Chemistry is going the way of computing. It’s getting smaller and faster. High-throughput experimentation, or HTE, is part of this push. Borrowing from biologists and biochemists, HTE has brought in microplates and multichannel pipettes to miniaturize reactions, as well as robots to run those reactions rapidly without sacrificing precision. But it’s also been around for decades. So why are so many in the field excited about HTE right now? Stereo Chemistry looks at the technology and culture shift behind the current buzz.
The following is the script for the podcast. We have edited the interviews within for length and clarity.
Matt Davenport: Welcome to Stereo Chemistry, everyone. I am Matt Davenport, and today we are going to be talking about high-throughput experimentation, or HTE. HTE has been around for decades, but we’ve been hearing a lot of buzz about it now. So in this episode, we’re going to talk about why that is.
Before we dive in, though, I want to introduce my cohost, colleague, and fellow toddler parent, C&EN science reporter Sam Lemonick. Sam, how’s it going?
Sam Lemonick: Hey, Matt. It’s good.
Matt: So I have a weird question for you. I was just grabbing a snack because I didn’t want to get hungry during recording.
Matt: And I was curious, do you ever run into this thing where I don’t know if it’s a moral thing or I just feel guilty when I’m like, “We bought this food for our child, but I would like to eat it”?
Sam: Yes, I definitely think about that and feel guilty. What helps me is that I realized we buy a lot of stuff for him that’s healthy, like we have more fruit around the house now than we ever used to. And I should be eating more fruit. So then I don’t feel so bad about eating his fruit.
Matt: [Laughter] Very cool. Well, I’m glad we had this talk. That’s really all I needed for the podcast, so I will let you go.
Sam: All right. Thanks so much. Great to talk to you, Matt. Really. A pleasure being on the show.
Matt: But seriously, we’re here to talk about HTE, specifically in synthetic organic chemistry, which is not something either of us knew a ton about to begin with.
Sam: Right. You picked the wrong person.
Matt: I think this makes sense. I’ve heard that one of the reasons HTE is sort of having a moment right now is because of its accessibility. And I thought we could test that hypothesis by having two relatively uninitiated science writers talk about it.
And really, HTE itself isn’t that hard to wrap your brain around. The idea is instead of designing one experiment to test one hypothesis in a beaker or flask, you design many experiments to test multiple hypotheses in a multiwell plate or microplate. So these plates are most commonly made of plastic, and they’re about the size of a pack of index cards. So they can easily fit in your hand, and they’ve got a bunch of reservoirs in them, or wells. They can come with 6 wells, 24 wells, 96 wells . . .
Sam: Yeah, yeah. And they get bigger, too, right? I mean, aren’t there well plates with more than 1,000 wells in them?
Matt: So I actually saw a 1,536-well plate while I was reporting, and I was like, “Those wells are so little. How do you put stuff in there?” The short answer is pipettes and/or robots, but I want to put a pin in that for a minute, because I think there’s a good chance that at least some people are wondering, “Wait. Haven’t we been using well plates for a while?”
And the answer is yes. But it feels like something special is happening now. Let me take you back to when I first started hearing some buzz about HTE.
Matt: That was at the national meeting of the American Chemical Society last August in San Diego. As a reminder, ACS publishes C&EN, who makes this podcast.
And I had a chance to sit down with a chemist at a pharma company and ask, What are the stories I should be paying attention to? What aren’t you seeing in C&EN that you want to see? And they told me, “HTE.”
Now, fast-forward to January, and this tweet from Ash Jogalekar caught my eye. Ash is the well known author of the science blog The Curious Wavefunction, and he’s also a professional chemist.
Ash Jogalekar: I started out as an organic chemist. My PhD is in organic chemistry. But I switched to computational chemistry. I always joke, one of the motivating factors was when I damaged a $2,000 piece of equipment by working in the lab and my adviser said, “Maybe you’re not cut out for lab work. Maybe you should look into computational.”
Matt: Ash now works for a company called Strateos, which is headquartered in Silicon Valley. So it’s probably not surprising to hear it’s a tech company, but it’s a tech company developing things like artificial intelligence and lab automation for biology and drug discovery. Ash is their product manager for medicinal chemistry.
So I took notice of this tweet he sent that said, “I am rarely zealous about new technology, but I am willing to place my bets on high-throughput experimentation using small volumes and plate-based chemistry having a significant impact on medicinal and process chemistry during the next decade.”
I am rarely zealous about new technology, but I am willing to place my bets on high-throughput experimentation using small volumes and plate-based chemistry having a significant impact on medicinal and process chemistry during the next decade. https://t.co/aaAwiFcWjj— Ash Jogalekar (@curiouswavefn) January 17, 2020
Sam: I saw that tweet as well back in January.
Matt: And you weren’t the only one. For instance, an organic chemist who works at AstraZeneca responded to the tweet, “We’re getting into it for real.”
Sam: And my first reaction was actually surprise because I thought to myself, “Well, HTE has already been around, right?”
Matt: Totally. I was emailing with the editor of C&EN’s business team, Mike McCoy, about this. Mike’s been covering HTE since the early aughts. Or early 2000s. Or whatever we call it.
Anyhow, Mike told me he hasn’t written about it in a while because it’s pretty established in industry, and there are actually companies out there specializing in HTE equipment and services, such as Avantium, Unchained Labs, and the aptly named Hte, which celebrated its 20th anniversary last year.
To avoid confusion, from here on out, whenever we say HTE, we’ll be talking about high-throughput experimentation, not Hte the company.
Anyhow, the question still remains, Why is HTE—the field—worth talking about now?
Sam: Yes, have you learned why? Do you have an answer?
Matt: Yes. The best answer I’ve heard is that the field is at an inflection point. Put another way, this podcast is a coming-of-age story. With robots.
In this episode, we’re going to be talking about the tools and technology of HTE, including robots. Those tools are becoming more accessible, and they’re also getting better at doing more complex chemistry. Then you’ve also got more chemists who are getting better at working with these tools, trying to solve harder and harder problems, especially in process chemistry and medicinal chemistry.
And so there’s this confluence of things happening now that also makes it feel like there’s something special about HTE now. Getting back to Ash and his tweet, I wanted to check that feeling and ask, Does it feel like HTE is having a moment?
Ash Jogalekar: Absolutely. I mean, just the amount of interest that I’ve seen both in the literature and different companies that we have been talking to. It’s just amazing.
Matt (in interview): What’s sort of, like, the state of the art in chemistry, and, you know, what is it that is different from what you see in bio versus what we’re seeing in chemistry?
Ash Jogalekar: I think that’s a great question. And that’s something that sort of confuses me a bit, too, to be perfectly honest. Because it looks like biologists have been doing their version of HTE for at least 2 or 3 decades now, right? And in fact, ironically, one reason why chemists are getting excited about it is precisely because they can sort of repurpose a lot of the devices, the workflows, the techniques that biologists have used for HTE.
Matt (in studio): I wanted to follow up on that idea. Like, is HTE becoming more of a hot topic because the tools and techniques of these other fields are becoming more accessible, things like robots and automated workflows? What Ash told me was yes, the robots and automation are driving the buzz. But there’s also a cultural shift that’s part of it.
Ash Jogalekar: Most of HTE is done using plates. It’s plate based, just like biologists were doing. And actually this is a whole discussion in itself, if you look at the education of most chemists, synthetic organic chemists, you would be very hard pressed to find a synthetic organic chemist who has used plate-based chemistry in their graduate school work or postdoctoral work.
And so I think just making that switch from doing everything the way it has been done for, like, 150 years in beakers and vials and round-bottomed flasks to doing everything in plates. I think that itself is a sort of mental paradigm shift. That too. Even now, it’s not that common. But it just took a while for that to percolate down into the chemistry community.
Matt: So I think that’s really the big thing happening now, that more chemists are feeling more empowered to explore more of experimental space because of HTE. So you have more chemists stepping into this realm where you aren’t constrained to do one experiment at a time.
Sam: Which is the way a lot of synthetic organic chemistry has been done. I mean, even if you’re talking about a 24-well plate, running 24 experiments at once is 24 times more than running one experiment at a time.
Matt: And that really sets us up for this episode. We’re going to talk to chemists who have witnessed and felt these cultural and technical changes firsthand to understand how they’ve lined up to get us where we are today. Before we dive in though, Sam, is there anything more you want to know now to get your bearings?
Sam: Yeah, so we already talked about testing more hypotheses at once with HTE. But are there any other big-picture reasons people are excited about it?
Matt: Totally. Ash had a really succinct way of summing that up. So yes, HTE lets you test multiple ideas at once and because you’re running those tests in teensy little wells, you’re using less material. That means HTE can be less costly and more ecofriendly than more conventional chemistry.
Sam: It’s sort of like bringing this smaller, faster, cheaper mentality we’ve seen in tech into chemistry?
Ash Jogalekar: And so I think ultimately sort of the reason why I’m getting so excited about this is that it allows chemistry to go in the same direction that computing and electronics went in the latter half of the 20th century.
Sam: That makes sense. So, Matt, I did have one more baseline HTE question for you. What does a good high-throughput experimentation result look like?
Matt: I’m glad you asked. And for the answer, I want to introduce you to Melodie Christensen.
Melodie Christensen: Hello?
Matt (in interview): Hello?
Melodie Christensen: Hi.
Matt (in studio): Melodie splits her time between Merck and the University of British Columbia. She’s been at Merck for about 10 years, where she’s on the automation team, which makes new tools to help accelerate Merck’s research.
Now she’s also a PhD student at UBC in Jason Hein’s group developing nontraditional methods for experimentation. For example, one of her projects is looking at using machine-learning algorithms in automated reactions.
So the HTE result she told me about is actually part of her Merck work, in collaboration with Jack Twilton, who was in David MacMillan’s group at Princeton. The team published it early in 2018. It was a hot paper in the journal Angewandte Chemie.
They were working on this project trying to find a quick route to make benzylic alcohols, a useful class of molecules in pharma. As an example, this particular project led them to a three-step synthesis of Prozac.
So the specific reaction they were trying to optimize was adding an aryl group to certain types of C–H bonds.
Melodie Christensen: So one thing with these reactions are you have a competing side reaction. You also have the possibility of a competing C–O coupling reaction. The addition of a Lewis acid could kind of modulate and reduce that side reaction.
Matt: Melodie says they had a good candidate for that Lewis acid, zinc chloride. But it turned out that zinc bromide worked better in combination with a particular base and solvent, something she and Jack discovered because they were able to test multiple reaction conditions using HTE.
Melodie Christensen: He had said that his initial lead was zinc chloride and he probably never would have really looked at zinc bromide if he didn’t have the opportunity to evaluate that method using an array of experiments. Because that really enabled him to find that kind of magic combination.
Matt: And I think this is a good time to come back to that question we put a pin in at the top of episode. That is, How do you do these experiments in these miniaturized vessels, 96-well plates or 1,536-well plates? And with miniaturized amounts of reagents and solvents on top of that? This is where we’ve seen a lot of automation, by which I mean robots.
And as we mentioned earlier, Melodie works on automation, so we definitely got into robots in our interview. I actually asked her if she names her robots, either at Merck or at the Hein lab at UBC.
Melodie Christensen: I do.
OK, so in the Hein lab, all the instruments are actually named after dragons. So yeah, in the Hein lab there’s Mushu and Smaug. And in the Merck lab, I really love The Hitchhiker’s Guide to the Galaxy series. So I’ve named all my robots after characters from The Hitchhiker’s Guide to the Galaxy. So I have Trillian. I have Arthur Dent, and then I have Marvin.
Matt (in studio): Despite all her work with robots, though, she really stressed the point that automation can really be a tool to free up the creativity of the chemist. And this is something I wanted to talk about with you, Sam, because you have covered all sorts of emerging tech in the lab, including robotics and machine learning, for C&EN. I guess my question for you is, you’ve interviewed lots of scientists about these things in different areas of chemistry; how do you see them fitting into HTE?
Sam:Yeah. You know, once you take the human out of the equation, if you don’t have to rely on our human size and our human imprecision, shakiness, I mean just the way we move, you can make chemistry a lot smaller. And I know that’s not just true in the high-throughput world, but that’s really fascinating to think about.
And sometimes it’s not a matter of asking a robot to do something better than us or to do something we can’t do. Sometimes you want a robot to do something we don’t necessarily want to do. You know, like sampling every well in a 96-well plate every hour or something.
I mean, one of the cool things about high throughput is that you could have your robot working after 5:00 p.m. when the lights are out, all the way until morning and through the next day.
And, to Melodie’s point, machines are freeing up smart, capable chemists to think about the results and what they might want to do next or just dream up new molecules that they might want to make in the future. I hear that a lot about machine learning as well, this idea that you can actually enable chemists by helping or by replacing some part of their job that’s more drudgery than actual scientific thinking.
Matt: And that’s something that came up a lot in my interviews, like, don’t forget the role of the chemist in all of this. Melodie said that some of the biggest successes have come from a combination of the humans and machines working together. Like, you use the robots to handle the easy stuff, say dispensing Lewis acid. But then you want to try this really specialized procedure and there’s no robotic workflow for it. There’s still an opportunity for a crafty chemist to step in and set it up by hand in a high-throughput experiment using pipettes.
Melodie Christensen: And I think this really has enabled us to be able to carry out more complex and complicated procedures. And it’s really led to successes.
Sam: And then, right, the thinking is that one day, maybe robots will be able to do that thing that was complex and complicated today, getting us to a new level of creativity in the future.
Matt: Yes. And that actually leads us to another reason why chemists think HTE is having a moment now. Because we’re starting to live in that future. Let me explain. And to do that, let’s put robots off to the side for a second.
Sam: Do I have to?
Matt: Just for a second. So even without robots, synthetic organic chemists are going to take on more challenging and complicated problems with more challenging and complicated chemistries, right?
Matt: And those chemists aren’t necessarily thinking “How do I optimize this for a 96-well plate?”
Sam: Right. They just want to get it to work.
Matt: Right. Now, even though the chemists developing these new methods may not be thinking about HTE, there are chemists in the HTE space who are thinking about these new methods. So, one, this means we can start thinking about robots again.
Sam: It’s about time.
Matt: And two, there’s excitement around HTE because some folks see it becoming even more powerful as it incorporates these new methods.
Sam: Gotcha. Do you have any examples of what these new methods might be?
Matt: Yes. I was talking to three chemists on the HTE team at the biopharmaceutical company AbbVie about this. And the head of that team, Ying Wang, pointed to a couple examples. One was C–H activation, where you pop off a hydrogen from a specific carbon in a molecule and add a different group.
The other example was photoredox catalysis, where you use light to help form or break bonds. These are techniques that Ying says are still young and in the developmental stages, which means they’re not straightforward to use in a high-throughput setting. Yet. So bringing these sort of emerging chemistries to HTE is a challenge, but one that folks are excited to overcome, especially in the pharma industry, where you have established HTE programs.
Sam: Now, what about academics? Are they doing HTE?
Matt: Great question. And the answer is yes, they are. But I think it’s time we dug into the accessibility of HTE a little bit more.
I talked to several HTE evangelists who pointed out that all you really need to do HTE is a microplate, a pipette, and a mind-set to design multiple experiments to run at once. Microplates and pipettes are pretty standard lab equipment, so the barrier to try any sort of high-throughput experiment isn’t that high. But that barrier looks different depending on what type of chemistry you want to do. Do you remember Ash mentioned two specific areas of chemistry where he thought HTE would make an impact in his tweet from the beginning of the episode?
Sam: Yeah. Process chemistry and medicinal chemistry, right?
Matt: Right. So with HTE in process chemistry, Ash told me the goal is to optimize the hell out of an experiment. So you know what you want to make and you’re trying out different conditions in parallel to give you the best yield. And you want to do this so you can find a viable way to make your molecule at a commercial scale.
That means you’re working with the same concentrations of reagents that you would work with at the commercial scale. Even though you’re using much smaller volumes in HTE, you don’t necessarily need specialized equipment to run the experiments beyond what synthetic chemists already have in their lab, plus maybe a pipette and a microplate.
Sam: What’s different with medicinal chemistry?
Matt: In medicinal chemistry, chemists are trying to make new molecules, new drugs. HTE lets you cast a wide net to create many new molecules at once.
But you just need enough of a new molecule to test it, right? You don’t need these new molecules at a commercial scale in the discovery phase. So medicinal chemists are working with ever-shrinking concentrations of reagents to the point where they’re so low, they do need robots to handle dosing.
Sam: And not everyone is going to have access to those robots.
Matt: Right. Then, whatever type of HTE you’re doing, you’re going to need analytical tools to characterize your reactions and your products. If you need really sophisticated robots and analytical tools for your experiments, those are going to be expensive. And, as AbbVie’s Ying Wang told me, that cost is going to be a barrier, especially in academic labs.
Ying Wang: The barrier, really the challenge, is a capital budget, right? Like the instrumentation. Or how do we access the automation?
Matt: Still, some universities and academic groups are finding ways over those barriers. The next voice you’ll hear belongs to Amanda Dombrowski, who works on Ying Wang’s team at AbbVie.
Amanda Dombrowski: A lot of places, how they’re getting around the large cost is they don’t have individual groups buy them, but core facilities at a university will buy the analytical instruments and/or the more expensive automation in order for the whole department can use it.
Matt: For example, centers at the University of Pennsylvania and Princeton came up frequently while reporting this story. Merck is involved with both of those. Which illustrates another way around the cost barrier. Academic researchers are collaborating with pharma companies.
But another thing that’s happening now in this field is that people are developing ways to make it more accessible, beyond access to cutting-edge robots and analytical instruments.
Sam: How so?
Matt: Let’s look at AbbVie, for example. In March of last year, they published a paper about what they call ChemBeads.
Sam: Oh yeah. I feel like I remember reading about that in C&EN.
Matt: You did. Mark Peplow wrote a great story about them, which everyone listening should totally read after this podcast. But I’ll give you the essentials here.
ChemBeads are microscopic glass beads that AbbVie coats with solid reagents used in HTE. A catalyst, for example. And these ChemBeads can help out in a couple of different ways.
For one, they create a reliable way to deliver small masses of reagents. AbbVie’s process to coat the beads results in a pretty uniform distribution of reagent per bead. And because the beads are so tiny, it can be a really small mass, like 100 μg. Imagine trying to weigh out that amount over and over again. ChemBeads ensure that you deliver the right amount efficiently.
Secondly, whatever you coat the ChemBeads with, they are all still glass beads. That means whatever you coat them with, they’re going to look the same to a robot that’s dispensing them. You don’t have to worry about changing a robot’s settings or maybe using a different setup altogether to dispense a different solid. You just pick a bead with a different coating.
Sam: So do you have a sense for how widely they’re being used?
Matt: That’s a real good question that I don’t have a very concrete answer for. Noah Tu, who is the first author on the ChemBeads paper, told me that the beads are making their way to industrial and academic labs.
Ying also told me that after the group developed ChemBeads in 2016, they’ve seen an eightfold uptick in requests for them. And Amanda told me that AbbVie’s HTE team is definitely helping accelerate drug discovery, at least internally.
Amanda Dombrowski: Just kind of due to how discovery goes, it’s very fast paced, so many of the compounds would never have been made because they wouldn’t be able to find conditions in a short period of time to make them. So because we use HTE, we’re able to find the conditions for them, they can make the compounds, they can get tested. So we’ve seen that it’s really moved projects forward.
Matt: This reminded me of Ash’s tweet, where he said we’d see HTE make significant contributions to chemistry in the next 10 years. It felt like maybe HTE was already making those contributions at AbbVie. So I asked Amanda, Noah, and Ying if they thought it was, too.
Amanda Dombrowski: Yes. I would say so.
Noah Tu: Agreed.
Ying Wang: Yeah, I think it’s started to show a lot of impact internally. At the same time, I also think there’s a lot of room for further improvement.
Matt: After the break, we’re going to get into that and so much more. We’re going to head into an HTE lab, make you accessories to a robot heist, and maybe bring a branch of chemistry back from the dead. Stick around.
Matt: Oh, hey. So this is still Matt, and we’re going to get back to HTE in just a moment. But I wanted to take a minute to ask you for a favor. We’ve started looking for this year’s Talented 12, presented by Thermo Fisher Scientific, and we need your help.
What do we need? It’s pretty easy. We want you to nominate an early-career scientist who is shaping the future of chemistry. Each year, we feature 12 brilliant chemists working in academia, industry, and government to tackle the world’s biggest challenges.
I’ve been helping out with the Talented 12 for years, and I am so proud of the work we do and diversity of amazing chemists we’re able to showcase. And that all starts with you, the chemistry community, nominating the folks we need to know about.
If you know of someone who deserves to be featured on this year’s list, head on over to our nomination form at cenm.ag/t12nom, N-O-M. It’s really easy to fill out, and it takes only a few minutes. Again that link is cenm.ag/t12nom, and your deadline is April 6 at 11:59 p.m. eastern. Thank you so much for your help, and we can’t wait to meet the chemists you nominate.
Matt: And now, the sound of 96 little reactions stirring.
I recorded this at the University of Michigan, where I was visiting Tim Cernak to learn what HTE looks like—and sounds like—in person.
Timothy “Tim” Cernak: There’s tiny little stir bars in every well.
Matt (in interview): Are there really?
Tim Cernak: Yeah.
Matt: Oh my goodness. My first reaction when you said that was, “Awww.”
Tim Cernak: Oh, they’re cute. They’re super cute.
Sam: Welcome back to our HTE episode of Stereo Chemistry. It’s time for an HTE lab tour. Matt, since you actually got to go to the lab, do you want to take it from here?
Matt (in studio): Oh, sure. Thanks, Sam. Tim is a medicinal chemist at Michigan, but we actually spent most of our time talking about HTE in process chemistry.
Tim Cernak: Within pharmaceutical discovery, there’s the development phase where we do process chemistry and where you’re trying to get a series of known reactions to work really, really well because we’re going to make a drug on ton scale for commercialization and the difference between a 92% yield and a 98% yield is significant. And one can imagine how having these high-throughput tools can help you push that 92 up to 98%. In med chem, there’s two yields, I was told. There’s enough and not enough.
Matt: So Tim showed me around his lab, which occupies several rooms. And, honestly, most of them look like regular chemistry labs. But then you go through a door, and things start feeling different. I actually told Tim it looked like what Hollywood thinks chemistry looks like. It’s the best way I could think of to describe seeing robots in glove boxes.
Tim Cernak: And so a lot of the reactions that we run are very sensitive to air and moisture. So we need to protect them from that, and that’s what this trusty little glove box is for. And so inside you can see this, this robot that is here is called the Mosquito. It’s got, like, this like Gatling gun of tips. And so we load up those tips with the, we load up those tips with our reagents and then dose them out into the plates. And over here, this is another robot, called an OT-2, or an Opentrons.
Matt: The robots were pretty sleek, which makes sense, right. Well plates aren’t that big, and the robots have to fit inside glove boxes to begin with. I thought the Mosquito kind of looked like a table saw and the OT-2 reminded me of, like, a consumer 3-D printer. We’ve got pictures of the robots on our website.
And not far from the robots, there was also a fancy looking HPLC/mass spec instrument giving them readouts on a 384-well plate.
Matt (in interview): That’s the plate then, right?
Tim Cernak: Exactly. So you’re looking at 384 different samples. And the more yellow it is, the better the reaction worked.
Matt: But like I said, outside of these tools, the lab actually looked like a bunch of other labs I’ve seen before.
Tim Cernak: A lot of what we’re doing is very traditional chemistry, but we use those tools to generate lots of information, and then once we know what we want to chase after, we just jump into the fume hood and do everything the way that a traditional synthetic chemist would.
Matt: Tim expanded on that in his office, holding a 1,536-well plate in his hand.
Tim Cernak: We try very hard to have our miniaturized reactions look just like traditional experiments. I run reactions in flasks, and so to me, miniaturizing down to this is just like a tiny little flask to me. I’m a classic synthetic organic chemist, and I love to do things in a very traditional way.
Matt: Which is kind of surprising to hear when, you know, he’s holding that microplate and there’s robots just around the corner. But it also maybe explains why it took him a minute to understand how important that well plate could be. This happened when he was working as a medicinal chemist at Merck about 10 years ago. At the time, Merck was interested in miniaturizing chemistry and working with small volumes. But Tim says that he and his colleagues kept running into the problem of evaporation.
Tim Cernak: When my colleague gave me the 1,536-well plate, I was staring at it all morning. I was like, “Oh, this is the coolest thing.” I just showing all my friends. And I put 1 μL of water in a couple of the wells, and I had it in, like, a ziplock bag, and I put it in my pocket, and then I went through the rest of my day and I totally forgot it was there.
So fast-forward like 6 h, and I was in the East Village in New York City—I used to live there those days—and having a lovely snack at a café and I remembered this thing in my pocket. And I pulled it out, and the droplets were still there. Like 6 h later and 1 μL of water is just sitting there. I mean we had done tons of experiments where we’d put a μL droplet in all kinds of different vessels and it would just disappear instantly. I mean within a few minutes. But it dawned on me, it was like a real eureka moment for me, that the vessel itself was the solution.
Matt: So now, Tim’s team has this vessel that preserves chemical solutions so they don’t evaporate and enables massively parallel experimentation. But the only tool the researchers have to fill all those wells is a manual pipette. They knew there had to be a better way.
Tim Cernak: We would dose for like 12 h. Like our lab mates would, like, bring us food to make sure that we had enough calories to keep going. And then at the end of the day, you’d hold up the plate and then finally see what you had done. And everything was in the wrong well. You could just tell by the colors. And you’d be like, “Dammit.” So that’s why we drove down to another site and threw a robot into the back of a truck.
I mean you know the trick that we’ve done is we just stole a bunch of robots from biochemists, and we actually literally put a robot in the back of a truck and drove it to a glove box at Merck one day. And that launched a lot of what’s happening right now.
Sam: So they borrowed some robots from Merck biochemists? If that’s not interdisciplinary research, I don’t know what is.
Matt (in studio): For real. And that’s something worth discussing. In the first half of this episode, we touched on this idea that HTE borrowed a lot from other fields, and I want to dig into that a little now. In fact, while I was working on this story, our editor, Lauren Wolf, was like, “This kind of sounds like combinatorial chemistry. What makes HTE different?”
Are you familiar with combinatorial chemistry, Sam?
Sam: A little bit. But I don’t know if I’m confident enough to say yes on tape.
Matt: That’s totally fair. I wasn’t either until Lauren brought it up. So in the ’90s, chemists had devised synthetic methods to prepare lots and lots of new molecules, potential drug leads, quickly and in parallel.
This was understandably very exciting. But it turns out, there are a lot of possible molecules you can make. Something like 1060 molecules, Tim told me. And what I heard is that, as combinatorial chemistry started probing that vast space, we found we could make a lot of useless molecules. So this field that was at first very exciting in the ’90s started to look overhyped, and over the next couple decades, combinatorial chemistry fell out of favor. Which explains why folks like you and me don’t know much about it.
Sam: Right. But combinatorial chemistry does sound like HTE. So I agree with Lauren here.
Matt: So does Tim Cernak.
Tim Cernak: Yeah. You caught me. We’re just combi chemists.
When I started in chemistry, combi chem was already dead, so I don’t really know what it was. I was just told combi chem is dead and we’re not supposed to do that anymore. And so the idea was that you could make all the molecules, which is beautiful. I think the beauty of combi chem was that you could do a lot of reactions, and certainly the modern flavor of HTE or high-throughput experimentation is definitely kind of a phoenix rising from the ashes.
Matt: There was also somebody else who I wanted to talk to about this, M. G. Finn at Georgia Tech. He’s a chemist and the editor of the journal ACS Combinatorial Science. He’s also been on the podcast before, and I wanted to see what he thought about HTE and combinatorial chemistry.
M. G. Finn: In many ways, they mean the same thing, although in my view, and I think in a more holistic view, high-throughput experimentation is a tool or a set of tools within combinatorial science and of course can be applied very often to chemistry. And so it is a way to do combinatorial chemistry. Let’s say it that way.
Sam: Which basically agrees with what Tim said.
Matt: Yes. Except on one important point that I haven’t played for you yet.
M. G. Finn:I should say that combinatorial chemistry is far from dead, but it has been incorporated as a tool for doing things in drug discovery, in materials science, and a variety of other fields.
Matt: So high-throughput or combinatorial thinking is playing out in other areas of chemistry, too. Researchers have new techniques and technologies to take on the seemingly limitless possibilities of stuff you can make with chemistry. To that point, Sam, you actually wrote a story about metallic glasses a couple years ago where researchers used high-throughput methods to make those materials.
Sam: You know, when I was doing that story, the sense that I got from the researchers was that the world of possible materials is really vast. I think they were talking about three-element combinations and the number of possible three-element combinations is enormous. And so for them, high-throughput was a way to make it possible to search such a vast area of chemical space.
Matt: Before we disembark this train of thought, I did want to bring in one more synthetic organic chemist who does see one important difference between HTE and combinatorial chemistry. That’s Spencer Dreher, who works with HTE at Merck.
Spencer Dreher: Computation has really evolved in the 20, 30 years since combinatorial chemistry came around. In that case, you were just trying to make as many molecules as you could, and they often looked very similar. And, now, because of sort of the “design, make, test cycle,” this whole idea of computational design, we’re able to come up with libraries of compounds that will, you know, meet the design criteria that we have. So we’re not just making things randomly; we’re making specific targeted sets of compounds.
Matt: You heard him mention the “design, make, test cycle,” and that came up a lot during my reporting for this story. And it’s exactly what it sounds like: you design something, make it, and then test it. But it’s become something of a mantra in medicinal chemistry over the last 10–15 years.
And Spencer’s point is that we’re a lot better at designing molecules now thanks to improved computation, like he mentioned. And so Spencer says, even if you’re trying to make 96 or 384 compounds at a time, you’ve probably spent a good amount of effort designing each of those.
Spencer Dreher: Now, to be honest, if we’re making 1,536 you probably don’t have 1,536 hypotheses that you’re going after. Maybe you are actually doing more of a combinatorial approach sometimes.
Matt: What I found really interesting, though, was he was reluctant to say we’re more informed now.
Matt (in interview): I’m going to try something and repeat back what I would have taken away from this, and please feel free to correct me if I’m wrong. It’s more informed now than it ever was.
Spencer Dreher: I’m not sure if I would use the word informed. It’s an interesting time because we can run a lot of experiments. But to be honest, we haven’t quite gotten to the stage of chemistry informatics that we can learn from all of the things we’ve done in the past.
That’s definitely a goal that we have going forward is to have all these big data sets and be able to do sort of machine learning, artificial intelligence–type work with that. The word informed, to me, is . . . I think what we’re doing is we’re able to run enough experiments now that we, that we can solve the problems, but we don’t really understand it yet.
Matt: So this whole thing may seem like semantics, is it combi chem or is it not? Is it informed or is it not? But I think they’re important and interesting points, and they also show people being really measured in how they talk about HTE.
Sam: In contrast to the claims of combi chem being overhyped.
Matt: Exactly. Like, people are aware and vocal that HTE isn’t going to solve all the problems and that it has limitations and obstacles to clear. Like you just heard Spencer talk about data.
Sam:Do chemists know how to deal with all that data?
Matt: Data is a huge issue for high-throughput experimentation. And there’s no good answer for how to handle it all right now. But it’s a problem that everyone’s aware of, and there are some cool things in the works.
This is where talk of AI and machine learning comes in. Tim Cernak’s lab at the University of Michigan is doing this gnarly thing where they’re trying to build a virtual-reality platform that would essentially let people swim through data.
Matt: I don’t know if you’ve seen Minority Report.
Sam: Yes, obviously.
Matt:You know, you have Tom Cruise up there moving stuff around on the screen. That’s what I think they’re envisioning, except you have an Oculus Rift over your face.
Sam: Do you have to put swim goggles on, too, or is the Oculus enough?
Matt: I think the Oculus is enough. Yeah.
But yeah, data handling is one bottleneck. Another bottleneck that people pointed to is, especially on the discovery side of high-throughput experimentation, you’ll need a way to analyze very dilute solutions really quickly. So you need cutting-edge analytical chemistry techniques.
Sam: What kind of techniques are they using? I mean, can you do, like, NMR or something else?
Matt: So one Spencer mentioned was MALDI, which stands for matrix-assisted laser desorption ionization. This is a technique where you use a laser to liberate and ionize molecules from a sample before sending them to a mass spec. And MALDI mass spec isn’t new, but there are some ultra-high-throughput versions coming on line now.
Then there’s another ultra-high-throughput mass spec technique that uses sound waves to create a mist of samples in a well plate. This one is capable of handling three samples per second, according to a report last year in Analytical Chemistry published by folks at AstraZeneca.
And guess where I heard about that.
Matt: Ash linked the paper in the tweet that kicked this story off.
Sam: Oh man. So we’re back where we started?
Matt: How about that, right? But I hope you learned something on this merry-go-round.
Sam: Oh totally. Is there anything else we should share with folks before we sign off?
Matt: Yes. The credits.
No, I do have something real quick. I learned a ton reporting this episode and had a ton of fun doing it, and I’m sure there’s a ton more to learn. Please feel free to tweet at me at @MrMattDavenport if you’ve got anything you want to share about HTE or combinatorial chemistry or just other stuff you think we should be talking about. You can also email us at firstname.lastname@example.org.
And now for those credits. I produced this episode and wrote it with my cohost, Sam Lemonick.
Sam: I’m @SamLemonick on Twitter.
Matt: It was edited by Lauren Wolf and Amanda Yarnell. Our fabulous copyeditor is Sabrina Ashwell.
The music you heard in this episode was “Origami” by Tom Goldstein, “Mindplay” by Roza, “Cold” by Anthony Lazaro, and “I Don’t Know What’s Gonna Kill Me” by A.M. Beef.
Sam: Thanks for listening.