Forensic familial searching may misidentify family members
MedWire News: The current parameters used in forensic familial searching techniques may incorrectly identify unrelated individuals as family members, say researchers.
Forensic familial searching is used to identify a suspect when a close genetic family member is found on an offender/arrestee DNA database. When a DNA profile obtained from the crime does not match the database directly, forensic familial searching techniques can be used to check for related individuals that may provide a link to a suspect who is not on the database.
The Combined DNA Index System (CODIS) is the federally approved system used in the USA, which is based on 13 particular short tandem repeats, the CODIS loci, for forensic identification.
"While all humans share a relatively recent common origin and the vast majority of their genomes, a small proportion of genetic variation differs in frequency across groups as a function of population history," lead author Rori Rohlfs (University of California, Berkeley, USA) told MedWire News.
"These variant (or allele) frequencies are used in calculating the rarity of a particular profile and the likelihood of a genetic profile match between unrelated people… These rarity calculations are contingent on background population genetic marker allele frequency."
In this study, published in the PLoS Genetics, the researchers found that when one particular background population is assumed, but the individuals involved are from a different population, they may seem more related than they really are, ie, a false-positive may be generated.
Rohlfs and colleagues calculated the confidence intervals for familial identification likelihood ratio estimates to investigate how successfully they could identify siblings and unrelated individuals. They used different population samples with varying allele frequencies and applied either accurate or inaccurate allele frequency assumptions.
The team found that socially defined groups have distinctly different underlying population genetics and that using allele frequency distribution assumptions for an incorrect population leads to decreased relationship distinguishability. In particular, Navajo and Vietnamese samples were found to be more genetically distinct to African-American, European-American, and Latino samples, and therefore more prone to inaccuracies.
The researchers say these results indicate that, even when the parameters for identifying relatives are set high, the investigation would result in increased rates of false-positive results.
"To fine-tune parameters, similar studies should be conducted using the actual forensic databases along with data from individuals in relevant populations (say, a random sample of residents), and the exact familial searching methods used in practice," said Rohlfs.
Forensic familial searching is currently formally used in three US states: California, Colorado, and Virginia, but its use is more common in the UK.
In light of this, Rohlfs added: "Our findings show that care is warranted in the application and interpretation of relative identification using the CODIS markers, particularly in the database application of familial searching when the appropriate background population for a found biological sample is unknown."
By Chloe McIvor