Binary logistic regression models enable miRNA profiling to provide accurate identification of forensically relevant body fluids and tissues

• NCJ Number: 307046
• Journal: Forensic Science International Genetics Supplement Series Volume: 4 Issue: 1 Dated: 2013 Pages: e127-e128
• Author(s): Erin K. Hanson; Kamel Rekab; Jack Ballantyne
• Date Published: 2013
• Length: 2 pages
• Grant Number(s): 2009-DN-BX-K255
• Agencies: NIJ-Sponsored
• Publication Type: Research (Applied/Empirical)
• Award: Identification of Forensically Relevant Fluids and Tissues by Small RNA Profiling

Annotation

In this paper, the authors describe the materials and methodology; they discuss results regarding newly identified miRNAs to further improve the sensitivity and specificity of the body fluid assays; and they present a novel statistical approach involving the use of the logistic regression analysis.

Abstract

Numerous studies have demonstrated the ability to identify the body fluid of origin of forensic biological stains using messenger (mRNA) profiling. However, the size of the amplification product used in these assays (100–400 bases) may not be ideal for use with environmentally degraded samples. MiRNA profiling represents a potential alternative to mRNA profiling, since the small size of the miRNAs (∼22 bases) might still permit their detection in degraded stains. Previously, the authors reported the first study involving the forensic use of microRNA (miRNA) profiling, which required screening of 452 candidates. Since their initial screening, hundreds of novel miRNAs have been identified. We have therefore evaluated additional miRNA candidates to further improve the sensitivity and specificity of the body fluid assays. Consequently, the authors have expanded their body fluid identification panel to include 18 miRNAs (comprising 5 original and 13 novel miRNAs). This panel permits the identification of all forensically relevant body fluids and, uniquely, includes miRNAs for the identification of skin. Using normalized miRNA expression data, the authors constructed body fluid specific binary logistic regression models to permit an accurate identification of the body fluid of interest. Using the developed models, the authors have obtained 100 percent accuracy in predicting the body fluid of interest. (Publisher Abstract Provided)