluid samples with an overall accuracy of more than 88%. The man v cube study showed that there was no significant difference in bacterial DNA yields and bacterial community compositions; however, reagent blanks associated with robotically extracted body fluid samples resulted in significantly different bacterial communities than their manual counterpart. When investigating the impact of storage time and temperature, storage temperature had a significant impact on the bacterial DNA yield, decreasing as the temperature increased regardless of storage time; however, the storage conditions did not significantly impact the bacterial communities associated with each body fluid. Finally, in the swab comparison study, results suggest that while differences in bacterial community structures were observed at later simulated timepoints, the swab type did not significantly impact bacterial DNA yields or the associated bacterial community compositions. This suggests that the previously identified body fluid indicators remain stable regardless of extraction method, storage conditions, or sampling method. This newly developed bacterial signature-based method is fast (no additional steps are needed as the same DNA can be used for both body fluid identification and STR typing), efficient (consumes less sample as one single test can identify all major body fluids), sensitive (needs only 5 pg of bacterial DNA), accurate, remains stable under various conditions, and can be easily added into a forensic high throughput sequencing (HTS) panel.
Rights
© Denise M. Wohlfahrt
Is Part Of
VCU University Archives
Is Part Of
VCU Theses and Dissertations
Date of Submission
12-16-2022