Kristen Lewis

Joined Program: 2005
Previous Degrees: M.S. Forensic Science, Virginia Commonwealth University; B.S. Biology, Oregon State University
King Lab
kelewis (at) u.washington.edu

Research:
Forensic DNA databases (e.g., Combined DNA Index System, CODIS) are routinely used to identify the perpetrator of a crime by comparing the unknown biological evidence profile to the known convicted offender profiles in the offender database to identify a perfect match at all loci. However, instances occur when no match results and no other case information exists to further investigate the crime. Forensic scientists and law enforcement personnel are now asking: How can we further use the offender database as an investigative tool? One answer is to search for relatives.

Although the current CODIS search algorithm can identify individuals with high allele sharing (in addition to perfect matches), CODIS was not purposely designed to identify relatives. An alternative approach that specifically aims to identify family members is called familial searching. Familial searching has been routinely used to identify victims of mass disasters, human rights violations, and missing persons cases. In this process, the unknown evidence profile is searched against the CODIS database to identify possible close relatives of the true offender, as these individuals would likely share more alleles with the true criminal than an unrelated person would. This method evaluates the number and frequency of shared alleles by calculating the likelihood that the evidence and offender profiles are from specified relatives versus from unrelated individuals. Typically, the true relatives have a high likelihood ratio, but unrelated individuals may generate high likelihood ratios by chance. A majority of these false positives will likely be dismissed after investigation or upon addition of genetic information.

The goal of my work is to provide the US forensic DNA and law enforcement communities with an objective assessment of the utility and limitations of familial searching as an additional investigative tool for solving crime. My research focuses on developing computational approaches to evaluate familial searching of forensic databases comprised of thirteen autosomal short tandem repeat (STR) markers. Due to the high number of false-positive matches resulting from searches for relationships other than parent-offspring, I am investigating the use of more informative marker systems than autosomal STRs.