Genetic profiling of skin microbiomes for forensic human identification

Date of Award


Degree Type


Degree Name

Doctor of Philosophy

Field of Study

Biomedical Sciences


Graduate School of Biomedical Sciences

First Advisor

Bruce Budowle

Second Advisor

Jianye Ge

Third Advisor

Johnny J. He


The field of microbial forensics has expanded from a focus in biodefense and biocrime attribution to include various metagenomics and microbiome applications made possible by advancements in sequencing and bioinformatics technologies. Recent developments in metagenomics and microbiome research with application to the forensic sciences, include post-mortem interval, body fluid identification, recent geolocation, and human identification. The primary goal of the dissertation described herein was to assess the feasibility of human identification from skin microbiomes using both shotgun metagenomic sequencing and targeted enrichment strategies. The main studies of this dissertation were conducted under the hypothesis that genes from stable, universal microbial species from the core skin microbiome can differentiate skin microbiomes of individuals and be applied towards forensic human identification purposes.

The initial study presented describes the development of a tool, AutoCurE, used to identify errors in bacterial genome metadata from public databases and curate the data for subsequent use in comparative genomic studies. This study highlights the types of inconsistencies and errors which may be present in public genome databases and describes the development of a curated local bacterial database for use in subsequent studies. This doctoral research herein presents the development of a novel approach for human identification using stable, universal clade-specific markers from skin microbiomes. Initially, publically available shotgun metagenomic datasets generated from skin microbiome samples collected from 17 body sites from 12 individuals, sampled over three time points over the course of ~3-year period, were mined to identify stable, universal microbial markers. Supervised learning, specifically regularized multinomial logistic regression and 1-nearest-neighbor classification, were performed using the nucleotide diversities of clade-specific markers to predict the correct classification of skin microbiomes to their respective host individuals. Reduced subsets of markers were developed into a novel targeted metagenomics sequencing panel, the hidSkinPlex, to generate individual-specific skin microbiome profiles to use for human identification. Finally, the hidSkinPlex was evaluated on skin microbiome samples collected from eight individuals and three body sites, in triplicate, to demonstrate a proof-of-concept to differentiate individuals with high accuracy.

The hidSkinPlex, comprised of 282 bacterial and 4 phage markers from 22 family-, genus-, species-, and subspecies-level clades, was used to correctly identify skin microbiomes from their respective donors with up to 92%, 96%, and 100% accuracy using samples from the foot, manubrium, and hand, respectively. Additionally, skin microbiomes were classified with up to 97% accuracy when the body site was unknown, and body site origin could be predicted with up to 86% accuracy. The hidSkinPlex is the first targeted metagenomics sequencing panel and method designed specifically for skin microbiomes with the intent of forensic human identification applications


Schmedes, Sarah E. Genetic Profiling of Skin Microbiomes for Forensic Human Identification. Doctor of Philosophy (Biomedical Sciences), September 2017, 162 pp., 12 tables, 27 figures, 130 references. Available for download December 2018.

This document is currently not available here.