Genetic distance to improve human identification from the skin microbiome

Human identification is commonly performed using nuclear DNA (nDNA) and sometimes with mitochondrial DNA (mtDNA), There are times when nDNA is too degraded to yield a full profile. Although more sensitive, mtDNA has limited discriminatory power. The skin microbiome is a supplemental source of DNA that can be exploited to accurately identify the donor of forensic biological evidence. Research in human identification using the human skin microbiome has focused on the abundance and unique taxonomic differences of populations of species residing on individuals. The data support that individuals can be differentiated based on their skin microbiomes. However, classification accuracies still are relatively low. This project proposes to use superior genetic metrics to improve classification accuracy. Improvements in attribution are feasible so that human microbiomes can be integrated into the forensic laboratory as another tool for human identification. The goal of this research project is to identify single nucleotide polymorphisms (SNPs) of selected microorganisms from the human skin microbiome that can be used for individualization of the host and develop a sensitive massive parallel sequencing (MPS) panel of microbial markers for human identification. The project proposes the use of genetic distance, such as Wright’s fixation index (FST) and population branch statistic (PBS), to select informative SNPs from the previously designed hidSkinPlex panel to improve the accuracy of attributing a sample to an individual. The proposed research study has two specific aims using a data set of 51 individuals sampled at three body sites in triplicate and sequenced using the hidSkinPlex panel. The aims of this proposed project are: (1) define informative SNPs from previously selected microorganisms that allow for the differentiation of individuals by (1.1) using FST to calculate the expected heterozygosity within and between populations of microorganisms on individuals for a particular SNP and then (1.2) using the PBS to estimate a genetic distance between two individuals at specific SNPs; and (2) evaluate the predictive power of selected SNPs by using machine learning techniques, such as single vector machines, logistic regression, and/or nearest neighbor to improve accurate identification of unknown samples. This project will lead to the development of an improved hidSkinPlex panel with increased classification accuracy for human identification. The resulting work will be disseminated at local and national conferences. Additionally, R packages created for FST and PBS calculations from MPS data will be published for public use for enabling reproducible science. Note: This project contains a research and/or development component, as defined in applicable law, and complies with Part 200 Uniform Requirements - 2 CFR 200.210(a)(14). CA/NCF