บทถอดความฉบับเต็มของ Ep. 6 Science SNiP: Forensic DNA Phenotyping, Predicting Human Appearance

that anyone who's had a DNA test come back with a non paternal event should take some solace in that it is actually a lot more common of an occurrence than, than people, think.  Now also in 2012, a husband and wife were gunned down in their own home. They woke to the screams of their 19 year old daughter who was in her bedroom on the second floor. Apparently there was a male intruder there that was threatening her with a knife. And her parents were at the bottom of the stairs coming to render her aid and the intruder rushed down the stairs and he changed a knife for a gun, and he shot both of the parents. When he was leaving he must have injured himself. They suspect that he cut himself with a knife when he was rushing down the stairs. So there was blood recovered at the crime scene. Now they did standard DNA profiling with that meaning they took it, they gave it to the crime laboratory. The crime laboratory did the standard DNA analysis. They generated a DNA profile. So that would have been using an STR or short tandem repeat technology. And they entered that DNA profile into the CODIS database and searched it, but they did not get any hits or what we would call an investigative lead. Now I talked about the CODIS database and how those work in episode five. So if you haven't listened to that one, you want to go back and give it a listen. Now additional DNA testing was done at the University of North Texas laboratory. They looked at, some familial markers, so they did some familial testing, they did some Y STR testing and they found, one potential suspect, but they were actually, a direct comparison between their DNA profile and the blood that was left at the scene, wasn't a match. So that, that suspect was excluded. The investigator in this case then learned about DNA phenotyping, which was a new service that was being offered by a private laboratory. And so they asked that laboratory, if they would do it, they call it a Snapshot DNA phenotyping test. And again, in the show notes I'm going to link to Parabon NanoLabs website, where they have information about this case you can read about it in full detail . So it was a combination of standard DNA testing, just the routine work that a crime laboratory does in conjunction with familial searching and Y chromosome testing, to see if somebody is of the same male lineage and then the DNA phenotyping, where they were able to have some information about what the possible hair color, eye color, whether or not somebody had freckles, what their skin tone was. Some information about the possible face morphology as well. And then of course, they also looked at some ancestry markers as well. Now we're going to deal with ancestry in a separate episode when we're talking about specifically more about biogeographical ancestry. But it was all of those things in conjunction where they realized that to the results of the familial search in conjunction with the phenotype and ancestry results told them that they might be looking for a male relative of that person of interest. And he did actually have a brother and they did a routine DNA analysis of the brother and found that he was a match.  The other big reason why this technology is very helpful is in cases of missing persons. In 2020, the remains of Christy Lynn Floyd, who was 16 years old at the time of her disappearance, and I think it was 1986, were finally identified. Parabon labs used their DNA phenotyping snapshot analysis on the DNA extract from another laboratory that had done initial DNA testing on her skeletal remains and they were able to come up with a picture of what the woman might've looked like when she was 25: hair eyes, skin, whether or not she had freckles. Again, they looked at a couple ancestry markers, and they were also able to possibly predict her face shape and her body mass as well. In this situation, it wasn't just forensic DNA phenotyping, and a little bit of ancestry markers. They employed forensic genetic genealogy, which again, we're going to talk about in more detail in a separate episode, but with that information, they were able to search the publicly available genealogy databases and find a possible third cousin. And that's how investigators were able to identify her remains.  You can see the utility of forensic DNA phenotyping. The idea that you can predict human externally visible traits from someone's DNA. And that's why it's become a fast growing subfield within forensic genetics. It can provide a lot of intelligence information from traces of DNA and help police focus investigations in search of unknown perpetrators or to identify unknown human remains. And of course there's also the anthropological aspect of it as well.  A little DNA background informationAngela: Briefly we're going to interrupt this discussion on phenotyping to make sure that we're all on the same page when it comes to DNA testing and understanding human DNA since I know that our listeners come from a variety of backgrounds. As humans, we have 23 pairs of chromosomes. 22 of them are called the autosomes. And the 23rd pair are our sex chromosomes, X and Y. And they're literally called X and Y because they actually look like an X and a Y when you look at a karyotype, which is what you do to visualize the chromosomes. Now your chromosomes are just your DNA molecule, all tightly packed and wound up. And when you need access to a particular part of your DNA to make something that your body needs it just selectively unwinds a little bit, makes a copy of whatever it needs, whatever instructions that it needs to get, and then twists it all back up and keeps it all packed up nice and tightly inside your cells. Your DNA is present in the nucleus or the brain of all of the cells of your body with the exception of mature red blood cells, because they don't actually have a nucleus. You also have in the cell of your body, you have mitochondria in the now they're just out in the main part of the cell.

They're separate from the nucleus. They have their own genetic component. Now their function in the body is to produce energy so there's quite a lot of them in each individual cell. Your mitochondria are inherited from your mother's side. And when we're talking about your autosomal DNA, half of it, you got from your mom and half of it, you got from your dad. And obviously if you have a Y chromosome, you had to have inherited that from your father. So you are a unique combination of each of your parents, cause what you inherit from each of them is random. Your siblings will be similar to you, but they are also a unique combination of the genetic material from both of your parents. So you can see how genetic genealogy is going to work and how we can tell familial patterns in people's DNA.  The standard method that is employed in crime laboratories for DNA testing is STR DNA profiling. STRs are short tandem repeats. They are repetitive stretches of DNA that are spread throughout your genome. In forensics we look anywhere between 15 to about 26 different locations Depends on what country you're in, what you're required to test for, but there's overlap between different countries and what locations we look at. When we're talking about a specific location in the DNA, in your DNA, it's called a locus. Plural would be "loci" or loci, depending on where you're from. For forensic DNA profiling those locations were specifically chosen because they were not related to any heritable traits. People were very concerned about genetic privacy at the time. And so they were selected to be in non-coding regions, right? Nowhere, that's going to say, oh, if you have this, you, therefore you must have this disease. So you can't be denied health insurance or medical care for preexisting health conditions. That was, it was a big topic at the time.  STR profiling or DNA profiling is different than what we look at when we're talking about phenotype and looking at those traits. STRs, like I said, they're short tandem repeats, what we're actually looking at are length polymorphisms. So if you think about your DNA as a train and everyone has an engine and everyone has a caboose at one of these locations that we're looking at, but the number of box cars inbetween the engine and the caboose are different for each person. So that's how you can tell the difference between people except for identical twins with this type of testing. Now there are limitations. So in a criminal investigation, you've got your DNA profile. You don't have a suspect. You don't have a person of interest to do a direct comparison to the evidential profile. So you put it into a database, you search it, but you don't always get hits. Now you can do familial searching, but it's really only going to help you find close relatives. It's not going to be good for more distant relatives. And that's of course, one of the powers of genetic genealogy is that you're combining the DNA technology with traditional genealogical investigations. PhenotypingAngela: Now, in contrast to the DNA profiling methodologies, looking at those STRs, when we start to talk about a phenotype, we are actually looking at the underlying genetic traits. So that previous concern everyone had about genetic privacy is back up in the forefront again.  It's important to note that I said genes, plural, because when you're talking about a phenotype, it is generally more than one gene, sometimes thousands along with external factors that contribute to a final phenotype. In the early 2000s, when research was being done into phenotypic associations to the underlying genetic information, that was primarily focused on disease research, right? So cystic fibrosis for example, was very heavily studied and a lot of different mutations were identified. And so it's helped the treatment of that illness. But the same technology that you were employing for disease research is what you would need to do for externally visible characteristics, which was what we're interested in,in forensic science to assist in the identification of individuals.  Pigmentation is the best understood of the externally visible characteristics, specifically eye color, and hair color, and to a lesser extent, skin pigmentation as well. Other types of traits like the facial morphologies, body mass index, those ones are not as well understood, which was why some of the information that Parabon NanoLabs is providing in their phenotype reports has been criticized quite a bit because they haven't actually released the underlying validation studies that support the conclusions. So you may be wondering how is it that, you know,a particular gene is, or genes are associated with a particular trait or sometimes sets of traits because sometimes they are all inherited together. The main way that we know is through wonderful advances in technology at first like the DNA microarrays and more recently massively parallel sequencing has allowed a tremendous amount data to be collected from thousands of different people. The more people that you have that can participate in the study, the more robust your results will be. So you get a large number of people that have the trait of interest and a large number of people that do not have that trait. For example, if you want to see what might be the cause of someone having blue eyes, you get a group of people that all have blue eyes and you get a group of people that had have not blue eyes and you sequence their DNA and you compare them and you see what's different and you analyze that data. And then once you think you figured out which SNPs it is that are causing that particular color or that particular pigmentation, then you go and you can test it and see. So you get another group of say 200 people and you use it and you test to see, is it consistent? Can you correctly predict blue eyes and not blue eyes? Now I may have made this sound like a simple process, but it's not. It's a lot of data to

collect and a lot of organization and a lot of technical know-how and a lot of money. It does to take quite a bit of funds to be able to conduct these types of studies. 2007 was the first time we saw a publication on eye pigmentation in 2009 was when we had a comprehensive study on eye color. The work was published in Current Biology and it is an open access article so you can read it at your leisure. I will link it in the show notes. So it was done by the, by Liu et al. at the Erasmus University in the Netherlands. They are very much leaders and the phenotyping arena, they collaborate quite a bit with Dr. Susan Walsh's group and Indiana University - Purdue University, Indianapolis Science Center. I think I got that correct. Based on previous findings together with some of the SNP prediction rankings that were observed by Liu et al in the paper that I just mentioned, her group put together the first DNA based eye prediction system for forensic usage. And that was published in about 2010 to 2011. And this system was called the Irisplex. It's a sensitive. multiplex , it's a word meaning we're testing for more than one location at a time. And it was genotyping of the six most eye color predicting SNPs. One of the other great things that they did was make this available in a format that was easy to implement for a forensic laboratories. In addition to creating the IrisPlex system, they also implemented the predictive model and they made that available to users of the Irisplex. It's interactive and it's easy to use. You just go to a website and you type in the SNP data that you have, and it will give you the categorical predictions. So it's very good with, blue eyes and brown eyes. And then you also have what's called intermediate eye color. So those are the categories that you have. So somebody has Hazel eyes, it's going to be an intermediate category. And it's better at blue and brown. And when it's been tested by other laboratories, they've actually of course, seen that the model holds up, even when they're looking at geographical areas other than where the European samples that were used in the studies came from. So even though, they built the system based on studies conducted on Europeans, they found that it did correctly predict from neighboring regions as well and distant regions like east Asian, Africa, Oceana, and Native Americans. Obviously in some of those locations it's only going to predict brown eyes, but that's consistent with what we observe anyway. So we know that, regardless of biogeographical origin, certain predictions are likely to hold true. They further tested it through a collaborative study with the International Society for Forensic Genetics, the European DNA Profiling Group that's part of the ISFG They conducted a multi center exercise involving over 20 laboratories with a wide range of experience, including novices. And they all found that this Irisplex was easy to implement and highly reliable. And so it says something that's highly reliable in the hands of people who are not that familiar with it.  Since then the research has expanded and the application has expanded. They continue to do more research into hair color. As you would imagine, determining genes associated with red hair were much simpler, or it was pretty simple. And, so that was, that's easy to predict. Some of the other types of characteristics are a bit more complicated. Hair texture, so if you're talking, whether it's straight or curly or wavy, these are a little bit more complex and, but they can also have significant impact when you're looking at it from an investigative standpoint. Knowing that somebody has curly hair or straight hair is good for differentiation. Face morphology is a bit more difficult. There's a lot of studies that are going on. They're looking at some genes that are known to be associated with cranial facial differences, but they're finding so far that the genetic variants don't have as much of an impact on the final phenotype as you would need, so there's research is still ongoing and likely it's probably surpassed, what I've found in the recent articles. Susan Walsh's group, working with her same collaborators, they've expanded Irisplex first to include hair, so it was the HIrisPlex or "high ris plex" . I don't know if that's how they wanted to say it or not. Now in 2018, they also added skin pigmentation to that. So they've been working to make it accessible for forensic laboratories to have this as a tool that's available to them on a much more regular basis. So there won't be a need to send it out to specialist laboratories because it is all run on the same type of equipment that forensic laboratories already use for their forensic DNA profiling. NowAngela: So where are we with phenotyping now? How reliable is it? On the website for Dr. Walsh's laboratory they state, " currently this tool provides categorical eye, hair, and skin color predictions with approximately 80% correct prediction output per trait." Is 80% enough? I think that it really depends on the circumstances of the case that you're investigating. Dr. Keyser pointed out in one of his articles that predicting the externally visible characteristics from the biological material that's left behind for whatever reason, whether it's a criminal matter, or if it's a missing persons or it's a disaster victim identification, or you're doing anthropological research, it's a "biological witness". It could actually be more reliable than other information that you might be able to obtain through more traditional investigatory methods. For example, would you take the biological witness and use that to possibly identify a particular possible perpetrator? Or would you rely on a eye witness? We know that there has been study after study that said that eye witness testimony isn't reliable, that people's memories are not always accurate. So is 80% on forensic DNA phenotyping, more accurate than eye witness? But of course,

then you run the risk of what happens if that biological material that was left at the crime scene had absolutely nothing to do with the crime that was committed? So you go and you do DNA phenotyping. You give a prediction on what the possible appearance of the person whose DNA was found might look like,and investigators start investigating that. They get focused on that particular appearance and that's all they're looking for and it's a false lead. So it definitely is a tool that needs to be used with caution. And of course people have the ability to change their appearance. How many women dye their hair on a regular basis? So maybe they're blonde, but they've been dyeing their hair routinely red for 10 or 15 years, if you're looking for a blonde and someone's red head, you probably not going to find them. And then of course, you've got other physical characteristics that your genetic information that's predicting your phenotype isn't going to take into account a lot of the environmental pressures that are also a part of determining what your phenotype is. You know, even if look at the facial morphology, right? So we've predicted that somebody whose face has a particular shape to it, their eyes are a certain distance apart. But what if they were in a disfiguring car accident? So that was an environmental thing that happened to them that changed the externally visible characteristics from what the genetic information would have predicted. That is a bit of an extreme example. And if somebody is very committed to altering their appearance so that this type of tool can't be used to find them, then they have to be very committed to it,plastic surgery that type of thing. If they wanted to go to that much of an extreme. Yes, people can change their appearance. It could make this less effective of a tool. But if you're going to start relying on DNA phenotyping, it's not your first step. It's the step that you take after your standard process of develop a DNA profile and search a DNA database, do some more investigations.And it wouldn't be the only tool that's available either because we have also investigative or forensic genetic genealogy that you can follow from a DNA profile that's found at the scene. You can also look at the biogeographical ancestry predictions and try to use all of these things, your standard DNA profiling in conjunction with all of these additional tools to assist the investigation. In the examples that I gave at the beginning of this episode, that's what was done. It wasn't just one technology that was used. It was multiple technologies combined with detective work. The investigators still has a lot of work to do but the science can help and the science can inform... with cautions, right? It's important that the investigators understand the limitations of the testing and understand that it's entirely possible that the genetic material that was found at the crime scene is there independent of the commission of the crime. So if you've got that information, don't be looking at it in a vacuum. You need to make sure that it is actually valuable and probative to that investigation. And if you're not sure you need to make sure you don't dismiss other avenues of investigation in favor of the phenotyping.  Now, in preparation for the episode, where I interviewed Dr. Matthias Weinroth and Dr. Carol McCartney, from the University of North Umbria about some of the implications of these new technologies and where they might lead and some of the complications from a societal perspective, what kind of effects do these technologies have on society? And I'm not putting on a tinfoil hat or anything, but I do like science fiction movies and if you think about all of this information that we're gathering from people's, from their gene genomes and we're collecting all of this data. Who has possession of this data? Yes. Some of it is publicly available, but not all of it. Some of it's housed by private laboratories. Does the government or different states, maintaining and data that's not publicly available? Who has control over it ultimately? And do we need to be worried about the dystopian future, like in Gattaca, right? Where there's forms of genetic discrimination. I'd be interested in hearing what your thoughts are on the subject. Please email me at hello@secretsfromthecrimelab.com or you can just go to our Facebook page and send me a message on there. And I'd like to know your thoughts on this episode as well. So please comment. And let me know what you think. Well, none of us really expected I'd manage this topic in under 15 minutes, but I hope you learned something new today and, have a great day.