FINDING KILLERS THROUGH GENETIC GENEALOGY

INTERVIEW

This is my interview with Dr. Ellen Greytak, Director of Bioinformatics and CeCe Moore, Chief Genetic Genealogist at Parabon Nanolabs, Innovative DNA Technologists talking about how they use genetic genealogy to identify the bad guys.

LYNN: How does genetic genealogy differ from familial searches in the CODIS database?

CECE: Familial searches in CODIS are based on a type of genetic marker called an STR (Short Tandem Repeat), of which the genetic profiles traditionally used by law enforcement are only made up of a handful of genetic markers. This type of search is performed only in law enforcement databases and can only detect very close relatives, typically first degree (parent/child or full sibling).

Genetic genealogy utilizes a different type of genetic marker called a SNP (Single Nucleotide Polymorphism) and necessitates hundreds of thousands of these markers across the genome. STR markers are not compatible with investigative genetic genealogy (IGG). This means that more advanced testing than has traditionally been used or available for forensic samples must be performed on the biological evidence from scratch. This type of search is performed in select consumer genetic genealogy databases and can detect very distant relatives, broadening the search and, thus, creating the ability to reverse engineer a person’s family tree from their DNA alone.

LYNN: Can you briefly describe the process you go through when approached by law enforcement to help with a case?

ELLEN:  Investigators come to Parabon when they have a case where there just isn’t enough information to find the right person, and they’re looking to generate new leads. Typically, their lab has generated a traditional DNA profile and searched it against law enforcement databases, but they haven’t gotten a match, so they’re wondering what else they can learn from that DNA.

Our first step is always a careful evaluation of the DNA evidence itself. We need to understand what type of sample it is, how old it is, whether it’s a mixture, and what prior testing has already been done. This helps us determine which Snapshot tools are appropriate and whether the DNA is likely to yield useful results.

Once we have that information, our laboratory partners generate the DNA data needed for analysis. From there, our team applies the appropriate Snapshot services to generate investigative leads. Throughout the process, we work closely with investigators to explain what the results do and do not mean, so they can use the information responsibly and effectively as part of their broader investigation.

LYNN: I saw that you’d helped to solve a case 44 years after it happened. What’s the oldest case that Parabon has had the privilege of solving?

ELLEN: The oldest criminal case we’ve helped to solve was from 1967--more than 50 years old. That was the murder of 20-year-old Susan Galvin in Seattle, WA. The crime happened long before the invention of forensic DNA analysis, but enough evidence had been preserved that cold case detectives were able to go back and find DNA. Using investigative genetic genealogy, we were able to tie that DNA to Frank Wypych, who died in 1987. He died before the advent of DNA databasing, so it’s possible that this case might never have been solved without Parabon’s technology. The Seattle Police had to exhume his body in order to perform a direct DNA comparison and confirm the identity of the perpetrator.

CECE: This question depends on the type of case we are discussing. We helped identify an individual from a 1940 burial who needed to be moved due to construction, so her identity was not lost in that necessary process.

I have helped solve a number of cases from before I was born, including the 1967 homicide of Susan Galvin in Seattle, WA and a case of a little boy John Doe found in Keene Creek Reservoir in OR in 1963. Susan Galvin was a young woman who worked for the Seattle Police Department. She was found raped and murdered in the parking garage near the Space Needle.

Stevie Crawford was an (almost) three-year-old (two years and nine month) boy with significant health challenges whose body had been carefully dressed, wrapped and disposed of in Keene Reservoir, presumably, by his parents since he was never reported missing.

LYNN:  Genetic genealogy has traditionally been used to discover new relatives and build a full family tree. How do you use it to discover the identity of an unknown individual?

CECE: Genetic genealogy was initially created and developed to identify our long dead ancestors, to extend our family trees further back than the documented “paper trail” can take us. We all have what we call “brick walls” in genealogy, where the documentary trail runs out, and we cannot go any further back in time. Genetic genealogy provided a way to scale those brick walls and gain insight into who was on the other side.

In time, a few of us early adopters decided to try using genetic genealogy to help adoptees and others of unknown parentage and grand parentage discover their recent unknown ancestors. The techniques developed for this work laid the foundation for IGG, identifying violent criminals and Jane and John Does. Using these techniques, we can “reverse engineer” the identity of a person who died without their name (and have been termed a Jane or John Doe) from their DNA alone in the same way we can use a person’s DNA to reveal their biological parents and build their family tree back in time. Further, these same techniques can be used to identify a violent criminal who left their DNA behind at a crime scene but has failed to be identified using other law enforcement tools.

LYNN: Can you briefly define for us the following terms:

Snapshot DNA Phenotyping:

ELLEN: With Snapshot DNA Phenotyping, we use DNA to predict a person’s physical appearance (eye color, hair color, skin color, freckling, and overall face shape), as well as their genetic ancestry (where in the world their ancestors came from). This allows investigators to develop a description of an unknown perpetrator or unidentified victim and narrow their search.

This kind of information can be incredibly valuable when there is no witness, as the DNA phenotyping can act as a kind of genetic witness. Instead of searching blindly, investigators can narrow their focus by knowing who they are (and, just as importantly, who they are not) looking for.

Snapshot Kinship:

CECE: In some cases, we require additional data points to zero in on the relevant branch of the family tree and kinship testing is used in those cases to help narrow down the scope of the research by asking member(s) of the family being researched to voluntarily contribute their DNA sample(s). This can also be called reference testing or target testing. Kinship testing can also be used to confirm or refute a theory developed through IGG before further pursuing an investigation into a potential person-of-interest.

ELLEN: With Snapshot Kinship, we determine how two people are biologically related. For example, we can assess whether a perpetrator may be related to a victim or compare unidentified remains to a potential family member to confirm or refute a relationship. A specialized version of Snapshot Kinship is currently being used by the U.S. government to help identify unknown soldiers dating back to World War II, so these fallen service members can finally be returned to their families.

LYNN: What is ‘innovative genetic analysis?’

ELLEN: Traditional forensic DNA analysis treats DNA like a fingerprint that can be compared to suspects or searched against databases to find a match. But if the unknown person isn’t already in a database, then that DNA can’t tell investigators anything beyond whether the individual is biologically male or female.

Our approach treats DNA like a blueprint - it contains a vast amount of information about a person’s appearance and who they’re related to. By analyzing different parts of the genome than what are used in traditional forensics, we can extract investigative clues from DNA even when no match exists in any database.

LYNN: You’ve said that DNA can reveal hair and eye color, race and even some facial features such as freckling. Can you describe in more detail how you predict physical appearance and ancestry from the unidentified DNA evidence.

ELLEN: When people say things like “you have your mother’s eyes,” they’re really talking about DNA. Much of our physical appearance is written into our genetic code - that’s why identical twins, who share the same DNA, look the same. The challenge is identifying which parts of the genome influence specific traits and then work to understand how those genetic factors work together. When I started working on DNA phenotyping in 2012, very little was known beyond light vs. dark eye color prediction. It wasn’t clear whether other traits could be predicted at all, or whether intermediate eye colors like green or hazel could be distinguished.

So, Snapshot began as a research project. We compiled genetic data and physical appearance information from thousands of anonymous research participants and analyzed hundreds of thousands of genetic variants to find those associated with specific traits. We then used machine learning methods to combine those variants into predictive models. Those models were rigorously tested on thousands of individuals with known appearances. Only traits that could be predicted with high accuracy (eye color, hair color, skin color, freckling, and face shape) were included in Snapshot. Each prediction is accompanied by a confidence statement, such as ‘this is a pattern we see in 90% of people with blue eyes but only 1% of people with brown eyes.’

There are limits, of course. DNA doesn’t tell us how someone styles their hair, whether they dye it, or their current age or weight. For consistency, we depict individuals as young adults at a standard body weight. If additional information is available, such as surveillance footage, witness descriptions, or the passage of time, our forensic artist can incorporate that context, including age progression when appropriate.

LYNN: What do you mean when you say that DNA phenotyping takes advantage of modern SNP technology to read the parts of the genome that actually code for the differences between people?

ELLEN: Traditional forensic DNA analysis uses ~20 short tandem repeats, or STRs, to create a DNA profile that can be used for identification. DNA phenotyping uses a different kind of genetic marker called single nucleotide polymorphisms, or SNPs. These are single-letter differences in the DNA sequence - one person might have an “A” at a particular position, while another has a “G.” While most SNPs don’t affect appearance, some are directly involved in physical traits. By analyzing around 1 million SNPs across the genome, we can identify combinations of genetic variants associated with traits like eye color or skin pigmentation. For example, certain SNP patterns are strongly associated with light versus dark eyes, while others contribute to intermediate colors.

LYNN: Please explain how ‘detecting relatedness out to 9th-degree relatives’ (or fourth cousins) helps to solve cases?

CECE: The DNA (SNP profile) of the unknown person-of-interest is compared against one or more genetic genealogy databases in order to find relatives. These “matches” are typically second cousins and more distantly related (even further out than fourth cousins). Obviously, the closer related the matches are to the Subject, the more straightforward the identification can often be.

The basis is this: The only reason that two individuals share a significant amount of DNA (can be less than 1% total) is if they have a common ancestor somewhere in their family trees. This means that the individuals who show up on the “match list” sharing DNA with the unknown Subject, share a portion of their family tree with the unknown Subject. Our job as genetic genealogists is to determine which portion of their tree is shared. From that analysis, we can then piece that unknown Subject’s tree back together ancestor-by-ancestor and can eventually zero in on one individual or set of siblings who are related to all of those significant matches. I like to say that we can reverse-engineer the tree of the unknown suspect based on who they share DNA with.LYNN: Can you talk a little bit about Parabon’s Snapshot Facial Reconstruction service?

ELLEN: Forensic anthropologists can estimate characteristics such as age, biological sex, height, and ancestry based on the skeleton of unidentified remains. A forensic artist can then use the skull as a foundation to create a facial reconstruction - essentially an informed approximation of what the person may have looked like in life. 

Snapshot Facial Reconstruction builds on this traditional process by adding information from DNA. Using forensic DNA phenotyping, Parabon can predict traits such as eye color, hair color, skin pigmentation, freckling, and genetic ancestry - details that cannot be determined from the skull alone. Our forensic artist integrates these DNA-based predictions with the anatomical structure of the skull to create a more detailed facial reconstruction. The result is not a photograph, but a scientifically grounded image designed to help trigger recognition and generate investigative leads that might otherwise be missed.

LYNN: The oldest DNA sample Parabon uploaded for analysis was 200 years old. Were you successful, and if so, what did that feel like?

CECE: The 200-year-old sample was a museum sample from an unmarked grave that was processed by the Armed Forces DNA Identification Laboratory (AFDIL). The DNA was in much worse condition than a typical forensic sample, and we worked with AFDIL on the laboratory methods and the bioinformatics to get as much information as possible out of it. We were eventually able to get a profile that could be used, which appeared to confirm a suspected identity for the individual. We’ve also explored even older samples as proof-of-concept research, including 2,500-year-old Egyptian mummies.

LYNN: How often are you disappointed in your results?

ELLEN: Not every case produces clear answers, but that’s simply the reality of working with forensic DNA. Some samples are too degraded, too limited in quantity, or too complex to yield useful information. We do everything we can to give each case the best possible chance of success, but on rare occasions, the data is just too poor to make any inferences. That is incredibly frustrating, and I will spend hours applying every technique I have, or even inventing new ones, trying to get useful information out of the data because I know how much hope investigators and families are placing in the outcome.

CECE: The ability to successfully apply IGG to an unsolved case is dependent on who has contributed their DNA to the databases we are allowed to utilize for law enforcement research. While some of that is luck, the population group of the unknown Subject is the most significant determining factor. Subjects who have more recent immigrant ancestors are much more difficult to identify through IGG because populations outside the United States have far less representation in the consumer DNA databases, especially the relatively small ones we are able to access for these purposes.

Another significant challenge we encounter in our research is when the person we are trying to identify is adopted or has an unknown/misattributed father (which happens surprisingly frequently). In those cases, we are able to determine the family tree of the Subject and their close family members, but they are “missing” and not associated with the key family socially or in public records. 

The good news about IGG is that it is a dynamic tool in that new people upload their DNA to the databases every day and just one new “match” can change the status of a case from unsolvable to solvable, and it often does.

I have been involved in a case of a John Doe from Oregon and another from Australia where we positively identified both of these men’s biological parents and their siblings and still do not know their legal name and identity. These men were, obviously, placed for adoption and raised by families other than their biological ones and, despite extensive efforts, we have been unable to discover the identity and name they lived under during their lifetimes. Very frustrating and disappointing for all involved!

LYNN: Parabon has helped to solve 370 cases since 2018. What does it feel like when you know you’ve hit pay dirt?

ELLEN: Whenever I look at the results of a DNA phenotyping analysis, there’s this exciting realization that I’m one of the only people in the world, other than the perpetrator himself, who knows this information. Especially if the results are something other than what the detectives expected, I imagine them diving back into their case file to see what new progress they can make now that they have this new lead. What’s most meaningful is knowing that the science is doing what it’s meant to do: helping investigators move forward and, in many cases, giving families long-awaited resolution.

CECE: It is very meaningful and satisfying when the genetic genealogy research all comes together and points us in the right direction, knowing that we will be helping investigators to provide answers for the survivors and families, resolution and, hopefully, justice. I feel incredibly fortunate to be able to do this type of extremely fulfilling work every day.

LYNN: What was the highest profile case in which Parabon played a pivotal role in identifying the murderer?

ELLE: Every case is high-profile to someone. However, there are a number of notable firsts that we’ve been involved in, such as the first conviction in a genetic genealogy case (William Earl Talbott II for the 1987 Jay Cook and Tanya Van Cuylenborg homicide), the first exoneration using genetic genealogy (Christopher Tapp exonerated, Brian Leigh Dripps convicted of the 1996 Angie Dodge homicide), and the first conviction of an identical twin using DNA (Russell Marubbio for a 1987 rape).

CECE: The April Tinsley case out of Fort Wayne was very well known, as was the Angie Dodge case from Idaho Falls, ID. I also identified a serial killer in 2018 who has now been found to be responsible for the very high-profile Yogurt Shop Murders from Austin, TX. Although we were not directly involved in working with investigators in that case, Parabon’s earlier genetic genealogy analysis to identify the suspect in several crimes in other jurisdictions laid the foundation for his identification in that case.

Prosecutions stemming from Parabon’s cases have set precedent for how investigative genetic genealogy is treated in court through rulings in dozens of cases. We have also been involved in groundbreaking cases in the forensic realm involving rootless hair analysis and sequencing twin DNA to identify unique mutations. Parabon’s Dr. Janet Cady’s analysis and testimony for a recent Virginia criminal case resulted in the court accepting evidence from full genome sequencing for identification purposes for the first time, setting an important precedent.

LYNN: If people want more information on the subject (especially writers), where would they go?

ELLEN: Parabon’s website is a good starting point, as it includes case summaries and explanations of the science. For writers in particular, Parabon’s publicly released case pages (“Featured Cases”) provide rare insight into how these tools are actually used in real investigations. We post the results of every DNA phenotyping case that law enforcement agencies have released publicly. When the person is eventually identified, we add their photo, so you can see for yourself just how accurate our predictions are.

LYNN: Many, many thanks to both Dr. Greytak and CeCe Moore for not only participating in this interview but for the important work they do. Please visit their website to learn more.