Show simple item record

dc.contributor.advisorGibson, Greg
dc.contributor.authorWilliams, Diana Wagner
dc.date.accessioned2018-05-31T18:13:59Z
dc.date.available2018-05-31T18:13:59Z
dc.date.created2018-05
dc.date.issued2018-03-29
dc.date.submittedMay 2018
dc.identifier.urihttp://hdl.handle.net/1853/59879
dc.description.abstractThe microbiome, specifically the human microbiome, is an area of intense research with potential forensic applications. If the microbiome of a given body area is individualizing, then the transfer of microbiomes between individuals could reveal contact between a victim and suspect. An important application would be in sexual assault cases where no semen is detected or analyzable. This thesis aimed to address whether the identification of individuals through their pubic hair/pubic area microbiome would support detection of biological transfer. Sequencing of 16S rRNA gene fragments derived from pubic hair and pubic mound swabs of 43 individuals, including 12 sexually active couples and 19 unpaired singles, each providing 2 to 5 sets of samples over two months, allowed me to evaluate under which circumstances forensic individualization may be feasible. Random Forest modeling demonstrated 90% accuracy in assigning replicate samples to individuals, but had variable accuracy in pairing sexually active couples. Hierarchical clustering also provided some support for couples being more similar than random pairs of individuals, and a trend toward greater sharing with higher sexual activity was detected. Couples who never reported sexual activity in the week preceding sample collection did not cluster together, whereas couples who reported at least one instance of sexual activity in the week preceding sample collection shared clusters at least 25% of the time. Unexpectedly, elbow and ear swabs were as similar within couples as pubic samples, although they clustered irrespective of the degree of sexual activity. These results suggest that detection of microbiome transfer during sexual activity is not guaranteed and will not generally identify the source from the general population. To explore the potential to identify or exclude suspects from a small pool of candidates, I performed in silico mixture modeling of pseudo-couples. These analyses implied that a minimum of 5% contribution from a donor is required to be able to detect any change in the host’s microbiome and at least 25% contribution is required for reliable detection. Evaluation of the SourceTracker tool for prediction of contributors to a microbiome profile consisting of Deblur OTU assignments also showed promise both for correctly identifying two or more contributors to a sample and for exclusion of non-contributors. While further work will be needed to refine these models, their potential is demonstrated for questions such as identification of a suspect from the general population, identification of a culprit from a small pool of suspects, and exculpation of a falsely accused individual from a small pool of suspects. Additional limitations and concerns, such as best practices for analysis of low biomass samples, how most appropriately to handle contamination, and other crime laboratory considerations, are addressed in the concluding chapter. This thesis shows the potential of the pubic hair/pubic area microbiome and represents the first steps towards the development of a new tool for crime laboratories.
dc.format.mimetypeapplication/pdf
dc.language.isoen_US
dc.publisherGeorgia Institute of Technology
dc.subjectMicrobiome
dc.subjectForensics
dc.subjectPubic hair
dc.titleIndividualization of the pubic hair/pubic area microbiome and its potential as an indicator of sexual contact
dc.typeDissertation
dc.description.degreePh.D.
dc.contributor.departmentBiology
thesis.degree.levelDoctoral
dc.contributor.committeeMemberStewart, Frank
dc.contributor.committeeMemberLachance, Joseph
dc.contributor.committeeMemberPaaby, Annalise
dc.contributor.committeeMemberKalafut, Timothy
dc.date.updated2018-05-31T18:13:59Z


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record