How Successful Are Kinship Searches (aka Familial Searches)?

Kinship analysis refers to comparing DNA profiles from different individuals to see if one individual might be a close relative of another. It is done all the time in child-support and missing-remains cases. It is done in criminal cases when a rape victim has a child or aborted fetus.

Using the same principles of genetics, kinship searches can be conducted in a law enforcement database of identifying DNA profiles. For brevity, we can call the convicted offenders (or, in some jurisdictions, arrestees) whose DNA profiles are recorded in a database “database inhabitants.” The profile derived from a crime-scene can be compared with all the database profiles to see (1) if a database inhabitant’s profile is an exact match (the usual “cold hit”) or (2) if there is a close enough match (as shown by kinship analysis) that the crime-scene DNA may have come from a very close relative of a database inhabitant. One group of scientists (Bieber et al. 2006) estimated that kinship matching could generate thousands of useful investigative leads nationally. However, the technique is almost never used, leaving proponents and opponents to rely on their choice of anecdotes about its value and accuracy.

A case in point is a recent law review article that emphasizes “one revealing fact: [Denver’s District Attorney, Mitch Morrissey‘s] familial searches did not work. None of the three matches turned out to point toward a relative, much less the source, of the actual crime-scene sample. . . . [I]t failed in three separate cases . . . .” Erin Murphy (2010).

However, these searches did not use a matching strategy designed and optimized to detect kinship. According to Moreau-Horwin (2011):

In 2009, Morrissey launched a familial search research project with the Denver Police Department. A familial search software program was designed by the DA’s office and the Denver Police Crime Lab. The program would only extend to siblings and parents. When a hit is made, family members could not be questioned unless investigators isolate a suspect using traditional detective work.

This software program resulted in the first case ever to use a deliberate familial search in the United States. In February 2008, several cars were burglarized in a Denver apartment complex. In one car, blood stains were left on the front seat. After extracting the DNA profile, the police ran it through the DNA database, but did not get a match. They then processed the sample as a familial search and a brother of the offender was identified. This led law enforcement to 21 year-old, Luis Jaimes-Tinajero. Police received a court order to take his blood and it was a perfect match to the evidence sample. Jaimes-Tinajero pleaded guilty on September 10, 2009 to criminal trespass and was sentenced to two years probation. Although this case was only designed to test the new familial search software, it would probably have received more publicity had it led to the arrest of a rapist or murderer and not a car burglar who apparently stole only $1.40 in change. 1/

So, is Morrissey’s record 0 for 3 or 1 for 1? It is neither. Surely, there were many cases in which the Denver kinship searches drew blanks or false leads. But the usual, full-match searches do not always produce cold hits either. The latter hit rate depends on the fraction of perpetrators of the crimes who are “database inhabitants.” The former hit rate is much more complicated. It depends on (1) sensitivity–that is, the ability of the kinship matching algorithm to hit on a true relative of the crime-scene profile source in the database; and (2) prevalence–the proportion of relatives of database inhabitants (a) who commit crimes with DNA traces and (b) who are not themselves database inhabitants. The sensitivity of various algorithms can be estimated (Bieber et al. 2006; Curran & Buckleton 2008), but the prevalence is more uncertain. The known fact that the prison population contains a substantial fraction of close relatives is high might mean that there are many more as yet unidentified relatives engaged in criminal activities. (Bieber et al. 2006).

Until a state implements a well designed form of kinship matching in a large number of cases, however, the real-life efficacy of the technique will not be known. The only reference I have seen so far to such information is in a news report of a study in the United Kingdom which “found that, out of 100 searches, more than a dozen led to a suspect.” (Miguel 2007).

Notes

1. Ms. Moreau-Howin is a forensic DNA database consultant who maintains a website, DNAforensics.com. For more details on the successful case from a traditional news source, see 9News.com, Car Vandal Nabbed Thanks To Brother’s DNA, Nov. 16, 2009, http://www.9news.com/rss/story.aspx?storyid=127140 (reporting that Luis Jaimes-Tinajero’s “brother was in the system because of a felony conviction for auto theft” and that the “car was one of many at his Denver apartment complex broken into [that] morning”).

References

9News.com, Car Vandal Nabbed Thanks To Brother’s DNA, Nov. 16, 2009, http://www.9news.com/rss/story.aspx?storyid=127140, last accessed Feb. 20, 2011

Frederick R. Bieber et al., Finding Criminals Through DNA of Their Relatives, 312 Science 1313 (2006).

James M. Curran & John S. Buckleton, Effectiveness of Familial Searches, 48 Sci. & Justice 164 (2008).

Ken Miguel, The controversies surrounding ‘partial DNA,’ ABC7 News, Nov. 13, 2007, http://abclocal.go.com/kgo/story?section=news/drive_to_discover&id=5760664, last accessed Feb. 20, 2011

Raphaele Moreau-Horwin States Using Familial Searches, http://www.dnaforensics.com/StatesAndFamilialSearches.aspx, last accessed Feb. 20, 2011

Erin Murphy, Relative Doubt: Familial Searches of DNA Databases, 109 Mich. L. Rev. 291 (2010).

1 thought on “How Successful Are Kinship Searches (aka Familial Searches)?

  1. David H. Kaye

    The California Attorney General announced the second successful kinship search under the restrictive policy in force in that state. This brings the score there to 2/13. The fact that California’s database is the fourth largest in the world could cut against the efficacy of kinship matching. Databases that span large areas are more likely to contain a geographically distant relative, but kinship matching identifies very close kin, who are likely to live in the same region. Broader searches should turn up more nonrelatives who happen to have high kinship likelihood ratios, producing more chaff in the harvest.

    The AG’s press release is at http://ag.ca.gov/newsalerts/print_release.php?id=2051. News accounts are at http://www.santacruzsentinel.com/news/ci_17618326?source=rss, http://www.mercurynews.com/central-coast/ci_17619970, and http://www.sfgate.com/cgi-bin/article.cgi?f=/c/a/2011/03/15/BAHE1IC089.DTL

Comments are closed.