Sano Genetics 067 - Anna Lewis ------------Dr Patrick Short: [00:00:00] Hi everyone and welcome to the genetics podcast. I'm here today with Dr. Anna Lewis who's a researcher at Harvard focused on ethics as it relates to the return of predictive genetic test results and gene editing. We're going to cover a lot in this episode, including those two. If you've listened to the podcast before you'll know that I love these kinds of ethics discussions, because technology developments. Simply don't happen in a vacuum and in particular today we're facing some very challenging, ethical dilemmas. There are many, many more people getting genetically tested through healthcare system. Through research biobanks, gene editing is now a reality. All of these technological changes are starting to come together to produce some really profound, ethical questions. So Anna, welcome to the podcast and thanks so much for taking the time to join. ------Dr Anna Lewis: Thank you for having me. ------------Dr Patrick Short: So you did your PhD in computational biology. You worked as a computational biologist and then a product manager for some time. And now you're an ethics research. Something that really jumped out to me from your bio is that I think you've said before that quote, and I'll quote you day-to-day conversation with ethical issues led [00:01:00] you to wanting to pursue ethics full-time I'd love to start there. What were you doing when you started thinking going to ethics research full time. Was there a single event that made you think you want to do this? Or, or what was it? ------Dr Anna Lewis: Well, I'll name two events. One was when a academic collaborator of mine, when I was in industry sort of bounced up to me and said, Hannah, I've worked out how to put a score on a genome. I was like, oh really? Yeah. That's all sorts of ways things could go from here with that kind of, um, that kind of idea. And the other one was the first time I wrote a consent form for a genetics research study. And here in the states, you have to include this very standard paragraph of text, which explains that although there's legislation that protects individuals from being discriminated against in the context of health insurance, it does not cover other forms of insurance, including life insurance. And it's just like, Hey, [00:02:00] this is a thing. No, no further information. And it just, that just felt to me that just felt really bad to be putting that in front of individuals with no further information and it seemed completely non-acceptable. And so I think I've, yeah, I've got interested in this world is called the LC world. It stands for the ethical, legal and social implications of genomics. Genomics is the original LC. There's now just starting to be LC of these other new technological areas like MLII, for example, and I view this, this field as the contextual wraparound to the, to the technology. It's lots of pieces that aren't right at the core. It's about how that . Technology is going to impact individuals and society. ------Dr Patrick Short: Absolutely. What has been the biggest changes since the very first time somebody said to you hey I think I could put a score on a genome. What, what, what score was that? Do you remember? And what's changed since then. ------Dr Anna Lewis: On analysis of rare variants and this idea of, of genomic [00:03:00] burden, um, which hasn't really taken off yet. I still think that might take off, but of course, the big thing, which has hit CenterStage a polygenic risk scores, which are based around common variance and an implication of it being the study of common variance is that it's accessible, justified at this much cheaper technology of genotype chips, as you know, and I'm sure you've discussed with many other people on this podcast, there's this is now a huge research endeavor, the publication of new polygenic risk scores and the underlying technology for creating them as basically the same for any trait. So if you've got yourself a phenotype and you've got yourself some genotype information, it's very, very easy to calculate one of these polygenic risk scores. It's not to say it's going to be a good such score, and we can talk about that. But then you've got this number that you can attach to an individual and in a Seminole paper from 2018, some researchers here at Harvard were like, uh, yeah, like the time to think about clinical application is now [00:04:00] because we can start to detect effects at the same sort of effect size as some of them monogenic variation that we return results for. And I've been involved in a couple of big studies, which are returning results, polygenic risk score results to individuals, and there's all sorts of Elsie things which come up in that context. Yeah. Yeah. ------Dr Patrick Short: What are some of the big issues that have come up or that you all are thinking about? And in particular, I think it'd be interesting. I know in a couple of your papers, you specifically contrast the framework for monogenic risk scores, you know, monogenic risk. They're not really scores. They're kind of scores. They're often zero or one, but, uh, with polygenic risk scores that really run a spectrum from the extreme end, where it's like monogenic to the other extreme end, where it's protective to. I'd love to hear what some of the big hitting issues that you think about a lot are. ------Dr Anna Lewis: So a polygenic risk score is just a number it's uninterpretable as it is. And so you have to [00:05:00] put that in context, somehow for the person receiving the information. And the simplest thing you can do is return a percentile. So you can say you are at the 98th percentile for coronary artery disease. You know, that sounds like a really high number. Maybe you should be concerned about that. But the problem with returning information like that is you have no idea how predictive the underlying score is. It could be that you're at the 98 percentile, but that's only an increased risk of some small amount. So maybe you want to frame it in some other way. And the next easiest thing to do is to frame it as a relative risk. So you might say, you know, you're at three times the risk compared to the general population. So. But notoriously, there are all sorts of issues with returning relative risk to individuals. It's very easy for risk to end up exaggerated. If you use that kind of language and then sort of better yet might be incorporating into absolute risk models where you could say, you know, combining your polygenic risk score, where these [00:06:00] other risk factors, your ten-year risk of a heart attack. X percent. So you've got to think about these different ways that you could integrate polygenic risk scores. And you've also got to, like, there are also some decisions about, are you just going to return right at the high end of the polygenic risk score range? Or are you going to make it a like continuous number that you include? So there's all sorts of things which are questions about how do you actually return these interviews? And the other really big one with polygenic risk scores is the impact of the study population and the validation, um, population. So no matter which way you cut individuals up into categories, whether you're somehow forming some notion of continental genetic ancestry, whether you're looking at self identified, race and or ethnicity, what you're going to find is when you can then take your probably genetic risk score and look at how it performs in one of these populations. You're going to see very different performance, depending on these different [00:07:00] ways that you cut up. So most infamously the scores before much worse in individuals of African ancestry than individuals of European ancestry. These are all population level summary statistics, I should say. And that's a problem when we think about clinical application, because the quality of the data that goes back to those with African ancestry is just worse. And it's not clear how to deal with that. You could, you could take the line that, you know, it's worse than we're just not going to return as a tool. Myriad, have a score for breast cancer, polygenic risk score for breast cancer. That's currently only available to individuals or to women of European or Ashkenazi Jewish ancestry. They're working on updating that, but that's still the current status right now. So that's one option you can take. Another option you can take is attempt to show that it performs less well on the report somehow. That also has issues. Yeah. And then a big issue, which I'm sure you've seen as well. I know you think about [00:08:00] as it's just very, very hard to convey any kind of probabilistic information to individuals. There were all sorts of gotchas and just, it's not just, um, people on the streets, physicians also struggle even with percentiles offense, which is kind of shocking. ------Dr Patrick Short: Absolutely. Yeah, it is. Um, well, and people aren't trained in genetics necessarily, right? Uh, the field has changed so quickly that medical school may not have had these concepts for most people practicing. Just, you know, even if 10 years ago you were in medical school, a lot has changed. ------Dr Anna Lewis: Yeah. That's, that's definitely true. And I feel like a lot of scholarship ends up by concluding that we need massive amounts of new education. You might say there's a, there's a flip side. Some people, yeah. Would argue, well, look, the more emphasis we put on this, the more exceptionalists were being about genetic information and that's not a good thing. Like that's a very active debate, the extent to which we [00:09:00] should treat genetic information very separately and put higher bars to its incorporation into clinical practice. Yeah. And I'd love to actually unpick that debate because that is a, it's an important one. What is the framing in your mind there? And whereas genetics, non exceptional, and it should be treated just like every other experimental thing in the healthcare system. And where is it except. Well, there was a lovely article out last year. I think arguing for genetic contextualism, which I think is appropriate. Typically the debate has been coming from the LC world, actually about genetics being exceptional for the sort of standard reasons implications for family members. The fact it doesn't change. Much closer ties to identity and these sorts of things. And on the flip side, there have been not so many voices who want to emphasize its continuity with other information. And, you know, there's lots of other information, which is also forward looking like a lot of clinical [00:10:00] measurements that we get also can be used to predict your health status later in life. We can say probabilistically quite a lot about your family based, not on your genetics, but on your, your phenotype and on your lifestyle. Just because those are also correlated to family members. But then yeah, there might be specific cases in which it is more appropriate to be more exceptionist about genetic information, anything where there's a chance, for example, that the sorts of genetic genealogy pieces are going to come into play. Al forensics are gonna come into play. You know, genetics I think is still real standout in those areas. Yeah. ------Dr Patrick Short: I'd, I'd love to maybe dive into a specific example and learn how you break it down from an ethics perspective. So if we took polygenic score for coronary artery disease, as an example, and we're thinking through, from an ethics framework, how should genetic test results be returned to an individual? For instance, I mean, I can, we can be more specific in order to break it down, but let's suppose somebody was enrolled in a research [00:11:00] biobank very general. So they enrolled to donate their genetic data to be used in research, not specifically for coronary artery disease, but the polygenic risk score is discovered that says we can find a small fraction of people, one or 2% that we know based on their genetics and a few other risks they're at really high, really high risk. How do you think about whether. The people running that biobank should, should get in touch with the 1% of people in the biobank that carry that risk or not. How would you frame that question from an ethics perspective? ------Dr Anna Lewis: Let's say, first of all, a lot of this is about how you set up the biobank in the first place. So the lion's share of the scholarship is on. Okay. You're about to set up a biobank. How did you think about this? It's not about you've got this biobank is already set up and now you're having to back think and untied a lot of messes made. But if, if, if you, if you take the prospective case, so you're, you're setting up a biobank and you're thinking about, basically what you owe your participants. [00:12:00] Yeah. I think that's a great way to put it. Yeah. So that's the first thing to say is this depends radically on which country you're in, what the legal framework might actually be the legal policies and regulations that are in place are surprisingly strong. I think in terms of saying that if you come across. Clinically actionable results, then you should return them. What we see though is with genetic data is a real mess between this idea of incidental findings and secondary findings. And these are separated clearly in the literature, but in my opinion, not clearly in practice. So incidental findings are the things that some researcher looking at that biobank data stumbles upon. They weren't actively looking for them, but they stumbled upon it. And you can easily imagine how this might be. Somebody opens like a genome browser and they're looking at BRCA and they see a really obvious pathogenic variant, like that would count as clinically actionable and should be returned in sort of the framework I was laying out. So in that case, they ought to return that bracket area. [00:13:00] But then again, they might've set themselves up not to do what's called the secondary findings, which is when you actively go and look for some set of, um, some set of conditions. Some have some set of variants that are clinically actionable. So I think this is like a dodgy division, because in practice, it's so easy to look for certain variants. Like if you're a variant, that's very well annotated as pathogenic and some database going and getting that piece of information is like a lookup function, right? ------Dr Patrick Short: Yeah. So it's often a routine part of analysis, right? Where you may be exclude people from an analysis that, you know, have a disease causing genetic variance. Right? ------Dr Anna Lewis: Right. Absolutely. Absolutely. So I think our sort of LSE framework for thinking about these things needs, uh, needs some updating, but it's because of the prospect of this kind of finding whether you're going to look for it or stumble upon it, that these, these legal frameworks, exist. And then there's also a whole lot of separate overlapping and [00:14:00] mutually reinforcing ethical arguments for why you should return these findings. Those include the rights of participants, right. To no right to access, etcetera. They also include duties of the researchers, including duty of ancillary care, duty to warn duty, to rescue. And then there can be some just straightforward appeals to what we call beneficence. So like as a . Researcher, you can do good by pointing out to somebody. Hey, do you use the BRCA example? You've got this pathogenic BRCA variant um, you should think about having much more screening than you're currently having. So, so that's kind of how that framework shapes up. I'll mention one other thing, which, and I think that, that you guys are part of this trend. Which is for more research, participant engagement with research, not just treating participants as sort of passive donators of data. That's sort of part of the process. That's another worldwide trend, which is [00:15:00] leading to the importance of return of individual research results. So that's some of the context, what should you then do with that? Well, the most important thing is that when you set up your bio bank, you should think it through carefully, you should think through what your protocol is going to be. That then of course has to get reflected in the informed consent document. You definitely need to budget to do this because it's not cheap. You need to sort of share your tools and processes to enable others to do a similar thing and bring down those costs. Barriers for research studies to do this, this sort of thing. ------Dr Patrick Short: So now moving from a top 1% highly actionable, or the BRCA case into the messy middle of things. How, how, how does it change if at all, to say, uh, returning information about 10 polygenic scores covering BRCA to coronary artery disease to Alzheimer's where I may be in the 50th percentile. In [00:16:00] one and the 95th and another. And the second, how does that change? And also, how does the participant kind of choice factor into all of this? Because this is almost one of the most interesting parts of it to me, because I think there's one camp that argues basically the participant choice is paramount. And if I give the information and say, here's what I'm going to tell you, if you agree, and here's exactly what it is. And isn't, if you agree, you get the info. If you disagree, you don't, or the other camp that says actually people, you know, Always be able to handle that truth. And even if I try to explain it, people may not understand, or I may not explain it well enough. So therefore there's some things that we just shouldn't ask them. I'm really interested when it gets into the, the messy middle part of this. How, how things change and how you think about it. Sure. ------Dr Anna Lewis: Well, let me, let me carve that into two. And the second part will be models for consent and in answer to the first part, like, okay. That you could potentially return. All these polygenic scores for a lot of different conditions. And you have to think about under what conditions to return. I [00:17:00] mentioned, um, if there is with a big, a big project over here in the U S it's called emerge for. It's a big NIH funded project. It's going to involve the return to about 25,000 Americans have about 10 polygenic risk scores. So it's, we're still in the planning stages. So in year one of this project and end of year one, and it's exactly those questions, which we've had to decide upon. And the thing that's really motivating the return of polygenic risk scores and that's returning and motivating return of results in research studies is this idea of clinical actionability. So one approach that you can take that I think is sensible and that's been taken by emerge is you can say, well, how do we decide what's clinically actionable? And there, at least for the time being genetics can look to how other risk information is incorporated into clinical decisions in certain phenotypes. So to take the example of colorectal cancer, there are preexisting guidelines [00:18:00] that say, you know, if you have a strong family history to find them whatever way for colorectal cancer in your family, then your screening intensity goes up in such and such. So then what you can do is you can look at well, what is the risk level associated with having a strong family history? It typically will be some kind of relative risk information that you can get. And then you can say, okay, well, we'll set the same threshold for return of apologetic risk score to that preexisting guideline. So that's an approach you can take. And for a lot, for a lot of conditions that are such similar. Preexisting guidelines. And of course the conditions we're focused on are the ones for which some kind of preventative measures are possible. Otherwise, why would you focus on them? And if you're going to take that approach, then you're not necessarily going to return any information about the middle of the spectrum. You're just going to focus on the people who are above that clinical actionability [00:19:00] threshold. This is not an easy choice. Mind you? Lots of people might. Hey, Hey. Hey, I want to know if I'm, so here it might depend on the, on the context and in your biobank example, that'll depend sensitively on how you've set it up in the first place and what you have told participants. You will return to them. If you have phrased that very strongly around clinical actionability. Then you should probably think about it in a similar way to how I just outlined. If instead you phrased it, like we're going to give you all this cutting edge cool stuff, which really isn't reliable at the moment, but you might be interested in looking in and like massive, like buyer beware, not buyer beware, but like big pinch of salt needed ------Dr Patrick Short: Yeah. Advisory that it may not be totally polished. ------Dr Anna Lewis: Yeah. To now segway into the consent conversation, even if they've sort of said, yeah, give me everything. There are some [00:20:00] things that could potentially have large impacts on people. So we have all sorts of frameworks for consent. Like regular consent is, is distinguished from broad consent, which is where you're just like, Hey, take my data and basically do anything. It's very, open-ended that type of consent is sort of increasingly sought by research. You can do granular consent where you ask for participants views or wishes for specific types of information. Like one, one typical thing would be like, what about something that's not treatable? So your outsider's example, you know, that's not, we tend to say it's actionable, but it's not really clinically actually. It's actionable in that you could go and buy life insurance, or you could go make plans for, for your care later in life and that type of thing. But unfortunately it remains the case that we don't have good treatments. There's a reasoning behind something like granular consent is that these are questions on which reasonable people can [00:21:00] sensibly differ. You might want to know your outsider's risk and I might not. And both of those are perfectly legitimate and understandable attitudes. And that's why it's important. Ask about. Another model for consent. It allows for dates of preferences. Like you might decide when you sign up for this buyer bank that you want everything. And then for whatever reason you might decide, actually, no, you don't, you don't want to know everything. So having the ability to go back in and say, actually, no, no, no. Um, I said yes, but I've changed my mind. It's not a straightforward process. This consent business. And I think one issue that we're having is that the answer to any sort of tricky question, when it comes to research participants is be transparent, put it in the consent form. And then you've got consent forms, which are pages and pages and pages and pages long. And do we think that participants read them all? Not so sure. And the cases in which [00:22:00] people have studied, do people actually understand what's in the consent form? The results do not look good. So, so you can solve some of these by always having a person in the loop, be it a counselor, a genetic counselor, or somebody else. All of this adds to the expense of doing genetic research. And I'll just mention one more thing and then I'll shut up. You mentioned some others who might have slightly contrarian views on these topics. And that line of thinking would say something like if we put all the emphasis on consent, then. We're inevitably going to end up with, like, that's not very compatible with an equity angle because it's going to select for people who have a lot of time and a lot of mental energy to put towards thinking through these things. So there's some kind of trade off there. And if we're talking about a situation where there really is potential [00:23:00] benefit to participants, then you're going to have certain groups of people. Systematically losing out. So there's a lot of talk now, rightly so about who gets recruited into research studies? We have big over-representation of educated white folk basically, but there are lots of good reasons why other folk do not participate in research. It's not just. They haven't been asked to participate. Yeah. ------Dr Patrick Short: Absolutely. We'll revisit that aspect of diversity and inclusion because you mentioned it earlier with polygeneic scores and how they don't work the same for everyone. And I think the numbers of people of European ancestry and genetic studies has more or less remained unchanged at 85% plus over the last. 10 years. And it's a huge issue. What do you see particular in terms of getting these tests into clinical practice? What do you see as the solution? You talked a little bit about it earlier. If it doesn't seem good to say no one gets it until it's equal for everyone, because then, [00:24:00] you know, even if it's doing something that it may be is worth them putting it out there, but it's also not good to say it's significantly better. for One group and another and not have any incentive to change it. So what role does policy have to play and think the role science has to play? It's pretty clear to me, figure out how to make these datasets more representative, but actually doing that in incentives. Don't feel like they're in place right now for anyone to do that, except to just because we feel like we should and must, but at the end of the day, it's not, it's not being required by the FDA, for example. ------Dr Anna Lewis: So I think so what we've been seeing recently, in fact, it's not recent, there's a very, very long history of this, but we're seeing a sort of renewed purse around diversity in genomic studies. I think these are like, not quite . So straightforward causes as they at first seen so diversity of what, right? So people talk about ancestry. And or genetic ancestry. And as if everybody knows what everybody else means, and this is emphatically, not the [00:25:00] case. If you try asking a geneticist to define ancestry, they're probably going to stumble over their words. If you ask a physician how to find ancestry, you're going to also get a similar stumbling, but arriving somewhere different. Likewise, for somebody in public health, suddenly likewise, for somebody in the social sciences who tend to see ancestry as part of our identity. About the stories we tell about ourselves and how . We're linked to our forebears. So it's a very, very Slippy concept, but even if we focus in on, you know, somehow it's something to do with our genetics and something to do with how that's come down through the generations then. Yeah. A there's there's no clear definition, but b the sort of definition that's being inserted is continental ancestry. So European ancestry, African ancestry, Asian answer. And these massive categories end up looking very much like the races, biological races of old. Um, and that's just like the critique of the hat is what [00:26:00] we continue to see with people being like huge debate and the, in the sort of clinical algorithms, literature, and in practice about how race should be incorporated into, for example, a score for kidney function or for whether you've suffered from concussion. A lot of these daily used scores in clinical medicine factor in race. And we're having a moment of reckoning around that. And in general, social scientists have been pointing out to us for a long time that, you know, biological race is hugely problematic in all sorts of ways. You can use it only as a proxy for racism, basically. So all sorts of problems with biological race. But then some people are like, oh, okay. But there's something to do with genetics and that's that's okay. And, you know, I fear, um, others join me in this fear that the use of these continental ancestry categories, we kind of bring biological race back in through the, through the back door.[00:27:00] And of course, what we know is that we are not these like well-defined groupings, we're spread out with some structure, but we are spread out through genetic similarity space. And anything where we carved that space up is problematic. So I think there are some valid uses of these continental ancestry categories, but I also think that they're, um, a massively overused and it's in genetics. If what we end up meaning by diversity is these continental answers to category. We're putting a lot, a lot of emphasis on genetic differences between individuals and we've removed entirely other differences. So there's. A much, much, much smaller picture, which points out that these polygenic risk scores, they actually vary in predictive power. If you look across economic class or if you look across gender. So I think we risk focusing on genetic diversity. Losing out on all these other things, which actually [00:28:00] are probably more important for health disparities. Uh, so I see, I see a real there. In terms of work that is being done, I do think that people, that funding agencies are putting money where their mouth is when it comes to recruiting samples. But I have my reservations about the way that is being done. ------Dr Patrick Short: Here's a wild card, a discussion. So here in the UK, there's plans to sequence 5 million people through a new project called our future health. It's probably conceivable that in the UK or in the U S they'll sequence, everyone I'd say by 20 by 2040. I'll be really surprised if we haven't sequenced everyone in some sense, we'll then have solved the, this, this problem we've just been discussing because at least in a given country, everyone will be sequenced, but it doesn't, it doesn't solve the underlying problem, which is if there are fewer people of a particular group, whatever you would, whatever defined as then, the scores are going to underperform in those groups, [00:29:00] I'd assume relative to the larger groups. So I wonder how. Whether there is a solution in insight for this and how, how we ultimately handle this. Even if we've sequenced everyone in the U S then, um, individuals of a particular genetic and ancestry group, I I'm, you've now rocked my world view of what ancestry actually even is. So I'm struggling to place words for it, but yeah. What, what do you think about these population scale programs? Does that change things. ------Dr Anna Lewis: Well you put your finger on, on many interesting things there. And one thing I like that you're using representativeness language rather than diversity language, because diversity is very, very ill defined. Something you can see to do is to be representative of a particular population, for example, the population of a country. So you can see. And another thing I wanted to mention is it's not just raw numbers of individuals that count. So at least as long as it's genotype data that we have, or these individuals, these differences in predicitive [00:30:00] Power, capture all sorts of things. But one of the things that they do capture, which is kind of unambiguously associated with genetics are patterns of, of linkage disequilibrium, which is a feature of population history. So I know many geneticists who believe. That even if we get truly, truly massive samples . Of like, say we do lots of recruitment in Africa, they're still not going to perform as well. Those scores just because that is where genetic diversity variation. So, so there were those, there were those various things. Another thing that you mentioned that I liked is kind of like, how do we know when we've got, when we've got equity around, for example, the performance of these scores is even the performance of the scores, the right sort of level to look at if we're, if we're getting to equity. And I think a lot of people in bioethics would be like, well, there were just, just considerations here. You know, equity is important, but it's, we need to be much more [00:31:00] precise than that. And being more precise then has implications for what studies should be funded, etc. I think bioethics has stuff to learn here. And the sort of machine ethics literature, which has really shown that there are multiple mathematical definitions of fairness when it comes to the performance of predictive algorithms and, um, yeah. Think them through and select. Right? You mentioned, you know, UK by 2040 might have everybody sequenced. And what, what world would that look like? This is where the sort of contextual pieces are so important around genomics, right? Because. The thought of everybody being sequenced in the UK versus the U S for example, where I'm currently based. That's just two very different worlds. Right? So in the UK, you can somewhat imagine some direct links into the NHS. You've got the kind of cradle to grave mentality. The NHS has your back. Um, they're hopefully not going to [00:32:00] integrate genetic insights until they're able to do stuff with it. Um, if it turns out to be the case that, um, genetic education is really a clear need for a particular sort of genetic education, as realized than then it can be added to key stage two and it rolls out across the country. Right. The UK is sort of, in some ways, optimally set up for this type of thing. The key question of trust is a big one. Like you'd have to trust the government. And there was still lots of open questions. There was a big report that came out earlier this month on sequencing babies, where they did a lot of stakeholder analysis in the UK and ended up, ended up with a fairly rosy picture. I have to say, particularly compared to previous scholarship on this question. Interesting. That's the UK. In the U S almost none of the things that I just mentioned are in place, and it's just a completely different ball game. And then, um, many other countries are going to have similarly different contexts in which you can imagine this technology [00:33:00] being inserted. Yeah. Like imagine. Imagine the Chinese government having sequenced data or on everybody, or is it just. It's going to be very variable. I think that a lot of the bioethics work we do, we assume that our norms and principles are then just going to roll out and live with the technology. But the technology is definitely going to spread. I don't doubt that, but the norms and principles, which we sort of assume are packaged with it are not going to be as sticky as they cross-boundary national borders. ------Dr Patrick Short: Yeah. Fascinating. I feel like I could talk about this for another hour, but I think we both probably have zoom meetings we need to attend. So I wanted to thank you for your time, but I'd love to just close out with one more question about what you're working on now that maybe we've touched on or haven't discussed. Cause we talked about a lot of your past work in particular. We, we didn't touch too much on some of the discussion that's going around with polygenic scores and selection. On an embryo level. Maybe we can revisit that at another time, but I'd love to hear about what's on the horizon for you. What you're working on it most excited about. ------Dr Anna Lewis: [00:34:00] I'm thinking about two things or more specifically writing two grants. One is around sequencing babies, and I think the context of. New genetic therapies really changes the landscape because in order for many of those genetic therapies to have an optimal effect, you need to know who would benefit from them as soon as possible. And that means sequencing babies. And we know people have all sorts of hesitations around that. So what does putting best foot forward look like for sequencing? And the other one that I mentioned this before is do you know, mixes, divisional, LC, ethical, legal social implications field. But I think that there are other fields. So generation of biological data for machine learning is an obvious one where what we need to do is sort of learn from the successes and failures of the LC genomics world and create these new LC fields. Um, for these other technologies, I'm also excited about that. Excellent. If people want to follow you and keep up with your work, I know you're on [00:35:00] Twitter and they can visit you on the Harvard website if they just type it in ACF lewis.com is my website and I met not a very good Twitter. I think I've been once in the last couple of years. ------Dr Patrick Short: Great. So go ACF lewis.com. I checked out your website before the call, and I know you said that you've stopped writing as much during the pandemic, but there's a huge amount of really interesting. It seems like for a long period of time, you were basically pulling all the most interesting things that were happening in a given month. In genomics and basically just doing a rundown. So it was a little bit of a trip down memory lane. I got to read about the head, Jan Quaid, gene editing issues, and all sorts of things that were kicking off, but one or two years ago, so definitely worth checking out. ------Dr Anna Lewis: Well, thank you. Well, um, this is a lot of fun. Thank you, Patrick, for having me on. ------Dr Patrick Short: Thanks so much for taking the time and hope we can do this again soon. Maybe we can do a deep dive into. Into baby sequencing. I need to read the outputs of the work that was done here in the UK. I haven't had a chance to check up on it yet, but, um, I'm interested in hearing what you're working on [00:36:00] and learning a little bit more about that. Thanks so much.