Note: I've recieved some critiques about the title of this post, from people that are more adept at spotting double meaning than I am. To be clear, 'killer app' is meant strictly as it would for a software platform. I should also clarify that this post is not meant to imply anything about my personal views on abortion, which I still struggle with.

Product/market fit applies to research too. The field of human genetics has a compelling product - insights about how DNA works - but to date it has never had much of a market. Unless you have a family member with a genetic disease, or you happen to be addicted to genealogy, you probably haven’t felt like DNA sequencing has changed your life[1].

I think this will soon change - drastically - because there is a looming killer app use case for genetics research: prenatal DNA sequencing.

What is prenatal sequencing?

Within five years, a pregnant woman will be able to read the full genome sequence of her future child in the first trimester of pregnancy. This will allow her to screen for potential genetic diseases with time to elect for the least invasive forms of abortion.

Screening for genetic disease in pregnancy is already commonplace - in the US, 72% of pregnant women are tested for Down Syndrome. Current screening methods can only detect certain well-characterized genetic conditions, though. Prenatal full genome sequencing will allow screening for the other 5000+ Mendelian diseases, including those caused by spontaneous de novo mutations that can't be predicted by carrier screening.

Full genome sequencing isn't simply an improvement on current screening methods, though. I think it is a step change that will transform the way we screen for complications in pregnancy, and here's why:

It won’t always give a clear answer

Current screening is highly specific - a positive test for Down Syndrome implies a 99+% chance that the child will have the condition. Full genome sequencing will inevitably reveal uncertain DNA variants - some women will be told that their child might have a severe disease, or might not.

There are two sources of this uncertainty. The first is limits in our understanding of the effects of DNA variants. Since the variants that cause severe disease are so rare by definition, and the rules of genetics so complex, incredibly large training datasets will be required before we can accurately classify the variants that cause disease. Secondly, some DNA variants appear to be truly nondeterministic, or have incomplete penetrance.

It might tell us too much

Current screening only reveals data relevant to a disease diagnosis. If a woman has her future child’s full genome sequence to screen for disease, she could possibly learn much more:

It's important to note that genetic prediction of this sort is in its infancy - we don't know how accurate predictions will become (or which current predictions are actually true). But it's safe to assume that some cosmetic predictions will be available from the data that is generated for disease screening. The question is whether a pregnant woman will want to look.


There is so much to discuss about prenatal sequencing, including fascinating recent tech developments, but in this post I want to focus on downstream effects - why prenatal sequencing is such a big deal for both consumers and producers of genetics research.

Genomics will become more relevant

Since the human genome project, genomics[2] has been a largely academic discipline, dissociated from the broader biotech industry. There have been successful genetics companies - Counsyl, Invitae, Ancestry, 23andMe - but they use 10-year-old technologies. Announcements at the annual industry conference, ASHG, haven't moved markets or captured public attention like they do in some other disciplines. (VR, cryptocurrency, stem cell research and immunotherapy come to mind.)

However, prenatal sequencing will be a cause to follow genomics research more closely:

Genetics will play a role in more pregnancies

Currently, less than 1% of prenatal screening tests are positive. If tests are negative, a woman usually won't have to consider the possibility of genetic disease until after her child is born. When prenatal full genome sequencing is available, a woman that elects for the most sensitive disease screening - if she wants to know about any plausible disease-causing variant in her child - has a much higher likelihood of receiving a not-completely-negative test.

Pregnancy will have a complex ethical decision tree

Since prenatal sequencing will produce probabilistic test results, and the space of diseases tested will increase, some pregnant women will face gripping ethical questions about whether to terminate a pregnancy. This decision is hard enough when a doctor reveals there is a 99% chance that your child will have Down Syndrome. What if she instead says there is a 45% chance that your child will have autism?

This is not a hypothetical - these are the DNA variants that prenatal full genome sequencing will expose. Women will be forced to define boundaries, explicitly or implicitly, around when to terminate an otherwise healthy pregnancy because of a mysterious DNA variant. Other health-related ethical issues, such as end of life care, suggest that this will be extremely difficult.

Crossover into social and political issues

Genetic disease is already revealing kinks in the traditionally polarized pro choice / pro life debate. As I mentioned earlier, 72% of American pregnancies undergo prenatal screening, and 46% of Americans identify as pro life - yet an estimated 92% of positive tests are terminated [3]. While these numbers are not perfectly comparable, something doesn’t add up - there are clearly some pro life women that would terminate a pregnancy based on the results of a diagnostic screen. But, again: at what threshold?

DNA sequencing data itself could also become a political lever. Abortion foes might try to restrict the genetic information available to pregnant women, since it would likely reduce the number of abortions, with the counterargument that people should always the right to access their personal data. It could be an effective strategy. As 23andMe found out, access to information isn't obvious to regulatory bodies. Pigeonholing one’s political beliefs about reproductive choice will only get more difficult.

New demands and opportunities

Prenatal seqeuncing will also create a whole new space of opportunities for those currently working in genetics. I'll quickly survey some of the areas that could be affected:

Academic research

With the increased focus on disease prediction, the field of Mendelian genomics will basically be inverted. Today, there exists an entire research enterprise around diagnosing patients with rare genetic diseases. This is rapidly approaching a solved problem, but the inverse - predicting whether a DNA variant will cause disease without any prior phenotype information - is an extremely complex problem that could generate research questions for decades.

Health care delivery

Communicating genetic data in a clinical setting is a already a tremendous pain point for the entire industry, and demands on OB/GYNs and genetic counselors will only increase, as sequencing causes more women to receive more complex test results. Further, since test results may inform termination decisions, prenatal seqeuncing will place extreme pressure on the speed of returning results.

Diagnostic testing industry

Prenatal sequencing could lead to shifts, and possibly fractures, in the products of diagnostic labs. Currently, diagnostic labs provide both an assay and an interpretation - they tell a patient what DNA vairant she has and whether it causes disease. Since disease prediction from full genome sequencing will inevitably be subject to interpretation, assay and interpretation may be forced to decouple.

As possible future customer of these services (I'm 27), I'd pay a large sum for the full genome sequence of my future kid. But, I'd demand to look at the raw data myself, and I'd probably also pay third parties to provide second opinions on any uncertainties in the initial report.

Rare disease research

One area of personal importance is the possible impact of prenatal sequencing on rare disease research. Rare disease patients and families, and their associated advocacy groups, are an extremely important part of the medical research ecosystem. Realistically, prenatal sequencing will likely lead to a decrease in the incidence of genetic disease. If fewer children are diagnosed with genetic diseases, will there be less appetite for research funding and patient services? There is actually an argument to the contrary - that current services will be less strained and more research efforts can be directed to each patients - but this is not a given, and should be addressed a priori.

What’s Next?

I've outlined why I think prenatal sequencing is a really big deal, but it's frustratingly qualitative. How is one to determine how big the impacts of prenatal sequencing will be?

I want to end with a list of questions that could help "size the market" for prenatal sequencing. These are questions that we haven’t been able to answer to date, though we are excruciatingly close.

  • What % of people carry a variant that could responsibly be diagnosed as the cause of Mendelian disease?

    Some DNA variants are incompletely penetrant - they cause disease in some people but not others. This means there are healthy people walking around with DNA variants that could have caused disease - and which, in a prenatal sequencing world, may have been cause for terminating a pregnancy. How many of these people exist? This will help us understand how many pregnancies will have to consider genetics. Improving databases of disease variants and population frequencies will help narrow in on an estimate. This number will help us size the market for interpretation services -

  • What % of severe common diseases are "caused" by a single mendelian variant?

    In some more common diseases, a small fraction of disease incidence is caused by rare DNA variants. Autism, type 1 diabetes and many pediatric cancers are likely examples. If we sum all the severe diseases that could plausibly cause concern for a prospective mother, what cumulative percentage of cases are mostly the cause of rare DNA variants? This again isn't known, but genomics research - in this case the genetics of common disease - is providing inroads.

  • How is the incidence of disease-causing variants changing over time?

    Geneticists often assume that the mutation rate - the expected number of new (de novo) mutations a child is expected to have - is constant. Is that a safe assumption? I want to clarify that this is purely personal speculation, but I wonder if increasing paternal age and/or the environmental factors that are responsible for decreased sperm counts could be driving an increase in the effective mutation rate. If so, prenatal sequencing will be proportionally more important.

  • What is the limit for how accurately a face could be rendered from a genome?

    Could a mother be given a projection of what her child will look like at age 30? To date, deterimining faces from genomes has been unsuccessful, but theoretically, a genome could predict one's physical characteristics as well as a hypothetical identical twin.

    This is not significant in a medical context, but seeing a picture of your child has a tremendous emotional appeal - and ethical risks - that disease prediction does not.

  • Will IVF providers be allowed to provide clients with genome sequence data?

    This post has assumed that women will continue to get pregnant, without considering how. One important consideration is how IVF embraces full genome sequencing, as is already happening.

    Could a woman seriously generate 12 embryos and choose to carry the one that is most likely to be tallest? This would be the closest we've come to "designer baby" doomsday scenarios that have often accompanied interpretations of genetics in pop culture - and particuarly if combined with the above quesiton.


[1] To be clear, genomics research has been invaluable in translational research and the care of patients with diseases. It just hasn’t been visibly useful to the general population.

[2] The nomenclature of genetics and genomics is subtle and honestly kind of silly. Basically: genetics is the study of individual genes; genomics studies the genome globally. I've tried to stick to the technically correct usage here, but it's not important for this post. More here.

[3] I've found that many people are shocked when they first hear this number, as I was.