You previously worked as a software engineer for a large company in Silicon Valley. How did you decide to become a PhD student in biology?
I remember finally reaching that decision while I was reading this Richard Dawkins book, The Ancestor’s Tale, which is an exploration of the tree of life. It works from humans backwards, and tries to find our common ancestors with all the other organisms that exist. I had been enjoying the book and all the little vignettes about life, ignoring his occasional rants about why God is dead and all that. Eventually, he got to bacteria, and I started getting super amped about all the things that he was saying about bacteria and viruses and archaea and whatnot.
It got to the point where I would strike up conversations with random people on the bus about what I was reading, and it dawned on me that I was excited about biology in a way that was distinct from the way that I was excited about my job at the time. I had wound up in a discipline where I got to solve cool problems, but I wasn’t really engaged with the physical world in any meaningful way.
To sanity-check my desire to change careers, I did informational interviews with biologists to get a sense of the possible jobs that would let me touch this stuff in my day-to-day work, and what I would need to do in order to have my hand on the tiller.
One of the things that quickly became clear about the field of biology was that people typically had answers to basic questions, but that there were a ton of questions no one had answers to. When I’d ask a deeper follow-up question, they might have an answer. And then, when I’d go one level deeper, they’d be like, “Oh yeah, nobody knows that.” Eventually, it became clear that I really needed to have a PhD in order to have the conversations that I most wanted to have, so I ended up pursuing that.
It’s funny: when I was on something like the ninth iteration of my application essay, I was sitting on the phone with my mother in a cafe in the Mission trying to justify this transition from computer science and artificial intelligence into biology, and having a hard time coming up with a cohesive connection. But she was like, “This makes perfect sense. When you were ten, you were excited about nanotechnology and talked about how you were going to have tiny robots clean your teeth, and now you’re talking about bacteria in the same way.”
If I had to trace the common thread, it would be the idea of tons of tiny agents interacting at a scale where the physics that we’re used to doesn’t even apply — that’s a super exciting world to think about.
So now you’re a PhD student working in a research lab. What does your work there look like, generally speaking?
We study bacteria, so we wind up learning stuff that’s also relevant to people who want to prevent infectious diseases. That said, we’re not a pathogen lab; we’re not focused on preventing infectious diseases. We certainly collaborate with labs that are — the ones that are working on pharmaceuticals and antibiotics, or that study pathogens and try to understand their biology. We can say things like, “We’ve now identified collections of genes in bacteria that, when you remove them, kill the bacteria,” and then somebody more directly working with pathogens can say, “Okay, we know that we’ve got pharmaceuticals that remove this one gene function in bacteria, others that knock down this different gene function in bacteria, and they’ve both been approved by the FDA.”
When publishing research papers, my lab is always very interested in what we call “having biology in the paper.” I’m sure that sounds tautological and opaque if you’re outside the discipline, but it means if I build a new tool, I want to be able to show you something new about the way that the biological world works; I don’t want to just describe the new tool I built. We’re often trying to develop new technologies that let us make robust and refined measurements about bacteria, and then elucidate the function of the elaborate web of genes driving life at this micro scale. When we can do that, that’s a great outcome.
It’s hard to understand what bacteria are doing and how they’re doing it just by looking at them. You can see these little pill-shaped things, and if we mess with them we can sometimes see them die in kind of interesting ways. But at the level at which we can watch bacteria, trying to make conclusions about them would be like trying to evaluate the well-being of France by counting the number of national monuments that were built over time in Paris — at this level you have no visibility into the lives of individual people or what the main commodities are or things like that.
And so rather than directly observing bacteria, a lot of what we’re doing is trying to create ways to perturb them, and then measure and record their lives in large datasets of observations. We want this data to be sufficiently nuanced that we can get a richer understanding of what is going on when we then later analyze it. That then provides tools for people to study a wide range of bacteria — pathogenic or otherwise. People use bacteria for all kinds of purposes, so for example many of the techniques we develop can be used to figure out better ways to get bacteria to produce stuff, whether that’s food products like alcohol and yogurt, energy products like methanol and ethanol, or enzymes like the ones in your cleaning detergents.
But for us, mostly we just care about how bacteria do bacteria, because it’s fundamentally fascinating that these tiny things are able to survive the vicissitudes of the world around them, where a slight shift in the salt concentration in water is this massive life or death situation that happens over the span of, like, three seconds. It’d be like your entire neighborhood being instantly flooded. Bacteria just have a program to deal with that. They don’t have a computer, they don’t have a brain — they’re smaller than one brain cell or a single transistor, and yet they’ve somehow got a plan of action to deal with that scenario in real time.
I find that fascinating, and I think that’s what brings a lot of people into the kinds of disciplines where I reside. And the same thing is true of the individual cells in the human body, and a lot of people who are “working on a cure for cancer” are mostly just fascinated by how in the hell these elaborate tools called proteins are being assembled by the endoplasmic reticulum and the Golgi bodies and whatever. There is this enormously complicated thing that’s happening a gajillion times per second in your body.
The way you get money from the National Institutes of Health (NIH) is by saying “and if we figure it out we’ll be closer to curing cancer” — and don’t get me wrong, the fraction of people who actually care about that in the biology community, including my lab, is non-trivial. A lot of people went into biology to like, I don’t know, avenge their grandfather or something. But for me, in the war between bacteria and humans, I’m probably gonna side with the bacteria. I’m not really in it for human health. It’s great if I can find funding by helping to facilitate that, but I’m mostly just curious how bacteria are doing their thing.
In terms of the research you’re doing now, what is the best-case outcome of what you’re working on?
My lab is very focused on basic science, which is to say that the goal is explicitly open-ended — it’s not like we either find the answer to a specific question or bust. We have a pretty good idea of where there are interesting answers to important questions, and so we come up with new ways to shine light into those dark corners, and then hope we find cool stuff. But we aren’t too dead-set on what that stuff is. Indeed, the story that we’re telling about my PhD work has changed pretty dramatically just in the last year, and I don’t think that’s atypical.
The way the course of research in a lab pivots is a bit different from how a startup might pivot, however. If a startup pivots, they are changing their plan from one thing to another. Whereas in research, it’s more like realizing you could do something that wasn’t possible before, or realizing that a question we had abandoned in the past turns out to be relevant to a number of other things. When those discoveries happen, you refocus in a new direction and see where it takes you. You need to be agile to find the best outcomes.
So I think the “best case” depends a lot on where the light winds up pointing us.
You Can’t Come in, Clive
The main biological thing I’ve been hearing about recently is the advances in our ability to edit genetic code using CRISPR, which has a lot of hype surrounding it. Is there hype among biologists about this in the same way? Does it feel like a real thing with potential, or is that far off?
CRISPR is actually just an acronym for “Clustered Regularly Interspersed Short Palindromic Repeats.” What the acronym means in and of itself isn’t exciting; it’s basically just a placeholder for a certain pattern we see in DNA.
We’ve sequenced DNA from lots of people, but even more bacteria and viruses. We don’t know what most of it does, but identifying genes is way easier than figuring out what they’re doing. It’s fairly easy to identify the boundaries of genes themselves, and we’re generally able to identify the basic structure of an organism’s genome, especially in prokaryotes like bacteria that don’t have more complicated genetic structures like introns. I mean, there’s still a fair amount of complexity in bacterial genomes with things like intergenic DNA and transcription regulators, but it’s nowhere near as complicated as eukaryotic genes in organisms like humans. Almost everything in a bacteria’s DNA is devoted to something that will actually be turned into a protein.
But within bacterial genomes, scientists would find these patterns where it wasn’t clear what they were doing. This pattern was given the CRISPR acronym, which basically says, “We see this banding pattern in the DNA where there’s a repeated palindromic sequence, and then it has a bunch of other stuff in between the repeated palindromic chunks, but that’s all we know right now.” Over time, people put together that the stuff in between the palindromic chunks seemed to match DNA that had been found in viruses — kind of like a virus’s fingerprint, wrapped up in the bacteria’s DNA.
How did they figure that out?
The breakthrough for understanding this came from yogurt manufacturing. Danisco is the company that makes Dannon Yogurt, and a big part of their business is to culture bacteria in giant vats in order to make their yogurt. But their bacteria can get sick: there are viruses called phage that infect them. If a phage infects a bacterial strain that’s in one giant monocultural vat, all the bacteria in that vat are getting sick. They’re all fucked, and you just have to nuke the whole thing. All the money you’ve put into that batch is just gone.
So Danisco was like: how can we better identify when this is going to happen, and reduce the frequency of these infections? Because we would like to not lose millions of dollars worth of dairy product whenever our vats get infected. They had a research division that was studying the yogurt bacteria’s CRISPR pattern to try to better understand how and when bacteria get sick. They saw viral DNA in this particular CRISPR pattern and decided to dig deeper.
It’s been a long time since I looked at the original papers, but they basically did experiments where they showed conclusively that CRISPR was involved with an antiviral protective mechanism. When the researchers infected the bacteria with viruses, the ones that successfully fought off the virus were modified genetically in such a way that their CRISPR pattern then contained clear genetic references to the virus that they just fought off. Furthermore, bacteria without the protective mechanism were vastly easier to infect with viruses. What they found was not just a generic protective mechanism — it was a reactive, adaptive immune system in bacteria.
How does that mechanism work, exactly? How does the bacteria fight off the virus?
Let’s say you go to your favorite nightclub. The bouncer in the back room has a wall of Polaroids with people’s faces on them. When he gets out front and sees that jackass Clive trying to come back into the club again, he remembers Clive’s face from the wall of Polaroids and keeps him out.
The CRISPR pattern is basically like a wall of Polaroids. It turns out that there are also all these genes that are adjacent to the pattern that are called “CRISPR-associated” or “Cas” proteins that vary in their details. There are actually lots of different CRISPR systems and patterns with different associated proteins; it’s an immensely elaborate network that is used by bacteria against viruses, by viruses against bacteria, by bacteria against themselves. It’s an incredibly complicated ecosystem, but fundamentally the idea is this: you have a bacteria’s immune system standing there acting as a bouncer to suppress the production of some gene, either in the same organism or a different organism, and they’re doing it based on this wall of Polaroids that they scanned when they started their shift. When a new viral infection happens, a lot of bacteria will just die. But the ones that manage to survive get a snapshot of the thing that almost killed them, add it to the wall, and the next time around when that guy shows up, they can be like, “No Clive, you can’t come in.”
So what practical applications does this mechanism have?
People realized early on that they could apply this in many different ways. It’s very common in molecular biology, particularly for people who study viruses and bacteria, to make connections between learning something new about what bacteria are doing and developing a new tool that we can apply to other areas of biology. For example, useful biochemical research techniques like making use of restriction enzymes and polymerase chain reaction (PCR) and ligases… all these technologies ultimately come from various viruses or transposons or bacteria, where we find a thing that’s doing something unusual like surviving in a circumstance we wouldn’t expect them to. We can trace that to some weird molecular thing that’s happening that we can then replicate.
When it comes to CRISPR, some researchers like Jennifer Doudna at UC Berkeley understood the import early on — there’s a lot of dispute over who got there first, and who should get the Nobel or the patent or whatever — because the researchers saw, hey, we could use CRISPR in combination with a protein in the system that acts like a pair of scissors, which would allow us to use the combined system to start cutting DNA in extremely specific places.
How would that work?
Going back to the bouncer analogy, if I want to change the behavior of the system, I can add a new Polaroid to the wall without having to kill the bouncer or replace him. I just have to sneak in and add the new Polaroid, and now he’s not going to let Jenny into the nightclub, even though Jenny has never done anything wrong. I’m just interested in seeing what happens when she’s no longer allowed into the club. In the same way, with CRISPR gene editing, you can make targeted changes to extremely specific places in DNA with very little effort.
This has greatly reduced the amount of money and effort that goes into modifying genes or knocking down the genetic expression of a virus or protein or something. We can now make edits that target arbitrary proteins, and proteins are responsible for most of the physical action inside a cell. So that’s really broadly exciting to biologists, and it’s transforming everything at the bench.
Where it gets public buzz is that it’s also potentially exciting in terms of biomedical applications. I think the average person wouldn’t be excited about most of the things the biology researchers are excited about. There’s huge buzz in the news, and there’s huge buzz in the labs, but they are about different things. They overlap because there are researchers who do care about the fact that you might eventually be able to use CRISPR to cut genes that would disable cancer without killing the person in which the cancer exists. Or you know, God help you, try to make babies smarter.
The thing that I find most exciting and most plausible in the relatively short term is fixing coherently understood genetic diseases — being able to fix a genetic disease where I know that my partner and I both have the same predisposition for a genetic disease, but if we had a functional copy of this one gene we’d be fine. And so if we can modify embryos in vitro using CRISPR gene-editing mechanisms, that’s a potentially exciting way to cure diseases before they start. There are people I care about for whom this would be a huge deal.
Some diseases you can cure by having a finite number of cells that do the right thing, even if they’re surrounded by cells that are doing the wrong thing. For example, if you’ve lost the ability to produce white blood cells due to any number of diseases, and I’m able to fix some of your bone marrow, or if I can give you insulin-producing genes inside your pancreas that is otherwise totally dysfunctional, then I might be able to make your life much better.
I think the media hype comes from this idea that we can probably now make modifications to an individual cell before it becomes a human being, and soon — or maybe already — we can modify cells or push pre-modified cells into an existing human being. That has the potential to fix problems that were previously unaddressable.
But that’s still really, really hard to do, and it’s not very illuminating relative to the real scope of what CRISPR can do. In basic science applications like the ones that I’m developing as part of my PhD, the excitement is more around using this system to engender outcomes that weren’t previously possible in the lab. It doesn’t just make science easier; it makes it profoundly easier, such that whole new categories of things are now possible.
Playing for the Lakers
Working for large software companies and working in academia have very different reputations around working conditions and generally how work gets done. Did you experience some culture shock in that transition?
Software companies and academia are different in a lot of ways. My particular program is kind of a weird home-for-lost-toys kind of a program, insofar as they were actually trying to recruit people who were not biologists by training. So I didn’t feel as on the outside by virtue of not having a biology background as I expected to.
In terms of the actual disciplines being different, there is a big difference between engineering and science in terms of the way that people talk about problems. That’s been the hardest bridge to cross.
One thing that I’ve noticed is that when confronted with a problem, I’ll try to bound the problem. My thought process is: okay, there’s this problem space we are trying to work in, so let’s figure out the best case and the worst case boundaries, and then shave away at them based on facts that we know until we get down to the space in which things could actually exist. Then we can do experiments or develop algorithms to refine that in the direction we want to go — knowing that we’re working towards this boundary where we can’t do any better, or this boundary where things can’t get any worse.
That feels to me like a very reasonable way to go about things. But, working within a scientific discipline, I find that when I start to frame the worst-case scenario, people will say, “What are you talking about? Things are not that bad. You have to calm down.” And I’ll start framing the best-case scenario and people are like, “Well that’s insane, and it’s a cute thought, but things are never going to go that well.” I’m still trying to figure out how to communicate in words that don’t send people into a panic thinking that I am totally untethered from reality.
That’s been the most jarring shift — not necessarily the biggest shift, but the most jarring shift. Some shifts were expected. A lot of the work that I do is still data analysis, partly because of my background and what I gravitate towards, and partly because that’s the direction the field’s going. There’s still a lot of time sitting in front of a computer: coding scripts to plan complicated protocols, writing a paper, generating figures or other visualizations, analyzing datasets.
But in biology there’s also a significant fraction of your time spent standing up in front of a flat surface with a bunch of — you know, it looks like what you imagine science looks like. You’ve got beakers of clear liquid that all look the same, and probably smell the same, and have totally different stuff in them, and you’re trying to keep track of them and put them through very different chemical processes, and track a bunch of different readouts, and use ultra-precise measuring tools. It’s like super high-precision baking or something. So, as you can imagine, someone who works at Tartine has a different experience from someone who works at Facebook. And if you’re in a biology lab, you’re moving back and forth between those things, often in the same day.
So that that was hugely different and I saw that coming, but it was still terrifying. When I was working in a lab over the summer before my program started, I turned to the guy I was working with and I was like, “Here’s the deal: I do not know how to do any of this. I would love to learn about the science you’re doing and contribute meaningfully, but mostly I need you to train me to do the most basic shit that you learned to do in Biology 101.”
On top of the way people work together, and the basics of the work itself, I imagine there was a big financial transition as well.
Being a grad student is a fundamentally shitty situation to be in, both financially and existentially. If you are in a PhD program you are almost by definition working on things that nobody else cares about — except, if you’re lucky, your advisor. But let’s say you’re working on something that lots of people do care about. In the best case, you’re in a high pressure race to be the first one to do it and you’re terrified every day that someone else is going to scoop you. Meaning your work becomes irrelevant because someone else got to it before you, thanks, try again.
And that’s horrible. I think the fact that you’re barely getting paid anything is pretty minor compared to that. We live in the Bay Area, which is awful no matter what you do — you have to be rich to be poor in this town. Living on a grad student income has been painful for me, and I had a bunch of savings because I was extremely fortunate to be in the tech sector during part of the boom, enough savings that I could buffer my passage through grad school. Even so, I’m running out of money now, so it would be good to get back to a real salary — I don’t know how the hell these kids who I’ve been going to school with who came straight out of college do it. It’s a difficult existence.
What are your job prospects once you come out of a biology program?
Here’s a roundabout answer to that:
The sense I get is that sometime in the 1980s or 1990s, the federal government put a whole bunch of money into biology. Someone decided that we needed more expertise in biology to come to the United States, and the way to do that was to double the funding to the National Institutes of Health — and they did that overnight. The result was this surge of universities creating aspirational biology programs, quickly building buildings, and hiring people to fill previously non-existent departments.
Initially, that was a huge shot in the arm for the discipline. But at some point, there weren’t enough places to install new faculty, even if you had people who were competent enough to fill the positions. We ran out of the NIH having extra money around to create new biology institutions, so we ended up with fewer biology faculty positions than postdocs, and fewer postdoc positions then graduate students, and so forth.
There’s basically an infinite amount of good biology to be done, but the discipline in its current form can only support so many biologists. There are something like five times as many people who graduate with PhDs in biology per year as there are total faculty positions, much less positions that are open right now. So becoming faculty at the kind of institution where you’re likely to get much NIH funding and be able to do cool research and live in an area that you want to live in, even after having studied at one of those top-tier research institutions, is like being in middle school and deciding you’re going to play basketball for the Lakers. You’re going to have to be really, really good to play for the Lakers, whereas most other people in your position will end up playing basketball in high school and that’s the end of the story.
The aspirational thing on the academic side is do a postdoc, where you continue to be an underpaid, over-educated researcher doing the actual job of research for another two to five years. That’s not an entirely raw deal because it comes with a lot of freedom. But beyond that, academia is a really tough row to hoe. You’re still trying to play for the Lakers. I’m at a sufficiently high-profile institution that there are people in my lab for whom it’s not unrealistic to think that they might actually become faculty. It’s not what I want for myself, or something I think I would be likely to achieve if I did want it, but it is something that people in my lab are actively pursuing.
For many people, the main takeaway from doing a biology PhD is just that you got to be involved in science for some period of time; you do some really cool research and then you go do something else totally unrelated to biology.
The other option that allows you to stay in the field is to go into industry. I’m just starting to get a sense of what kinds of jobs are available, but they range from working for Big Pharma and established biotech companies like Genentech, to smaller startups, to founding your own thing and looking for VC funding directly. The work you’re doing still looks like biology at those places.
There are also hybrid, semi-academic research labs, like the Broad Institute and their ilk. A number of places in the Bay Area are pushing in that direction. They often center around cutting-edge technologies, where there’s value in having a lot of people do this stuff in a more stable way than academia affords. Because whereas in academia you need to constantly be doing the new thing, and in business you need to constantly be meeting the bottom line, these hybrid labs let you work on a well-established research area without being beholden to the bottom line.
These kinds of institutes can develop expertise and partnerships in a way that still feels academically rigorous, but isn’t tied to whether we make quota next month, and at the same time isn’t dependent on the novelty of the work. They just need to incrementally improve so that the labs or businesses they partner with and sell services to can achieve a much higher throughput, and do so much more cheaply than they could otherwise.
They can also provide an avenue for private-sector companies to come in and say, “Hey, we want to learn how to do this technique that was published two years ago. We’re just reading the paper and to reinvent it from scratch would be really challenging, and there’s not an option to send someone to go work in a research lab at MIT to learn it, so can we hire you to help us implement it?”
So these independent labs can form these elaborate partnerships, and some pretty exciting work happens in them. In terms of salary, they’re never going to be fully competitive with Genentech or something, but they can often thread the needle in terms of exciting work and mostly competitive pay.
Then there are many largely tangentially disciplines you could pursue as well, like public policy or various forms of intellectual property law.
Within academic research labs, how does the flow of money work? Where does funding come from, and who decides how it is spent?
I’m in an institution that is very heavily NIH-funded so I may have a skewed perspective on this, but my sense is that a lot of the funding for biology research in the country comes from the NIH. That shapes how people pitch their projects and what kinds of projects get funded.
Mechanically, the way it typically works is that there are grants — and I admit that I have only the loosest understanding of how those grants work — and there are different tiers of grants. I think the most fundamental one is called an R01.
If you are a principal investigator (PI) of a lab at a major institution, you submit an application making a case for why the work you’re doing is important, you’re the right person to do it, the place you’re doing that work is the right place for it to be done, and you’d be doing even more good things if you had more money. As part of your application, you also have to lay out your lab costs: do you need a giant, thirty-thousand-dollar centrifuge, or a PCR machine, or to buy time at a sequencing facility? And then you also have to account for the cost of the people that you have working in the lab.
If you’re awarded a grant, then the NIH effectively earmarks this big lump sum for you. They dole it out in increments and, depending on the grant, it can last several years before you have to renew. But it’s much easier to maintain and renew an R01 than it is to get a new one. That’s part of why it’s hard to make it in academic biology: not only is there a finite amount of money, but if you’re a new researcher trying to get your hands on an R01, it’s this zero-sum game where incumbents have a leg up.
What is the breakdown and hierarchy of the roles in a lab like that?
Fundamentally, there’s a PI and then there’s everybody else. After the PI the hierarchy is pretty flat, but the PI has absolute authority over the lab; it’s basically a dictatorship. The NIH gives them money, and it’s like a VC giving money to a CEO: if you’re a researcher in the lab, you’re an employee.
As a PI, you basically get a grant based on what you said you were going to do, or had already done at the start of the grant. Then you give updates over the course of the grant, saying that you’re doing useful stuff. It doesn’t have to be the same stuff you said you’re going to do, as long as it’s useful. And then when you need to apply to re-up the grant four years down the road, there’s a renegotiation and it’s important that you’ve done impactful scientific things in the interim. In theory, it’s possible to lose your grant at that time, but in between those goalposts you’re otherwise pretty safe. If you were completely delinquent, they might strip your grant midstream, but it’s really unlikely. It’s like how you may not get elected for a second term as president, but you’re less likely to get outright impeached.
Postdocs are senior to graduate students, insofar as postdocs already have their PhDs. Often, they are pursuing and acquiring their own sources of funding through lower-level grants that the NIH gives out to support nascent researchers to foster the next generation.
Graduate students are also pursuing different sources of money. The more money you bring in, the more independence you have because, at some level, the PI can tell me what to do with their money, whereas if I have my own money, I can spend it the way I want. But most of the money is coming from the top-level grants and from the PI.
In terms of autonomy and responsibilities within the lab, generally speaking, postdocs have more experience and are planning more of their own projects. They have a lot more flexibility to pursue their own ideas. The graduate students are beholden to their thesis committee, and also to what the lab can support. So if I have a fundamentally new idea as a graduate student and I can get my PI on board, that’s fine, but it’s going to be hard for me to get permission to access the slush funds to explore unilaterally.
There’s another level below graduate students — well, parallel to or below depending on the culture of the lab — which is technicians. Their fundamental job is to analyze the science you’ve done, or perform an established scientific protocol at the bench. You know, turn the crank. If we need to process plasmid DNA extracted from bacteria four hundred times over the next three weeks, that’s the technician’s job, so they just do a lot of that. It’s not specifically engaged with the intellectual pursuit portion of the work. In a lab that doesn’t have a lot of money, that may be a graduate student or postdoc’s job. But if you can afford a tech, you’d rather have them doing that work.
You also have people who are in a nebulous region above postdocs: research scientists who are not PIs, but who help do research. They’re not a canonical part of the lab. But there are labs, for example, where both members of a romantic partnership are scientists, and rather than founding two separate labs, they just decide to work together — so one of them becomes a researcher in the lab, getting paid by the research grants of their partner who is the PI. I know someone like this who is an extremely senior researcher who probably could have founded their own lab, but they’d rather work with their partner than prioritize that prestige. I get the appeal.
How Not to Wipe Out the Human Race
What is the conversation in your field around the ethics of what you’re doing? Is that something that explicitly comes up?
It varies based on what you’re doing. There are some areas where, in order to understand what’s happening, you have to work with humans. The biology of Bacillus subtilis and the biology of humans are extremely different, so I can do basic science on bacteria all I want, but we don’t really know how this thing works in a human being until we’ve done it in human beings.
The cost of actually doing that is very high, in every possible sense. So people slowly work their way up, starting with experiments on eukaryotic models like yeast. Then if it looks promising, you move to mice, and then if it looks promising with mice, you try monkeys, and then you do stage one clinical trials with humans. All of this is far outside my ken, but the point is that the more you move up this ladder, the more you have to justify that what you’re doing will ultimately have positive health outcomes for human beings.
If I’m doing studies on mice, that usually means I have to kill them after a month. There are whole review boards just making sure that if we’re doing something harmful, it’s within reason. As soon as you climb north of yeast, there are more controls — reviews to make sure you have thought about what you’re doing and convinced a panel of ethical experts that it’s responsible, and is aimed at the right ends. There are medical ethicists, bioethicists, and, if you want money from the NIH, federal review boards that evaluate your research before you can get funding.
Even working with bacteria, there’s still a certain amount of oversight, but it has less to do with ethics. If the things I wanted to do to bacteria were motivated purely by the desire to see them suffer, nobody would need to thwart me by founding People for the Ethical Treatment of Bacteria — I’d just never get funding, because who the hell would pay for that?
So ethical oversight is part of receiving funding.
Yeah, it’s part of receiving funding and continuing to get funding and just having permission to do what you’re doing.
Safety is the bigger issue for us when it comes to working with bacteria. It’s not like anyone cares about how badly you treat them — our worry is that the bacteria might actually come kill us. So as long as you’re doing a thorough job of killing them, review boards are fine with it. It’s not about how badly they suffer on the way down the sink; it’s about how much bleach did you pour down the sink with them.
All of those things involve different kinds of ethical questions and safety criteria, and there are good and bad systems in place. It’s very good that we have protections for safety and for the ethical treatment of organisms, but there’s a lot of random hit-or-miss stuff that winds up percolating through OSHA (the Occupational Safety and Health Administration).
For example, there are chemicals that intercalate into DNA. When you’re trying to identify things about DNA, often you want to “stain” it; you want to put something in it that becomes visible when you illuminate it with fluorescent light. And so you use these intercalating chemicals that get woven into the bands of the DNA, like hairs threaded through a comb. That’s great, but it turns out that the presence of those chemicals makes the process of replicating the DNA less robust; as the DNA replicate themselves, errors get introduced. These cells that have random mutations might be cells in your body. Mostly, DNA errors just make cells less functional and they die, but occasionally these “mutations” interact with the cells in a way that makes them cancerous.
So scientists have identified that sometimes intercalating chemicals can be carcinogenic under certain circumstances, though we don’t understand that very well. In the few cases where we’ve identified that a chemical might be a carcinogen, we’ve created these massive, Old Testament-style fences around it, like, “Thou shalt not come within a nine mile radius of anything labeled ethidium bromide, one of those fluorescent tagging chemicals, unless X and Y and Z and W conditions are met in the lab.”
And then there are something like nine other chemicals like SYBR Gold and SYBR Safe, which are brand names for things that perform the exact same function as ethidium bromide, and I can keep those in a little cardboard box in my desk drawer, and nobody gives a shit because it’s not ethidium bromide. Meanwhile, we’re giving ethidium bromide to cows as an antibiotic because it’s more harmful to bacteria than it is to eukaryotes. I could probably go drink it and be fine.
So I have to jump through these elaborate hoops to use this thing that probably isn’t going to kill me, whereas there’s almost no regulation at all of this other thing that’s probably just as toxic. It’s really arbitrary.
To take another example, there are probably ways in which, as an ethical researcher, you might be personally inclined to treat mice well even though there’s no actual stipulation for how you make their lives more or less comfortable in the lab — and then other ways in which their lives are incredibly finely regulated, and the mouse can’t actually tell the difference between a Level Eight and a Level Nine mattress in terms of how comfortably they are sleeping, but you need to always pick one or the other because there’s a regulation about mouse mattresses.
I would guess that there’s probably some deep bureaucratic misfirings that go into how this stuff gets regulated, because it’s hard for laws to keep up with the pace of what the science is doing. It’s this constantly moving target.
Regulations aside, how do scientists themselves talk about these issues?
While I’m sure there are callous assholes out there, for the most part biologists care about doing things in a way that is ethical, and healthy, and lets you sleep at night — even if it’s just about cover-your-ass self-preservation. In my work with bacteria, theoretically we might do the wrong things and produce organisms that are harmful to humans, but we care a lot about not screwing up. I didn’t sign up for this to wipe out the human race.
Scientists are particularly careful when technologies become clearly relevant to human health, like with CRISPR. There are several consortia, globally and nationally, where scientists are sitting down and saying, “Hey, we need to think about this, and here are our current thoughts. We need very clear boundaries and even moratoria in some cases to forestall negative outcomes.”
The unfortunate reality is that those initiatives are like test ban treaties, in the sense that they work insofar as people are paying attention to you, but you don’t have supreme dictatorial control over the world. Plus, there are guidelines, but the interpretation and implementation of those guidelines isn’t uniform. And in a cutting-edge field of research like CRISPR, there are people out on the fringes in countries with less oversight who think, “That’s cool, but if I’m the first one to do this research, I will be taken seriously. So I can either follow your regulations and nobody will ever care about the work I’m doing, or do this thing right now, in which case you guys will pay attention to me.” So there’s an obvious incentive to behave badly.
In your experience, how do these conversations around ethics or oversight differ in the private sector?
The two worlds differ substantially when it comes to the role of ethics in funding. When you’re trying to get VC funding, questions about ethics and safety are probably not the first ones that come up. But if you’re trying to get money from the NIH, you’ve had to address ethics in a form as part of the application process, and they’re going to keep checking in on it.
In tech, a company’s vibe early on might be, “Let’s just make a cool product for people.” Then, as soon as you have accounts and authentication and enough money flowing through your system that there can be meaningful fraud, you start worrying about privacy and security. When you’re starting to worry about those problems, you’re already succeeding.
In biology, you have to worry about those problems before you even have a chance to scratch the surface of your research, because you’re often trying to work on some disease that’s killing people. We should worry deeply about whether our work is effective. If it’s not, the best-worst case is that we’re wasting money and implicitly killing people who have this rare illness because we’re not finding a cure fast enough. The worst-worst case is that we’re creating a thing that will actively kill people.
The way you get funding for VC-driven startups isn’t necessarily by being super reliable and thoughtful and ethical. And I don’t think there’s a better example than Theranos. What Theranos was trying to do was at least noble in spirit. The people who funded them were a bunch of Silicon Valley VC firms, based on Beltway DC people vouching for them. Actual biologists were like, “That doesn’t make sense.” But it didn’t matter.
I’m not saying all biotech startups are terrible — I think there’s a lot of great work being done — but it’s very hard to evaluate right now. I won’t name names, but I can think of a couple of other companies where I’m like, well, that’s either gonna be a scandal or a very quiet fizzle sometime soon. I’ve had good friends go to work for companies who are like, “What the fuck are we doing in my company? We’ve literally been injecting mice with saline solution and then looking at RNA sequencing to see what’s happening. As far as I can tell, we are doing a deep dive on the placebo effect.”
I don’t know what’s happening there. That’s not even a joke. That’s a real example from people I know.