When teaching this personal genomics seminar for the first time in Fall 2014, I contemplated the potential negative consequences of consenting to doing a 23andMe (or similar) test. Essentially, I wanted students to consider potential outcomes of testing that, if the outcome were to become a reality, it would make them resent having done a test. I'm enabling students to take a test that may produce findings that could drastically change personal perspectives and I want each student to contemplate how they, and family members, might respond. I wrote several different scenarios that a student could potentially face; finding an unknown close relative, discovering a surprising ancestry composition, revealing a genetic disease, etc. During the first class meeting students discussed the different scenarios. Furthermore, we discussed several other potential concerns that, at the time, I considered hypothetical. One such hypothetical concern was that your genetic data could be used by law enforcement for discovery purposes in a forensic investigation.
Now, a story has emerged that turns this hypothetical into a reality. This is the story of Michael Usry, a filmmaker that apparently became a suspect in an Idaho Falls murder investigation by police identifying a DNA match in a consumer genetic database that included one of his relatives. This case clearly demonstrates that law enforcement can, and will, access consumer genetic data to investigate cases where they have DNA available from a crime scene, but no suspect. Enabling police to develop a suspect list of you or your close relatives based on a 23andMe or similar genetic test is no longer a hypothetical concern, but a reality. Before consenting to a personal genetic test, consider whether you are okay with your data being used to develop a suspect list in a criminal case where a DNA sample is available. If a nephew or cousin ends up on this suspect list, and with further investigation becomes the primary suspect and is convicted of the crime, your use of a consumer DNA test provided the initial tip that a relative was the perpetrator. Are you okay with that?
Details of this case are also enlightening. The police in this case used markers on the Y chromosome, analyzed from a DNA sample from the crime scene, to search the database of Sorenson Molecular Genealogy Foundation. A match was found and a suspect list of related males was generated. To me, this illustrates a gross lack of understanding of Y chromosome matching on the part of the police. For any particular Y chromosome type, there will many men that share the same type. In most cases, the same Y chromosome type will be shared by hundreds if not thousands of different men. This is also likely the case for Michael Usry, because subsequent DNA analysis using the standard test for forensic investigation demonstrated that the DNA sample from the crime scene did not match his DNA even though there is a matching Y chromosome. Genetic testing of the Y chromosome is not sufficiently fine grained to develop a small list of suspects when a match is found. In other words, matches and their close relatives will most often be falsely identified as potential suspects.
However, the fact that the police turned to a database of typed Y chromosomes demonstrates an eagerness to utilize commercial genetic testing platforms for forensic purposes. SNP tests available through AncestryDNA, 23andMe, and Family Tree DNA provide sufficient resolution to clearly identify exact matches with a DNA sample. In reality, these genetic tests are much more powerful for individual identification than the standard DNA forensics tests. Given that these tests analyze over 700,000 variants spread across the genome, whereas the standard forensics test analyzes 13 markers, each individual's results are completely unique (excepting identical twins). Based on some blog posts on this subject, this is a fact that either is not widely understood, or is being downplayed, in the genetic genealogy community. Moreover, just as these tests are used to discover relatives in support of genealogical research, a non-exact match has a predicted relationship and the prediction is quite accurate for close relatives. For example, these tests are able to clearly distinguish between an uncle and a 1st cousin. If law enforcement identify a match in a commercial database, the relationship of the offender to the individual in the database can be predicted. If the offender is a close relative, a small list of suspects may then emerge. Developing lists of suspects based on relatives to a DNA match is not new to forensic investigations. Law enforcement has been doing this for years with their own DNA database. Investigation of suspects related to a DNA match may uncover additional evidence linking one suspect to crime.
Combined the databases of 23andMe and AncestryDNA are now approaching 2 million genetic profiles. For the customers themselves, and apparently also for the police, these users provide a lot of opportunity for discovery of relatives.
The future is here.