Should Police Use DNA to Investigate a Suspect’s Family Members?

A shorter version of this was published in Nature Reviews Genetics [abstract].

In July 2010, California police used a new forensic technique called familial searching to capture the “Grim Sleeper,” a serial killer who had evaded them for a quarter century. With DNA obtained from a discarded pizza crust, investigators found in the state’s offender database not the profile of the killer himself but instead, through a partial genetic match, that of his son.  Based on the identification of a close genetic relative, a principal suspect was identified (1).

Forensic familial searching is practiced in some US jurisdictions and a number of other countries, and its use is increasing. Not surprisingly, this has raised significant scientific and social concerns. Careful consideration of these issues is especially urgent now, in light of the recent Supreme Court decision allowing the collection of DNA from arrestees prior to conviction (2,3).

How does familial searching work? In most forensic cases, investigators analyze DNA from a crime scene at 13 genetic markers determined by the federally managed Combined DNA Index System (CODIS) (4). They compare this genetic information to profiles stored in large databases, in an attempt to identify a complete 13-locus match.

In familial searching, analysts instead search the database to find profiles that are partial matches to a crime scene DNA sample. These partially matched profiles may come from genetic relatives of the person whose DNA was found at the crime scene. Police can then investigate these family members in an effort to identify the suspect (5).

In the Grim Sleeper case, familial searching led police to the suspected killer. But it is also expected that database searches for partial matches will find profiles of one or more unrelated individuals that coincidentally partially match the crime scene sample. Depending on the likelihood of these coincidental partial matches and the follow-up investigative procedures used, the relatives of those people would then be investigated, exposing innocent individuals to unwarranted criminal justice attention and posing risks of unjustified search and seizure.    

Three main concerns surround the increasing use of partial matches and familial searching in criminal investigations.

• First, familial searching is, by definition, less certain than DNA matching at 13 loci.  

• Second, the increased uncertainty, and hence greater extent to which an otherwise unrelated individual may be erroneously identified as a relative, makes familial searching more prone to errors.  Potential errors are related to assumptions underlying the methodology (described below), which may render some individuals more vulnerable to being wrongly implicated than others.  

• Third, because current forensic state and federal databases over-represent individuals of certain racial backgrounds, particular ethnic groups and/or communities are disproportionately exposed to errors associated with familial searching.