As submitted by the proposer: Soil is a valuable form of trace evidence owing to its potential to link a victim, suspect, or evidence to a crime scene. Traditional soil analysis only assesses physical and chemical characteristics in soils, making individualization difficult or impossible. Bacterial populations within soil offer deeper resolution, allowing more objective, definitive characterization of soil to a location of origin. A snapshot of the bacterial population in a soil sample can be produced through next-generation sequencing, which creates massive amounts of data while remaining cost effective. Recent NIJ-funded research at Michigan State University revealed the potential of next-generation sequencing in producing soil bacterial profiles. Soils from diverse and similar habitats, across time and space within a habitat, and from stored evidentiary items were tested, all of which resulted in successful differentiation of diverse soils and accurate assignment to locations of origin. However, only a limited number of variables that may exist at a crime scene could be considered. The proposed research will expand on these findings and examine previously unconsidered factors that may affect bacterial profiles. First, the minimum amount of soil needed to produce a bacterial profile will be determined. Next, depth samples from various habitats, as may occur with burials, will be tested. Prior research showed that woodlot soil on evidence had predictable bacterial changes with time, however the potential of this biological clock needs to be examined using other soil types and over longer periods of time. The effect of perspiration or blood on soil profiling will also be examined. Finally, several blind studies assessing all of these factors will be performed.
Data analysis must also be rigorously examined if any new technique is to gain a footing in forensics. Previously, combinations of analysis methods proved to be effective for interpreting the massive datasets produced from next-generation sequencing. Bacterial abundance charts and nonmetric multidimensional scaling plots provided visualization tools, useful for an expert witness, while a supervised classification technique provided objective assignments of soils back to their location of origin. None of these produce a statistical measure of soil similarity/difference or confidence in assignment however, thus identifying and testing several supervised classification techniques is critical for determining which best meets the needs of the forensic community.
This project contains a research and/or development component, as defined in applicable law.