Week 10 Genetic Engineering

This week we discuss current and near-future potential applications of genetic knowledge. Recent technological developments have led to an increased interest in the broad field of personalized medicine, including attempts to make predictions about a person on the basis of their genotype. We will consider three issues in particular: Could we? Should we? And, would we?


  • Introduce current technologies in genetic engineering.
  • Consider the potential effectiveness of these technologies (could we?).
  • Consider the ethical issues around use of these technologies (should we?).
  • Consider the social acceptability of personal or policy applications of these technologies (would we?).

Lecture Notes

The movie Gattaca (1997) is about a near-future world where people are judged on their genetic profiles. It’s very smoky, includes at least one too many unnecessary plot twists, and is poorly lit with a score that is inexplicably loud. But its scientific basis is essentially what we now call polygenic scores. This future society uses polygenic scores to predict an individual’s health and behavioral outcomes at birth, screen for employment eligibility, and evaluate romantic partners. Parents can choose to selectively implant embryos with the best polygenic scores (and in a deleted scene, given the option to edit the DNA beyond that for an additional fee): https://youtu.be/PFjaOnCp0lo

Could we?

The technology to attempt this already exists. Direct-to-consumer genotyping services, like 23andMe, offer users genomic ancestry estimates and predictions of probabilities for a variety of health and behavioral outcomes. Although the major DTC companies don’t venture too far into unusual or non-disease-relevant traits, a nearly infinite supply of third-party services now allow you to upload your DNA files obtained from these companies to platforms that will estimate your status on any polygenic score that exists.

But how accurate are these services likely to be? For outcomes that are due to a single gene effect, they can be hugely useful. Screening for major disease carrier status is common among individuals who are known to be at risk. For example, individuals of Ashkenazi Jewish ancestry are often screened for Tay-Sachs carrier status, which is a recessive neurological disorder that leads to a horrific, painful death usually within the first few years of life. Israel has a nationwide screening program available to encourage carriers to know their status and use it to make reproductive decisions (for example, who to date or whether to adopt instead) - and the program of socially-supported widespread screening is effective. 1 out of every 30 people of Ashkenazi Jewish ancestry is a carrier of Tay-Sachs, but the birth rate of affected individuals is substantially below what would be expected (only 1 in 3,500 births, less than the expected rate of 1/30 * 1/30 = 1 in 900). This suggests that, with widespread screening and even in the absence of policy mandates, people are using genetic information to make reproductive decisions. But this is an extreme example - how far would such screening go?

One question would be: how certain can we be? For behavioral outcomes, the absolute upper limit for how well we could ever predict from a person’s genome is the estimate of heritability. If you take the square root of a heritability estimate, you get the absolute highest correlation that could ever be achieved by a polygenic score for that trait (assuming that we could identify all the genes to go into that score and that all of the assumptions that go into estimating heritability are correct - so it would be an optimistic assumption. If the heritability of the trait were around 50%, that would give us a maximum potential correlation between an absolutely perfect polygenic score and the trait of around 0.7. And remember, this is a VERY optimistic prediction. Polygenic scores are limited to common variants, and often heritability estimates from GWAS data are around half of what’s obtained in twin data - meaning that construct that’s 50% heritable might only be 25% heritable due to common genetic variants (where rare variants will likely always remain too rare to properly score - at least until we understand function a lot better than we do now) which would put its absolutely most optimistic maximum correlation with the trait outcome at around 0.5.

Would we?

Early selective implantation technology was, like many reproductive technologies, originally developed for use in agriculture, to produce more efficient ratios of males-to-females and to breed more productive animals overall. The same basic technology that is used to screen livestock animal embryos for sex or size or temperament could in principle be applied in humans, as well - although the process does necessarily involve in vitro fertilization (IVF) to allow screening before implantation. It’s common for fertility clinics to screen embryos for major genetic diseases before implantation. It’s also not uncommon for the selection of embryo sex chromosomes to be an option. There have even been recent reports of clinics offering screening on the basis of polygenic scores, including those for height and intelligence.

One major deciding factor that comes up in screening, in general, is whether there are any effective interventions. That is - if you know about something, would you be able to do something about it? And if that something is genetic, does the intervention necessarily need to be genetic, too? Myopia (nearsightedness) is heritable, but we have a very effective environmental intervention: glasses. There’s a major metabolic disorder called PKU (phenylketonuria). It’s caused by a single genetic mutation that usually arises de novo (not inherited from either parent) and if left untreated will cause severe intellectual disability and behavioral problems (a result of the body being unable to break down phenylalanine, an amino acid that is present in most foods). It occurs about 1 out of every 10,000 births - and I can guarantee that every single person in this class was screened for it at birth. The PKU screen is non-genetic - it’s a blood spot test that measures metabolite levels. It’s administered at or shortly after birth in every country with a developed health care system - because it’s treatable. As long as someone with PKU avoids foods containing phenylalanine, they will be largely phenotypically normal - a genetic disorder with an environmental intervention.

Should we?

The discussion about whether or not to apply genetic engineering to humans is almost certainly less a question of “Yes or No?” and more “Where’s the line?”. The history of technology consistently follows a trend where new technologies are first demonized, then accepted (see for example: bicycles, writing on paper instead of slate tablets, radio, in vitro fertilization). Very few people working on these issues argue either there should never be any application of screening or even modification (see: major single-mutation-determined disorders) or that there should be no restrictions. But where to draw that line varies substantially between individuals, and is an issue that we need to address with increasing urgency. As the technology becomes cheaper and more reliable, there will continue to be a market for services like direct-to-consumer genotyping and embryo selection, and even embryo genetic modification. What we choose to do with this technology may have sweeping consequences - and it may turn out that we’ve spent a bunch of money and human pain on predictions that were never going to be accurate in the first place.

“There is no gene for the human spirit.” - the actual tagline for Gattaca

Prep Work

Participation Activities

  • Read & Discuss via Perusall: Kirby 2000 The new eugenics in cinema- genetic determinism and gene therapy in GATTACA. https://www.depauw.edu/sfs/essays/gattaca.htm
    • This chapter was published 20 years ago. As you’re reading, consider: how relevant are the issues raised in it today?
  • Read & Discuss via Perusall: Karavani et al 2019 Screening Human Embryos for Polygenic Traits Has Limited Utility. https://doi.org/10.1016/j.cell.2019.10.033
  • Read & Discuss via Perusall: Andorno et al 2020 Geneva Statement on Heritable Human Genome Editing- The Need for Course Correction. https://doi.org/10.1016/j.tibtech.2019.12.022
  • Journal Response: Direct-To-Consumer Genetic Testing
    • I’ve posted some of my own results from various companies to Moodle. Everyone is welcome to use my results. If you have your own direct-to-consumer genetic info, you may use that instead to respond to the following prompt.
    • Journal Assignment: Select any three themes from behavior genetics (see Lecture Notes, the bottom of Week 1). For each theme, write a brief (100-200 words) description of how that theme relates to the presentation of direct-to-consumer genetic testing results. You can address the science, communication, ethics, any aspects that strike you as important.
  • Journal Response:: Themes in Fiction
    • Find a fictional portrayal of genetic testing or engineering and connect it to three of the themes in behavior genetics.
    • You can pick anything - feel free to be creative and/or esoteric. A few examples:
      • Gattaca (1997): There are sadly no non-rental options available right now, but this playlist of 8 short clips + the trailer hits the important points in <20 minutes.
      • Rampage (2018) or any of the Jurassic Parks: Not about humans, but there’s plenty of discussion of complex behavioral phenotypes, genes, and environments.
      • You could absolutely analyze three themes relevant to the opening line of Lizzo’s Truth Hurts, "I just took a DNA test, turns out I’m 100% that b*tch."
    • Journal Assignment:
      • At the top, state what fictional work you are discussing (give me enough detail that I could look it up if I’m not familiar).
      • Select any three themes from behavior genetics (see Lecture Notes, the bottom of Week 1).
      • For each theme, write a brief (100-200 words) description of how that theme relates to the fictional portrayal of genetics.
  • Class Chat, Thursday, 11:00 am - 12:20 pm (CT)