From Genetic Association to Forensic Prediction: Computational Methods and Tools for Identifying Phenotypically Informative Single Nucleotide Polymorphisms

Pigmentation genetics has become an important pillar in the field of forensic genomics for its application in DNA-based prediction of externally visible characteristics (EVCs). EVCs such as hair color, eye color, and skin color are complex traits that are influenced by several loci. When traditional short tandem repeat DNA profiling does not reveal any matches, pigmentation-associated loci can be informative of an individual's EVCs through a process known as forensic DNA phenotyping (FDP). Current FDP panels contain a combined set of over 40 polymorphisms that have been identified as being significantly associated with skin, hair, and eye color. A comprehensive understanding of the genetics underlying pigmentation traits is required to improve the precision and accuracy of FDP estimations. Presented herein is a summary of methods and tools for conducting a genome-wide association study (GWAS) to identify forensically relevant, phenotypically informative single nucleotide polymorphisms. The pipeline described focuses on post-genotyping (i.e., in silico) analyses, with emphasis on association analyses, post-association analyses, and first-pass functional annotation. Using eye color as an example, we demonstrate how the pipeline uses GWAS data to draw preliminary conclusions regarding the location and function of pigmentation-associated variants. The experiment specifically investigates eye color associated variants in individuals with a blue eye color background (rs12913832:GG genotype) in a Canadian dataset. While methodologies and tools available for GWAS and post-GWAS processing continue to evolve and advance, the presented approaches have been applied successfully in numerous association analyses among hundreds of thousands of individuals in a wide range of disciplines. As such, they may offer a road map for future genomics investigations of pigmentation traits as well as other EVCs, ultimately serving to improve statistical predictions of phenotypes in forensic settings.