Wednesday, 25 September 2019

US Department of Justice Interim Policy on Forensic Genetic Genealogical DNA Analysis and Searching

The US Department of Justice have issued an Interim Policy on Forensic Genetic Genealogical DNA Analysis and Searching. The press release can be found here:

Here is a direct link to the policy statement:

The announcement was made by Ted Hunt, Senior Advisor to the Attorney General on Forensic Science at the Department of Justice, during a talk given at the International Symposium on Human Identification in Palm Springs, California.

Investigative genetic genealogy or forensic genealogy is a powerful tool which can help to solve crimes and identify missing persons but, as with any such tool, it needs to be used responsibly. Earlier this year I wrote an article for Forensic Science International on Using genetic genealogy databases in missing persons cases and to develop suspect leads in violent crimes. I explain in the article how the methodology works but I also discuss some of the privacy implications and the need for ethical and regulatory oversight. These new guidelines from the DOJ are long overdue and this document will be a good starting point for further debate but there are a number of important issues that have not been addressed and which I will discuss briefly below.

Informed consent
The following sentence in the report is highly misleading: "The FGG [forensic genetic genealogical] profile is... compared by automation against the genetic profiles of individuals who have voluntarily submitted their biological samples or entered their genetic profiles into these GG services (‘service users’).

FamilyTreeDNA changed their terms of service in March this year. All European Union customers who tested prior to the change were automatically opted out of law enforcement matching. Going forwards, all customers, including all EU users, were automatically opted in to law enforcement matching. New settings were introduced so that customers could opt out. Previously if you didn't want to share your results with law enforcement you had to switch off matching altogether, thus losing all the benefits of the genetic genealogy database. While FTDNA customers have voluntarily uploaded their samples for their family history research, because of the automatic opt in they cannot in any way be considered to have given consent to have their profiles matched with law enforcement kits. The DOJ perhaps doesn't realise that FTDNA has an international database. There are international conventions on the transfer of DNA data for people in criminal databases (eg, through Interpol or the PrĂ¼m Convention in Europe). Innocent people in genetic genealogy databases should expect similar protections if their data is to be used by law enforcement agencies in foreign countries. I would like to see the DOJ make a requirement that, at the very least, DNA profiles of non-US citizens can only be used if the customers have given explicit informed consent. Ideally they should insist that FTDNA follows the standard convention and adopts an opt in policy for everyone.

GEDmatch customers have voluntarily uploaded their profiles to the database and, following the decision in May this year to require all users to opt in to law enforcement matching, anyone who has exercised this option has given their consent to have their data used for law enforcement investigations.

The DOJ guidelines include the following criteria which must be met before a genetic genealogy search can go ahead:

Before an investigative agency may attempt to use FGGS [forensic genetic genealogical DNA analysis and searching], the forensic profile derived from the candidate forensic sample must have been uploaded to CODIS, and subsequent CODIS searches must have failed to produce a probative and confirmed DNA match.

The investigative agency with jurisdiction of either the crime or the location where the unidentified human remains were discovered (if different) must have pursued reasonable investigative leads to solve the case or to identify the unidentified human remains.

Proportionality is a key concept in the criminal justice system and priority should always be given to the most effective and least intrusive methods. It is notable that in a number of the forensic cases where genetic genealogy databases have been used, the police have failed to use existing and well proven methods.

In some cases, the suspect had already had prior contact with the police or had even been in prison but the police had failed to get a DNA sample. See for example herehere and here. Jerry McFadden was identified through genetic genealogy as the murderer of Anna Marie Hlavka. But McFadden was executed in 1999 without having his DNA taken.

If matches are to be found in the CODIS database then surely it makes sense to ensure that all convicted offenders are in the database. Yet, according to a report from Forensic Magazine in 2017 there is a big backlog of prisoners who have not been tested: states hold prisoners whose DNA had not been collected, and who were not in CODIS. Most often, these states had no retroactivity conditions in their DNA laws, which were generally enacted in the 1990s and were never extended into the past to include criminals already locked up. But there are other cases where prisoners refused to give samples, or authorities simply didn’t get the testing done for logistical reasons. For others, there were simply collection delays.
There is also a huge backlog of untested sexual assault kits in the US. Astonishingly it's not even possible to determine the size of the backlog but there are probably tens of thousands of kits which still haven't been tested. The End The Backlog website keeps track of the problem.

I cannot understand why these backlogs have been allowed to develop. There must be many victims of crime who have yet to receive justice and potentially many innocent people who could be exonerated if this testing were done.

Familial searching has far fewer privacy implications than genetic genealogy because the people whose DNA is being searched have already committed a crime and can therefore be considered as having forfeited the right to privacy. Familial searching involves a search of the police database to identify people who are first-degree relatives of the suspect (eg, the parent, child or sibling). It works on the basis that crime tends to run in families so, even if the perpetrator himself is not in the database, it's quite possible that he could be identified through a match with another family member.

Some of the cases where genetic genealogy has been used could have been solved much earlier if familial searching had been used. A number of the suspects, including the Golden State Killer, had a brother who already had a criminal record but sadly his DNA was not in the police database.

Familial searching is currently only done in a handful of states such as California, Texas and Colorado. The practice is banned in Maryland and in Washington DC. Familial searching could potentially produce tens of thousands of useful investigative leads. The Florida Department of Law Enforcement have set an excellent example by adopting a policy whereby genetic genealogy is only used if a familial search has produced a negative result. It would be good to see the DOJ updating their policy to follow Florida's example and insist that genetic genealogy should only be used after an unsuccessful familial search.

The National DNA Index System (NDIS), the federal CODIS database, has over 16 million DNA profiles. DNA searches in the national database would probably produce valuable leads in thousands of cases. For some reason the FBI has resisted allowing the use of NDIS for familial searching. However, if the DOJ is happy for genetic genealogy databases to be used then the FBI's position is clearly untenable and familial searching should be introduced as a matter of priority.

Ethical oversight
The policy makes no provision for ethical oversight. Genetic genealogy searches are left to the discretion of the investigative agencies and the prosecutors. Agencies are required only to use companies which "provide explicit notice to their service users and the public that law enforcement may use their service sites". Law enforcement and prosecutors are necessarily going to want to push for the use of genetic genealogy. There is an urgent need for some form of independent oversight to provide balance in the system. Again, Florida is setting an excellent example. Because genetic genealogy is only used in Florida when familial searching has been unsuccessful, all searches must be approved by a Familial Search Review Committee. I would like to see other states follow this example.

Mechanics of genetic genealogy searches
The policy recommends that "The investigative agency shall, if possible, configure service site user settings that control access to FGG profile data and associated account information in a manner that will prevent it from being viewed by other service users." At GEDmatch kits can be uploaded as research kits which means that the user will receive a list of matches but will remain invisible to the people on the match list. At FamilyTreeDNA law enforcement kits appear in the match list of everyone who has not opted out of law enforcement matching. They are not distinguished in any way from those of ordinary users. I understand that some people are trying to encourage FTDNA to adopt a similar system to GEDmatch so that the law enforcement kits cannot be seen. The DOJ missed a big opportunity here and could have insisted that the police only use companies which restrict matching in this way which would put pressure on FTDNA to adopt best practices.

The report discusses the procedures for target testing which is sometimes done when the matches are more distant. Testing someone who is potentially a second cousin to the suspect helps to narrow down the search and confirm that the investigators are pursuing the right line.  The report says that "An investigative agency must seek informed consent from third parties before collecting reference samples that will be used for FGGS". However, they do not discuss the tricky subject of incidental findings, such as the discovery that a target tester is not related to his or her presumed biological family in the expected way.

I am sure that there is much more I could write but I just wanted to set down a few initial thoughts about the areas which I think are the most important.

Further commentary

No comments: