Accurate Trace Evidence Using Regression Approaches in Forensic Studies

Abstract

Human microbiome data has become popular in forensic study due to its use in estimating the time since death, examination of trace evidence and human identification. The human microbiome is commonly presented independent of seasonal and environmental changes and is highly individual. With the growth of next-generation sequencing (NGS) technology, there are more human microbiome data available to use. 16S rRNA sequencing helps to detect and identify bacteria in human microbial communities. The wide use of human microbiome data in the forensic studies calls for powerful statistical and computational methods for analysis. Due to the human microbiome being highly individual, even varying in different body sites, it can serve as an alternative method for human identification and trace evidence in a criminal investigation, particularly, when human DNA samples are absent from the scene of the crime. In this research, we proposed a new analytic method FourStage to link the human microbiome (e.g., on palm) to the microbial evidence sample. The FourStage method has four procedures: 1) select the contributors from a pool of suspects and exclude the innocents at same time, 2) construct a “unknown” profile for possible missing suspect, 3) determine the status of a “missing” suspect, and 4) estimate proportions for each contributor. Through comprehensive simulation studies, we demonstrated that our new method surpasses the currently available approach, even for a situation in which a suspect was a contributor but is excluded from data analysis. This new method will be helpful for researchers/investigators in forensic area to analyze their own data and also provide them a new research angle/direction in trace evidence.