# Lesson 7 — Genetics, Genomics, and Health (v3 expanded)

*Companion-podcast transcript • Sarah & Kiffer*  
*~4953 words • ~26.9 min audio*

---

**Sarah:** Welcome back to Office Hours. I'm Sarah.

**Kiffer:** And I'm Kiffer. Today we're working through Lesson 7, Genetics, Genomics, and Health. And I want to flag up front that 'genetic' might be the most overloaded word in health science. When somebody says a disease is 'genetic,' that single word can mean six or seven completely different things, and the rest of the conversation usually goes sideways unless you stop and ask which one.

**Sarah:** Let's set the stage. The lesson covers a lot of ground. It walks from Mendel's pea plants in the 1860s through the dark eugenics history, the Human Genome Project, the distinction between single-gene and complex disease, and the contemporary world of direct-to-consumer testing. That's a huge arc.

**Kiffer:** It is. And the reason we cover so much ground in one lesson is that all of these threads are still alive in present-day conversations. Eugenics isn't just a museum piece. Polygenic risk scores aren't science fiction. The twenty-three-and-me data breach is a real privacy story. So we need to see the whole shape, and we need to see why the contemporary debates inherit problems from the older ones.

**Sarah:** Let's start at the beginning. Mendel.

**Kiffer:** Gregor Mendel was an Augustinian friar in what is now the Czech Republic, growing pea plants in a monastery garden through the 1850s and 1860s. He was a careful experimentalist, tracking seven discrete traits across thousands of plants over multiple generations. He published his laws of inheritance in 1866 in the proceedings of the Brno Natural History Society, and the paper was basically ignored for thirty-five years.

**Sarah:** Why was it ignored? That always strikes me as one of the strange features of the history.

**Kiffer:** Two reasons. The journal had small circulation. And the scientific establishment of the 1860s was preoccupied with Darwin's On the Origin of Species, which had appeared in 1859. Darwin needed a mechanism of inheritance for natural selection to work, and Mendel had actually provided it. The work just didn't reach the people who needed it. Mendel was rediscovered around 1900 by three botanists working independently — de Vries, Correns, and von Tschermak. By 1910, Mendelian inheritance had been established for fruit fly traits in Thomas Hunt Morgan's lab at Columbia. And from there modern genetics could begin in earnest.

**Sarah:** The next big chapter is the chemistry. We don't even know D N A is the genetic material until the 1940s.

**Kiffer:** Right. Through the 1920s and 1930s, the assumption was that genes were made of protein. Protein was the only macromolecule complex enough to seem like a plausible carrier of hereditary information. The 1944 Avery, MacLeod, and McCarty experiments at the Rockefeller Institute, building on earlier work by Frederick Griffith, demonstrated that D N A — long dismissed as structurally simple — was actually the genetic material. That was a big shift.

**Sarah:** And then 1953, the double helix.

**Kiffer:** Right. April 1953, three back-to-back papers in Nature. Watson and Crick at Cambridge, working largely from theoretical reconstruction. Wilkins and colleagues at King's College London. And Rosalind Franklin and Gosling, also at King's, who produced the X-ray diffraction images that revealed the helical structure. The famous Photograph fifty-one was Franklin's. And the credit dispute that followed is one of the most contested in twentieth-century science.

**Sarah:** Worth pausing on. Franklin's data was shared with Watson and Crick without her knowledge.

**Kiffer:** Without her knowledge or consent. Wilkins shared it. The structure itself was immediately suggestive of both a replication mechanism — the two strands separate, each templates a new partner — and a storage mechanism, where the linear sequence of base pairs carries information. The 1962 Nobel went to Watson, Crick, and Wilkins. Franklin had died in 1958 of ovarian cancer at thirty-seven, and Nobel rules don't allow posthumous awards. But her contribution was substantially minimized for decades, including in Watson's 1968 memoir, which painted a picture of her that subsequent historians have thoroughly refuted.

**Sarah:** And it's now a standard case study in scientific ethics.

**Kiffer:** On credit, collaboration, and the use of others' data. It also shaped how women in science were perceived and treated for a generation. The canonical history of a discovery can be constructed to favor certain participants over others, and that construction itself becomes part of what the field passes on.

**Sarah:** Skipping forward, 1990 to 2003 brings us the Human Genome Project.

**Kiffer:** Coordinated international effort, led by the U S National Human Genome Research Institute under Francis Collins, with parallel teams in the U K, France, Germany, Japan, and China. And a private competitor, Celera Genomics under Craig Venter, raced to the same finish line with a different methodology. They announced essentially simultaneous completion in 2003 — the fiftieth anniversary of the Watson-Crick-Franklin papers, which was nicely symbolic.

**Sarah:** And the cost was about two-point-seven billion dollars.

**Kiffer:** Two-point-seven billion dollars in 1990s and early 2000s funding. Sequencing a single genome was a major project. The big surprises from the project: humans have roughly twenty thousand protein-coding genes, not the one hundred thousand people had been expecting. And over ninety-eight percent of the genome is non-coding sequence, which was initially mysterious and is now understood to do a tremendous amount of regulatory work — transcription factor binding, non-coding R N As, chromatin organization.

**Sarah:** And then the cost collapse is its own story.

**Kiffer:** It's astonishing. By 2014, a complete genome was about a thousand dollars. By 2024, two hundred to four hundred dollars. That's faster than Moore's Law in computing, and it's what has enabled everything that's followed — biobanks at the half-million-person scale like U K Biobank, the U S All of Us program targeting one million participants, China's Kadoorie Biobank, the Quebec CARTaGENE cohort. We've moved from 'where is the gene for X' to 'what is the joint distribution of small effects across thousands of variants, and how does that distribution interact with environment.' That's a completely different research enterprise.

**Sarah:** Okay. Now we have to do the part of the lesson that's harder. Eugenics.

**Kiffer:** Yeah. And this is one of the sections where I think the course has a responsibility to be direct. Genetics has been used in service of human dignity and in service of state violence, and both stories are part of public health's inheritance. The eugenics movement was not a fringe view. From roughly the 1880s through 1945, it was the scientific mainstream in Europe and North America.

**Sarah:** Francis Galton coined the term in the 1880s.

**Kiffer:** Galton was Darwin's cousin. He was an extraordinary scientist by some measures — foundational contributions to statistics, including correlation and regression. And an authoritarian eugenicist by others. He proposed that human populations should be improved through selective breeding, encouraging desirable traits and eliminating undesirable ones. The proposal combined statistical sophistication with extraordinarily naive assumptions about the heritability of complex social traits.

**Sarah:** And the movement around him.

**Kiffer:** Major universities established eugenics departments. The American Eugenics Society ran 'better baby' contests at state fairs, judging infants on health and racial features. Cold Spring Harbor's Eugenics Record Office, founded in 1910, became the institutional center of U S eugenic research. Progressive politicians of the era supported eugenic principles. This was not a fringe view, it was the scientific and political mainstream of its time.

**Sarah:** And then in 1927 there's the U S Supreme Court decision, Buck v Bell.

**Kiffer:** Buck v Bell. Justice Oliver Wendell Holmes — typically remembered as a progressive jurist — wrote the majority opinion upholding Virginia's forced sterilization law. His famous sentence, 'three generations of imbeciles are enough,' is canonical. The case involved Carrie Buck, an eighteen-year-old who had been raped at seventeen and institutionalized. Subsequent historical investigation has shown that Carrie Buck was almost certainly of normal intelligence, that her daughter Vivian similarly tested as normal-intelligence, and that the whole case was substantially fabricated by Virginia officials to produce a test case. Buck v Bell has never been overturned. It's still good law.

**Sarah:** And there are roughly seventy thousand forced sterilizations in the U S under those state laws.

**Kiffer:** Roughly seventy thousand between 1907 and 1937, with some state programs continuing into the 1970s. The Nazi defendants at Nuremberg specifically cited Buck v Bell as legal precedent. The German Sterilization Law of 1933 had been substantially modeled on U S state laws. The U S eugenics movement provided technical, legal, and ideological support for Nazi medical policy. Eugenicists like Harry Laughlin received honorary degrees from German universities in the early 1930s. That part of the history is regularly omitted from popular accounts.

**Sarah:** And now the Canadian story, which I really want students to learn.

**Kiffer:** Alberta's Sexual Sterilization Act, enacted in 1928, not repealed until 1972. Approximately twenty-eight hundred people sterilized over forty-four years. B C had a parallel act from 1933 to 1973. Saskatchewan considered similar legislation but did not enact it. Manitoba operated less-formal sterilization programs through provincial mental health institutions without specific enabling legislation. The Alberta act was enthusiastically supported by progressive Alberta governments, including Premier Ernest Manning for most of his tenure from 1943 to 1968.

**Sarah:** And the population affected was not random.

**Kiffer:** Not at all. Indigenous women were targeted at rates far above their proportion of the population. In some Alberta institutions, Indigenous women were the majority of those sterilized. Eastern European immigrants, particularly Ukrainian Canadians, were heavily represented. Catholic women more than Protestant. Working-class women more than middle-class. The program was eugenic in a specific colonial, racialized, and gendered sense. And the reckoning is incomplete. Forced sterilization of Indigenous women in Canadian healthcare settings — not under the specific sterilization acts, but in regular hospital and clinical encounters — has been documented continuing into the 2000s and 2010s. Investigations by Senator Yvonne Boyer and journalist Karen Stote have laid out the pattern. Class-action litigation is ongoing.

**Sarah:** And then there's Nazi Germany, which is where this logic was taken to its conclusion.

**Kiffer:** Over four hundred thousand Germans were forcibly sterilized between 1934 and 1945. The T-four program from 1939 to 1945 — named after the address of its headquarters at Tiergartenstrasse four in Berlin — murdered approximately two to three hundred thousand disabled people in psychiatric institutions, including children. And the personnel and techniques developed in T-four were subsequently used in the Holocaust. The continuity between eugenic medical practice and genocide is direct, not metaphorical.

**Sarah:** And the post-war response was the Nuremberg Code.

**Kiffer:** Drafted as part of the Nuremberg Doctors' Trial, 1947. The Code articulated principles for research with human subjects — voluntary informed consent, scientific justification, proportionate risk-benefit. It's the founding document of modern research ethics and the conceptual ancestor of the Belmont Report and the contemporary T C P S framework we use in Canada. The Code was a direct response to the medical experiments Nazi physicians had conducted on concentration camp inmates. But the reckoning was uneven. Alberta's law remained in force for twenty-seven years after the Nuremberg Code. Sweden's compulsory sterilization continued until 1976. North Carolina conducted eugenic sterilizations into the 1970s.

**Sarah:** Why do we make students sit with this material?

**Kiffer:** Two reasons. First, the field's amnesia about its own history is genuinely dangerous. Eugenics wasn't pseudoscience operating outside public health. It was inside the tent. If we let ourselves imagine the field has always been innocent, we miss the patterns that produced the harm. Second, contemporary genetic medicine raises analogous questions in updated forms — prenatal screening for Down syndrome with most detected pregnancies terminated, embryo selection through preimplantation genetic diagnosis, polygenic embryo screening that's now offered commercially. None of these have settled answers.

**Sarah:** And the disability rights critique here is important.

**Kiffer:** The mainstream view distinguishes between individual reproductive choice, which is protected with autonomy framing, and state-imposed reproductive policy, which is rejected because of the eugenic legacy. That distinction matters. Coercion is a different ethical category from choice. But the population-level effects can look similar, and the social signal that some lives are less worth living is real. Disability scholars — Alison Kafer, Rosemarie Garland-Thomson, Catherine Frazee in Canada — have argued this for decades. The critique deserves serious engagement rather than dismissal.

**Sarah:** All right. Let's move to the substantive distinction the lesson sets up between Mendelian and complex disease.

**Kiffer:** This is the part where 'is this disease genetic' starts to mean something specific. Mendelian diseases follow single-gene, high-penetrance inheritance. Cystic fibrosis, with C F T R mutations. Huntington disease, with H T T trinucleotide repeats. Sickle cell disease, with beta-globin mutations. Phenylketonuria, with P A H mutations. BRCA-1 and BRCA-2-associated cancer syndromes. Familial hypercholesterolemia. There are about seven thousand known Mendelian diseases. Individually rare, collectively important — roughly five to seven percent of the population carries a recognized pathogenic variant for one of them.

**Sarah:** And these have driven much of the success in genetic medicine.

**Kiffer:** The first F D A-approved gene therapies have all been for Mendelian conditions. Luxturna in 2017 for a form of inherited blindness. Zolgensma in 2019 for spinal muscular atrophy. Casgevy in 2023 — the first C R I S P R-based therapy — for sickle cell disease and beta-thalassemia. These therapies are real and they're transformative for affected patients. They're also extraordinarily expensive. Zolgensma is about two-point-one million dollars per dose. The technical feasibility of correcting single-gene conditions has substantially outrun the policy infrastructure for making the treatments available to those who need them.

**Sarah:** And complex diseases are the polygenic ones.

**Kiffer:** Most chronic disease. Type two diabetes, coronary heart disease, schizophrenia, depression, most cancers, hypertension, asthma, type one diabetes, A D H D, autism. Hundreds to thousands of common variants each contribute a tiny amount of risk, and environment matters as much as or more than genetics. There is no 'gene for type two diabetes' in any clinically meaningful sense. The framing you sometimes see in popular media — 'scientists find the gene for X' — is just wrong for these conditions.

**Sarah:** And the methodology that opened this up is G W A S.

**Kiffer:** Genome-wide association study. You genotype large samples of cases and controls at hundreds of thousands or millions of common variants — single-nucleotide polymorphisms — and you statistically identify variants associated with the disease. First G W A S was published in 2005 for age-related macular degeneration. Sample sizes have grown from thousands to hundreds of thousands to millions. The findings are consistent: many variants, each of small effect, distributed across the genome. The cumulative effect of any single variant typically explains less than one tenth of one percent of population variance. And the combined effect of all identified variants explains a substantial but minority fraction of total heritability.

**Sarah:** Let's pause on heritability, because the lesson is firm that this concept is constantly misinterpreted.

**Kiffer:** It's maybe the most-misinterpreted construct in popular health science. Heritability is the proportion of variance in a trait, in a given population, that can be attributed to genetic variance. It's a population statistic. It is not an individual decomposition. If somebody tells you height is eighty percent heritable, that does not mean your height is eighty percent determined by your genes.

**Sarah:** Give me the standard example.

**Kiffer:** The cleanest example is P K U. Phenylketonuria — a single-gene Mendelian condition with extremely high heritability. And essentially completely modifiable through dietary phenylalanine restriction. The condition is 'genetic' in every sense. And the difference between catastrophic intellectual disability and normal development is determined entirely by diet. Heritability is silent on whether environmental intervention works.

**Sarah:** And the other big misuse is comparing groups.

**Kiffer:** Within-group heritability tells you nothing — literally nothing — about between-group differences. Height heritability is high within most populations, but a substantial fraction of between-population height differences over the past century has been driven by nutritional improvements, not by genetic change. The same logic applies to other traits where between-group differences have been wrongly attributed to genetic causes. This is a technical statistical point, and it's the foundation of careful work on race, ancestry, and genetics. It's also the point most consistently butchered in popular discourse.

**Sarah:** And heritability depends on the environment of the population studied.

**Kiffer:** Right. The heritability of height can be high in one country and low in another based on nutritional variation. When nutrition is uniformly good, most height variation reflects genetic differences and heritability is high. When nutrition is highly variable, more height variation reflects environmental differences and heritability is lower. The same gene, the same person, would produce different heritability estimates depending on which population is being measured. The statistic is about the distribution of causes of variation in a population, not about the causes of any individual's trait value.

**Sarah:** Polygenic risk scores.

**Kiffer:** P R S. They combine effects across many G W A S variants into a single risk estimate per person. They're improving fast. They're scientifically real. And they have several limitations worth knowing. Predictive accuracy is modest — even a 'high' P R S typically shifts your absolute risk modestly, not dramatically. A high P R S for type two diabetes might shift you from ten percent lifetime risk to fifteen percent, not from ten to fifty.

**Sarah:** And the equity issue.

**Kiffer:** Accuracy is unequal across populations, because most G W A S data come from European-ancestry cohorts. As of 2024, about eighty-six percent of G W A S participants globally were of European ancestry. P R S derived from European-ancestry G W A S predict substantially worse — by some estimates two to five times worse — in African-ancestry, East Asian-ancestry, and other populations. So deploying P R S clinically before that gap closes risks widening health disparities, not narrowing them.

**Sarah:** And clinical actionability is unclear.

**Kiffer:** Yeah. Telling someone they have high genetic risk for a condition where they should already be doing the recommended preventive behaviors — eat well, exercise, don't smoke — provides limited new information and may produce fatalism or motivation, depending on the person and the framing. The evidence base for P R S predicting outcomes is much stronger than the evidence base for P R S-guided interventions actually changing outcomes.

**Sarah:** There's a callback here to Lesson 1, where we set up the social determinants framing.

**Kiffer:** Yeah. The risk with P R S, and with precision medicine generally, is that it pulls resources and attention away from population-level interventions that work. A perfect cardiovascular precision program would produce small population-level effects compared with reducing tobacco use, improving diet, increasing activity, and managing blood pressure — all of which work at population scale and don't require individual genetic testing. The opportunity cost of precision investment is the population-health investment foregone. That tension recurs across this course.

**Sarah:** Let's move to the part where genetics in public health has actually worked. Newborn screening.

**Kiffer:** This is one of public health's quiet successes. It began in 1962 when American microbiologist Robert Guthrie developed a bacterial inhibition assay for phenylalanine that could detect P K U from a heel-prick blood spot. P K U is autosomal recessive. Affected infants lack a functional enzyme to metabolize phenylalanine. They accumulate it to toxic levels and develop severe intellectual disability if untreated. Detected within the first few days of life and treated with a phenylalanine-restricted diet, they develop essentially normally. The Guthrie test transformed P K U from a devastating childhood condition into a manageable dietary condition.

**Sarah:** And it spread fast.

**Kiffer:** By 1970, all fifty U S states had mandatory P K U screening. Canadian provinces followed through the sixties and seventies. Today every newborn in Canada is screened for twenty to thirty conditions, varying by province. The Wilson-Jungner criteria from 1968 govern which conditions qualify. The condition has to be serious without treatment. It has to have a recognizable preclinical phase that screening can detect. Effective early treatment has to be available. And the test has to be accurate and feasible at population scale. It's a quiet, institutional success. A heel-prick on day one to three of life, sent to a provincial lab, processed automatically. Most parents in Canada don't even know it happened. And it catches roughly one affected infant per thousand screened across the panel.

**Sarah:** Precision medicine in oncology has been real.

**Kiffer:** In oncology it has genuinely transformed care. HER-two testing in breast cancer identifies tumors that respond to trastuzumab — Herceptin. E G F R testing in lung cancer identifies tumors that respond to specific targeted therapies. BRCA testing identifies elevated risk for breast and ovarian cancer and can guide prophylactic surgery, intensive screening, or PARP inhibitor therapy. Microsatellite instability testing in colorectal and other cancers identifies tumors responsive to immune checkpoint inhibitors. These are real and they have transformed prognoses for specific patient populations. Comprehensive tumor genomic profiling is now standard in many advanced cancer cases.

**Sarah:** Outside oncology it's been slower.

**Kiffer:** Much slower than the rhetoric. For most chronic disease, the strongest interventions remain population-level — blood pressure control, cholesterol, behavior, environment. The simpler genetics of cancer, where tumors often have driver mutations targetable by specific drugs, just doesn't have a clean analogue for diabetes or depression. Mental health treatment selection remains largely empirical rather than precision-guided. Type two diabetes prevention is still mostly about diet, weight, and activity. The precision-medicine promise has consistently outrun delivery outside oncology.

**Sarah:** Epigenetics is another piece the lesson treats carefully.

**Kiffer:** Yeah. Epigenetics studies heritable changes in gene expression that don't involve changes in D N A sequence. The main mechanisms are D N A methylation — adding methyl groups to cytosine bases, generally silencing nearby gene expression. Histone modifications, which alter the proteins that package D N A. And non-coding R N As that regulate gene expression. The public health significance is twofold. Epigenetic marks respond to environment — diet, stress, toxin exposure, social conditions — and can persist after the exposure ends. That gives you a mechanistic pathway from social determinants to biological outcomes. The Dutch Hunger Winter cohort, which we touched on in Lesson 6, shows persistent methylation differences in people exposed in utero decades earlier.

**Sarah:** And the second piece.

**Kiffer:** Some epigenetic marks can be transmitted across generations. Extensively documented in animal models, increasingly characterized in humans. But the field has been overhyped. Popular claims that 'your trauma is in your D N A' or 'your grandparents' experiences shape your gene expression' or 'you can edit your epigenome through meditation' are oversimplifications that obscure what the actual research shows. Most epigenetic marks turn over rapidly. Most transgenerational effects in humans are small. The core science is real and is reshaping how we think about the gene-environment boundary. The pop-science framing is mostly noise. And epigenetic clocks — Horvath's clock, GrimAge, PhenoAge — are increasingly used as biological-age measures in aging research.

**Sarah:** Direct-to-consumer testing. Twenty-three-and-me, Ancestry, My Heritage.

**Kiffer:** Over thirty million people have used commercial genetic testing services. Ancestry information is reasonably accurate. Single-gene tests for known variants are accurate when validated. Health-risk predictions vary widely in quality. And the privacy questions are sharper than the clinical questions. These companies build genetic databases that have been used in research — millions of people have contributed to research data, though not all consent processes have been adequate. They've been used in law enforcement — the 2018 Golden State Killer identification through the G E D match genealogy database was a watershed and has been followed by hundreds of cold-case identifications. And increasingly for commercial partnerships with pharmaceutical companies.

**Sarah:** And there was a substantial breach.

**Kiffer:** The 2023 twenty-three-and-me data breach exposed genetic data and ancestry information for millions of users. Class-action litigation produced settlements. The company filed for bankruptcy in 2025, which raises questions about what happens to genetic data when a D T C company fails financially. Canada has the Genetic Non-Discrimination Act, passed in 2017, which prohibits insurance companies and employers from requiring or using genetic test results, with criminal penalties for violations. The Act survived a 2020 Supreme Court of Canada constitutional challenge. But it doesn't regulate what D T C companies do with the data, doesn't restrict law enforcement use of genealogy databases, and doesn't address the privacy of relatives who can be identified through one family member's test.

**Sarah:** That last point is important. Your privacy isn't just yours.

**Kiffer:** If you submit your saliva to a commercial database, you're partially exposing every biological relative — siblings, parents, children, cousins, distant cousins — who shares enough D N A to be identifiable. They haven't consented. The standard advice from medical geneticists is: be cautious about D T C testing, get genetic counseling from a qualified clinician for anything you'd act on, and remember that the privacy implications extend beyond you.

**Sarah:** Let's pull this together. What do you want students to leave this lesson with?

**Kiffer:** Walk me through them.

**Sarah:** First takeaway. The word 'genetic' is overloaded and lazy on its own. When you hear it, ask what kind of genetic claim is being made. Single-gene Mendelian? Polygenic risk score? Heritability estimate? Population-level claim about ancestry? These are radically different things, and conflating them produces sloppy thinking and sometimes harmful policy.

**Kiffer:** Second takeaway. The history of eugenics is part of public health's inheritance. The Canadian story — Alberta and B C, the targeting of Indigenous women, the slow reckoning — should be taught alongside the success stories. Contemporary debates about prenatal screening, embryo selection, and germline editing don't repeat the explicit coercion of historical eugenics, but they raise analogous questions that haven't been settled. The disability rights critique of contemporary genetic medicine deserves serious engagement.

**Sarah:** Third takeaway. Heritability is a population statistic, not an individual decomposition, and it tells you nothing about between-group differences. If somebody is using heritability to argue about race or class or fixed group differences, they are almost certainly misusing the construct. P K U is the cleanest counterexample — high heritability, fully modifiable by diet.

**Kiffer:** Fourth takeaway. Newborn screening is one of public health's quiet successes. Mendelian genetics translated into population practice with measurable benefit. Twenty to thirty conditions screened per province in Canada. Cheap, institutional, mostly invisible. Worth knowing exists, and worth understanding why the Wilson-Jungner criteria matter — they're why we don't screen for everything we could screen for.

**Sarah:** Fifth takeaway. Precision medicine is real in oncology and slower outside it. The opportunity cost question — what we don't fund when we invest in expensive individualized interventions — is the question the field hasn't fully reckoned with.

**Kiffer:** Sixth takeaway. Polygenic risk scores are scientifically real and clinically modest, with substantial equity concerns because the underlying G W A S data are mostly European-ancestry. Deploying them clinically before diversifying the training data risks widening, not narrowing, disparities.

**Sarah:** Seventh takeaway. Direct-to-consumer genetic testing has expanded much faster than the policy infrastructure to govern it. Privacy implications extend beyond yourself to your biological relatives. The standard advice from medical geneticists is: be cautious, get counseling on anything you'd act on, and remember that the D T C report is informational, not diagnostic.

**Kiffer:** Eighth, and I want to put a slight gloss on the synthesis. The pace of technical change in this field is genuinely faster than the policy and ethical infrastructure that should govern it. Sequencing cost has dropped over a million-fold in two decades. C R I S P R-based therapies for sickle cell disease and beta-thalassemia were approved in 2023. Polygenic embryo screening is being marketed commercially. The 2018 He Jiankui case in China — the germline editing scandal — illustrated how easily boundaries can be crossed. The students listening to this are going to spend their careers working in a field where the technology continues to outrun the governance. That's worth being aware of, and it's part of why the historical context we covered in section two matters: it's a reminder that 'respectable scientific opinion' has confidently endorsed practices we now recognize as horrific. Confident contemporary judgment isn't always reliable either.

**Sarah:** Before we wrap, I want to underline one of the methodological points the lesson makes carefully. Mendelian randomization.

**Kiffer:** Yeah, M R. It uses genetic variants as instruments for modifiable exposures. The logic is that alleles are randomly assorted at conception, which is a kind of natural randomization. If a variant is reliably associated with, say, B M I, and B M I is causally related to coronary disease, the variant should be associated with coronary disease through the B M I pathway. And if the variant affects coronary disease only through B M I, you can use the variant's association with the outcome to estimate the causal effect of B M I — without the confounding that conventional observational studies suffer. Modern M R has tools like M R-Egger and weighted median estimators to handle assumption violations. It's reshaped what observational research can claim causally, and you'll see it in H S C I three-forty-one.

**Sarah:** And another point worth a quick callback. Mendel was rediscovered in 1900. Eugenics had been building since the 1880s. So the eugenics movement actually predates the rediscovery of Mendel's laws.

**Kiffer:** That's a good observation. The first wave of eugenic enthusiasm was running on Galton's statistical work and on broad Darwinian intuitions, without much understanding of how inheritance actually worked. Once Mendelism was rediscovered, eugenics adopted it enthusiastically and badly — applying Mendelian inheritance models to complex social traits where they didn't fit. The bad science amplified the bad politics. It's worth knowing that the movement didn't follow from the science. The movement was already there, and it pulled in scientific authority where it could find it.

**Sarah:** Next lesson we move to behaviors and mental health. The tobacco playbook and where it's been adapted to other targets.

**Kiffer:** Right. And there's a nice thematic continuity, because the question of how much individual behavior matters versus how much structure matters is going to come up again. Genetics is one of the things that gets used to argue 'this is intrinsic.' The behavioral lesson will push hard on what 'intrinsic' is doing in that sentence.

**Sarah:** See you next time.

**Kiffer:** Take care, everyone.