Product safety is critical in the chemical and pharmaceutical industries. Everyone needs to trust the medication they take, the food they eat, and the personal care products they use. Unfortunately, this has required the use of animal testing, which has been a controversial subject for decades.

Thanks to scientific advancements, the age of animal testing may soon be in the past. Episode 2 of the Analytical Wavelength explores the role of chemical prediction technology in reducing the need for animal testing. We will talk to experts in the science of animal testing, and the use of QSAR in product safety.

Read the full transcript

Chemical Predictions and Animal Testing – How Computers Saved the Bunnies

Jarrod Bailey   00:00

The same is true when it comes to testing, whether it’s chemicals or pharmaceuticals. Animals are not sufficiently predictive of humans with regards to toxicity and to safety. You can never make an animal that you’re testing human relevant enough.

Charis Lam   00:28

Product safety for pharmaceutical and chemical products is critical. We need to be able to trust that our medications foods, cosmetics and soaps aren’t going to harm us or the environment.

Jesse Harris   00:40

The science of product testing and toxicology led to massive increases in product safety over the course of the 20th century.

Charis Lam   00:48

Unfortunately, much of that testing was done on animals, which has been a controversial topic for decades.

Jesse Harris   00:55

In the 21st century, we may be able to advance past animal testing with the help of computer modeling and advanced cell-based methods.

Charis Lam   01:03

Hi, I’m Charis.

Jesse Harris   01:06

And I’m Jesse, and we’re the host of the Analytical Wavelength, a podcast brought to you by ACB Labs.

Charis Lam   01:12

In this episode, we’re going to be talking about the role of chemical predictions in toxicology and product testing.

Jesse Harris   01:18

First, we’re going to be talking to Dr. Jarrod Bailey, director of science at Animal Free Research UK. He’s going to explain the scientific argument against animal-based testing.

So hello! Today I have Jarrod Bailey here with me. He is the director of science for Animal Free Research UK, and he did his Ph.D. in genetics at Newcastle University. Is that correct?

Jarrod Bailey   01:47

That is correct. Yes. Newcastle in the far north of England.

Jesse Harris   01:50

Excellent. I wasn’t aware, my geography of the UK is actually surprisingly bad. I should probably learn a little bit more about that at some point.

Jarrod Bailey   01:58

Well, Newcastle seems to be most famous for its beer, for Newcastle Brown Ale, which arguably is bigger in North America than it is in England these days. But yeah, it’s famous for its beer and for a pretty bad football team, but it’s a nice place to be.

Jesse Harris   02:16

Okay, glad to hear that. Well, thank you so much for joining us today. Our first question, of course, is what is your favorite chemical?

Jarrod Bailey   02:27

I gave this a little bit of thought and kept coming up with chemicals that probably aren’t particularly exciting or are many people’s favorite chemicals. But I’m going to go for this one. One of my passions since I’ve been at school, which is quite a long time ago now, is skiing. I love the mountains and I just love skiing. So it’s actually probably something as simple as paraffin wax. This is something that you can put onto your skis and go out and just have a fabulous day in the mountain sliding around on the snow. So a pretty simple answer, but one that that’s brought me a lot of joy over the years.

Jesse Harris   03:02

Yes, that’s a good one. That’s a good one. Lots of applications for chemicals in our life. So I want to talk a little bit about your background and kind of how you came into the field of animal free research. Can you tell us a little bit about how you ended up there?

Jarrod Bailey   03:17

Yes, of course. And well, my degree was in genetics, my PhD was in virus genetics, which of course is very topical at the minute with COVID. I then spent seven years as a researcher researching premature births in humans. Why do women have premature babies? And of course, this has lifelong consequences for millions of children and people every year. It’s a pretty serious issue, and during that time I realized my own moral issues with animal experimentation. I didn’t think it was right to cause other beings (who happen not to be human) pain and suffering for science.

That was augmented by a scientific argument that I was realizing, that people were using animals to study the same thing I was studying, using human tissue and human cell culture. We were getting different results that conflicted with one another. And I started to think, maybe this isn’t the best thing we could be doing scientifically either. And of course, if it’s not the best thing we can be doing scientifically, there are human ethical issues because we can be doing better things, more human relevant things, more humane things. And if we’re not delivering cures and treatments, understanding of human diseases and biology, then there’s a human ethical angle too.

Cut a very long story short, I spent one lunchtime in the lab writing to, I think, about 35 different organizations that campaigned on this issue to try and get a change in how science is done. And for the last 17 years, I’ve worked for a number of organizations, both in North America and in the UK and Europe as a whole, underpinning the scientific argument to this, asking scientific questions: “Are animals good models for human beings, both in research and in the testing of chemicals for chemical safety and new drugs for efficacy and safety?”, publishing lots of papers, book chapters, making the case to governments and inquiries to the FDA, the EPA, the European Commission, and many, many more.

And that’s how I got into this.

And very briefly, currently with Animal Free Research UK, I started work with them in July last year, so not too long ago. And as well as continuing to make the scientific case to try and encourage a shift from animal science to human-specific science through public affairs programs, through political avenues too. We fund a lot of human-specific research; we funded more than £10 million worth of science in the UK, more than 260 projects. I think the last year for which figures are available, about three quarters of £1,000,000 were given to scientists to do human-specific animal-free research, and to educate early career scientists that this is the future of science and to excite them about that and to get them into that way of thinking.

Just to finish off, I firmly believe, as more and more scientists seem to, that we need to maintain a human perspective from start to finish, whether it’s in research, disease research, basic research, or whether it’s in testing of chemicals and new drugs for safety. There are too many species differences between one species and another. There are many differences in the same species between individuals to worry about. But we need, from start to finish, to maintain a human perspective to do the best science. And I’ve been convinced of that for a long time.

Jesse Harris   06:47

So let’s get right into that then, the meat of it. So what would you say is your argument on that? Because I think that there are many people who have opinions on the ethics of animal research just from the perspective of animal welfare. And people would probably have made decisions or have thoughts on that. But I think that this idea that there’s a scientific argument for not using animal models might be new to some people.

Jarrod Bailey   07:13

Yeah. I mean, obviously for the purposes of our discussion, I guess it’s going to be more interesting for the listeners to hear about testing, testing of chemicals and drugs. But it’s important to remember all the way through that this also applies to research. Are we really understanding, for example, many different types of cancers and how cancers metastasize and spread? Are we really getting to the bottom of Alzheimer’s disease and Parkinson’s disease and many, many others? Never mind finding treatments and potentially cures for them. I think the answer is no, we’re really, really not.

We’ve been led down too many dead ends because of the use of animal models that are of poor human relevance. And the reason for that poor human relevance comes down to genetics. We are genetically too different. We, of course, share many of our genes; but even when we share genes between rats, mice, dogs, monkeys, cats, humans, everything else, those genes are expressed differently. They’re doing different jobs in different ways and that really, really, really matters. Those gene expression differences have been linked to all sorts of biological processes and diseases.

The same is true when it comes to testing, whether it’s chemicals or pharmaceuticals. We need to know how a drug or chemical is absorbed, distributed around the body, metabolized, excreted, and so on and so forth, and all of those things differ, often significantly, between different species, in between different individuals of the same species.

So the point is that we know that animals are not predictive, sufficiently predictive, of humans with regards to toxicity and to safety. We know why; it’s because many of the drugs that are involved in those processes—absorption, distribution, metabolism, excretion—those genes are different. And so you can never make an animal that you’re testing (a new drug or a new chemical on) human relevant enough. We really have to maintain a human perspective.

So it’s actually a relatively simple principle scientifically and there are two things that come from that—one is, should we be using animals at all? Because they’re really not terribly predictive. And the second thing is, if we don’t use animals, what should we be doing instead? And we can talk about this; there are some amazing scientific technologies that would have seemed like science fiction to me 20 years ago, when I was in the lab. They are enabling us to do this research and this testing in a human context, and it’s much, much more predictive for what is going to happen to an actual human being. And if you like, I can tell you about some of the evidence my own work generated to show that testing new drugs on animals is not adding statistical evidential weight to the likelihood of toxicity or safety and what that means and why we should go from that.

Jesse Harris   10:11

Yeah, I definitely want to hear more about these alternatives. You know, it’s something that I imagine that the science has been moving very quickly between the advances in the cell cultural methods and then also in silico methods that’s been changing a lot recently. So I would love to hear kind of where things are at and how these compare to animal testing models.

Jarrod Bailey   10:32

Yes. So what’s happening? I think it’s important to say from the outset that, you know, maybe 20 years ago or so this was I guess, a relatively niche point of view. There weren’t too many scientists shouting about this and demanding change. And that’s really, really changed. And now this is a mainstream issue. It’s something that’s being discussed and debated, research written about in mainstream scientific journals and conferences.

There is an appreciation of a need to move to more human-specific research and testing within the regulatory agencies, the Food and Drug Administration, the Environmental Protection Agency in Europe, to the European Parliament recently made a resolution to phase out animal testing because there was a scientific need to do so.

So what do we do instead? What is giving science and scientists the confidence that we can do this paradigm shift, if you like, that we can move to better science? And I think it largely pivots around two or three things. One is advanced cell and tissue cultures, so-called organoids—the culturing of mini human organs, if you like, they’re the size of a full stop or a fly’s eye, they’re three dimensional. They are composed of many different cell types. In other words, they have real structural and functional physiological relevance to a real human organ.

There’s organ-on-a-chip and body-on-a-chip technology. So for those of your listeners who might not know what these are, they are very small chips. They’re made of glass or silicon. They used to be the size of a smartphone, they’re even smaller now in many instances. And they, again, they have these advanced three-dimensional cell and tissue cultures, human specific, with circulatory systems that simulate blood flow. If you’re looking at lung tissue, they have air flow that mimics breathing. If you’re looking at a kind of mini heart cultures, they can physically beat. They’re really, really mimicking what human organs, tissues are doing and how they interact with one another.

You can factor in the liver and the metabolism of new drugs and chemicals. So this is really, really exciting. And they’re proving their worth in many areas of research, for example, neurodegenerative diseases, like Alzheimer’s and Parkinson’s that I’ve already mentioned, and also in the testing of new drugs and chemicals. So there have been a couple of papers recently that have really demonstrated the superiority of these methods. One of the biggest issues is drug induced liver injury or chemicals, including pharmaceuticals, often if there’s an issue, they damage the liver. That’s where they go, that’s where they metabolize, the metabolites damage the liver. Now, that traditionally hasn’t been predicted well by the use of animal tests, and experiments have shown that these liver-on-a-chip technologies are vastly superior to animal methods. I think there was a paper published recently by a company called Emulate that were one of the pioneers in these organ-on-a-chip and body-on-a-chip technologies, that showed their chips could predict liver toxicity with 87% sensitivity and 100% specificity. This is just on another level compared to what any batch of animal tests could do.

And finally, this is now feeding into artificial intelligence, too. There are AI companies that are getting tens of millions of dollars of investment venture, capital investment, because their technology is so promising. So a lot of these data can feed into AI and, allied with computers, they can predict even better what real human beings are going to do, how they’re going to respond to this particular chemical or this particular drug.

One final point, these kinds of technologies have also been used in repurposing existing drugs for the battle against COVID with great, great success. So it seems clearer than ever, objectively and among the scientific community and increasing numbers, huge numbers of scientists, that this is the future of science and that there’s a real need for the benefit of humans, not just the 200 million animals a year used around the world in science. There’s a real need for this shift to a more a more human focus. And everyone’s going to benefit from that.

Jesse Harris   14:59

Yes, it sounds that way. So with that, I think that you could, if we’re talking specifically on AI as one of these frontiers of animal free testing, and I’d be interested to hear what else is there on the near horizon, of either like what are the challenges that you think might still be out there that need to be conquered? Or what are the technologies that are, you know, maybe not quite ready for primetime, but you’re excited about, to see how it performs?

Jarrod Bailey   15:26

Yeah, I think it’s important to say that we’re really a lot further on than many people have previously thought. We’re already at a point where we can do much, much better than animal-based approaches.

Another critical point is that we can constantly improve human cell and tissue cultures, organoids, organ- and body-on-a-chip technologies; constantly improve computational methods. That’s not something you can do with animals—animals are animals. You can only genetically modify them to be a little bit more humanlike to a certain degree. And there’s always a huge ethical issue with that.

So we’re already in a better place. We’re already in a place where, you know, these technologies that can replace animals are hugely advanced and can be improved even further. And there’s another thing that feeds into a shift towards human-specific technologies, which is that you can factor in human variation and diversity. And you can do that by taking samples of blood or skin, for example, possibly even urine from people, and harvesting cells, harvesting cells from those samples, coaxing those cells to go back in time to something called stem cells.

And if any of your listeners might not know what they are; when we were developing embryos, we were composed of cells that could turn into any of the specialized cells in our body. A lot of the different specialized cells could stem from those cells. You can coax developed cells in adult humans to think that they’re stem cells again, and then coax them to be any type of cell that you want to investigate.

Now, that’s important because if you if you can take cells from healthy individuals, very healthy individuals, particular patients with particular diseases, with specific different kinds of genetic causes, then you can model all of those different human individuals using the technologies I’ve spoken about. So not only are you being human-specific, you’re factoring in human variation and specific genetic causes of human diseases that you can compare directly with people who don’t have those diseases.

So it’s a tremendously exciting time. And yeah, I think this is why when I speak to early career scientists, young undergraduates, they are arguably more excited about a career in science than I’ve ever seen in the last few decades. And so they should be.

Jesse Harris   17:53

Yeah, that sounds also particularly interesting, but the implications are something like personalized medicine of being able to do with your specific testing on this and see how it would fare for a particular individual.

What about resources for people who want to learn more about this? Because this is, obviously, we’re doing a survey of the field here. I’m sure that you could talk more about any one of these technologies or any one of these points. So what resources would you recommend for people who want to go deeper into the science of these questions?

Jarrod Bailey   18:19

Yeah, I guess it stands to reason what I’m going to say, social media is a huge goldmine of information. Many of these biotech companies who are developing these technologies, many scientists and institutions who are using these technologies and generating amazing data. So get on social media, get on their websites and have a look at our website,

See what sort of science we’re funding, see what research that we’re pointing towards, that we find is amazing. Have a look on scientific repositories like PubMed, do PubMed searches. Find out what, where organoids and organ-on-a-chip technologies are being used, what sort of data they are generating, and how excited scientists are about the data that are coming out with these technologies.

So there’s all sorts of resources, largely Internet based. But you could spend years looking through the information already and it’s tremendously exciting.

Jesse Harris   19:18

And I’m sure that there are people who do! Spend lots of their time looking, but that’ll be great for this as an introduction to that. Thank you so much for your time. And it was lovely chatting with you.

Jarrod Bailey   19:30

It is a pleasure. Thank you for the opportunity.

Charis Lam   19:32

Jarrod offered a fascinating take on the scientific challenges about animal testing.

Jesse Harris   19:37

And it isn’t a fringe view. In fact, companies like Unilever have made it clear that they feel animal testing is unnecessary.

Charis Lam   19:46

Well, not everyone may recognize the name Unilever. You’ll certainly recognize some of their brands Dove Soap, Ben and Jerry’s Ice Cream and Axe Body Spray, to name a few.

Jesse Harris   19:56

Dr. Steve Gutsell is a team leader and computational scientist at Unilever. We talked to him about the role of computational chemistry in product testing.

Charis Lam   20:06

Here with us today is Steve Gutsell, who is a computational science and team leader at Unilever in the UK. Steve worked for 18 years with Unilever, and he received his Ph.D. in organic chemistry from Swansea University. Hello, Steve.

Steve Gutsell   20:22

Hi. Nice to talk to you today.

Charis Lam   20:23

It’s great to have you. So let’s start today with our usual icebreaker question. What is your favorite chemical?

Steve Gutsell   20:28

Yeah, I gave this one some thought, and it’s a little bit cheesy. It’s a little bit corporate, but forgive me. So there’s a class of surfactants—the alkyl acetylates. I’ve spent a long time in my career working on surfactants, studying their properties. They’re a tricky bunch of chemicals, but we’ve got some really cool models now to predict how these molecules interact with biological membranes and cause toxicity. So the science has really moved on even within my career. So these alkyl acetylates are a series of chemicals that I’ve kind of looked at and studied throughout my career, and they just happen to be the active ingredient in one of our leading products—the Dove bar. It’s probably one of our oldest products, but, you know, if it’s not broken, don’t fix it. It’s a great product. So yeah, alkyl acetylates for me have been good fun over the years.

Jesse Harris   21:25

That’s really interesting to hear. And I imagine that even this idea of picking compounds just because they’re new versus, you know, whether you’re going to things that are old and still work, is probably something that comes up in your work a lot. But let’s get into that. Can you tell us a little bit about how you got into computational toxicology?

Steve Gutsell   21:41

Sure. So my PhD, as you mentioned in Swansea, it was actually organic chemistry, so I was based in the lab. So it was actually part-funded by Unilever and it was linked to collagen. And so even from my postgraduate studies, I was studying how can you link the properties of chemicals to some kind of toxic effects.

In this case, it was skin sensitization—will a chemical result in an allergic reaction in people? And the fascinating thing to me, that kind of crystallized during my Ph.D., was that all of that information is wrapped up in the chemical somehow; in the chemical structure. So if you understand enough of the chemistry, and pretty clear if you understand the biology, the toxicology side as well, you can actually start to relate the two things together.

So what I did was, I was measuring stuff in the lab. I was measuring kinetics—rate constants. So, small organic molecules reacting with model nucleophile; so a model of a protein sidechain. That’s the fundamental step in causing skin allergy. And what I found out was that, yeah, you can fairly easily measure these things in the lab, and it can inform you about the biological properties.

So not only could you predict whether something would cause dermatitis allergy or not. The very cool thing was by understanding the magnitude, in this case, the rate constants, you could do quantitative predictions. And the link to Unilever really came into it. So I came and did placements at the lab where I am now in southeast of the UK, and the guy that I worked with was a computational chemist.

So like all computational chemists, there’s a streak of arrogance there and he’s like, well, you’re measuring all this stuff in the lab, and that’s really difficult. Surely, we can do this in the computer. So he taught me how to do some really cool quantum chemical calculations and semi-imperial calculations, molecular orbital energies, effectively recreating what I’ve done in the lab in the computer, which had I known that three years earlier, maybe my Ph.D. would have taken a slightly different course. But that was the hook for me, that working out that not only could you measure these things in the lab and relate them, you could then recreate those experiments effectively in the computer and build these predictive models.

And then it was kind of, bit of serendipity, right place, right time, that particular scientist moved on from Unilever, creating, a vacancy. And I thought, well, why not? Why not me? Why not put my CV in? I remember the hiring manager saying, well, this is a computational position. You’re not really a computational chemist, but we think you could be. We’ve invested in you, and we want to continue that. So there’s a combination of things, but that real insight gained through my Ph.D. I think is what hooked me in.

Charis Lam   24:30

That’s great. Going into the computational side; so far my understanding is a core part of that is quantitative structure activity relationships, or QSARs. Can you give us a bit of an introduction of how QSAR works in predictive toxicology?

Steve Gutsell   24:47

Sure. As the name suggests, what you’re trying to do is relate the structure of a molecule to some kind of activity. And in our case, that’s biological activity, and it’s something related to the toxicology. So really what you need to build a QSAR is three things. You need a set of chemicals with measurements of that property that you want to predict. So let’s use the skin allergy is an example, as I already mentioned it. So what you would need is a set of chemicals that you already know the answer for. You already know whether there’s going to be a skin allergen or not. But it can equally be a different property. It could be physicochemical properties, it could be environmental properties, anything that you want to predict, something that’s already been measured.

The second thing that you need is a way to describe your chemicals, and you need to be able to do that in numbers, because what you’re trying to do is build a predictive model within a computer and computers, for all their benefits, they don’t understand chemistry. So you have to introduce the chemicals into a model using numbers. So what we do is we basically translate chemical structures into descriptors. Now, they could be from the lab. So I mentioned in my previous work, I measured things in the lab, I measured rate constants. An experimental rate constant is a very good descriptor of the chemical in terms of how quickly it reacts with something.

But of course, you can calculate these things, and the advantage with the calculations is you can calculate many, many more of these, and you can even calculate for chemicals that you don’t physically have in the lab. So now it’s possible to calculate thousands of discrepancies on thousands of chemicals at the touch of a button. So they’re your two bits of information that effectively allow you to produce a data matrix.

So you have property that you want to predict and a whole bunch of numbers describing the chemicals. And then you need some math, you need a statistical method to go look for comparisons, because what you’re trying to do is show that there’s a relationship between some of those properties of the chemical that you either measured or calculated, and the property that you want to predict.

So do chemical kinetics relate to skin allergy (in my case)? And the general rule is with mathematics, to keep it as simple as possible. If you can draw a straight line between the points, you’re in a really good place, that’s often not the case, and you do have to go to more and more sophisticated methods. The QSARs can start from literally a simple linear regression to show that your x variable is minimally related to the y variable. So as reactivity increases, allergy increases, as an example. Or they can be extremely complicated, using lots of descriptors and something like neural networks or random forests, more machine learning, AI techniques, and it’s important to consider really when you’re using these things, what are you using them for? Because it may be perfectly acceptable to have a very complicated model with really complicated AI, but if you have to explain the predictions, and you need some transparency in that model, that may not be the best route.

So for example, if you’re predicting toxicity and you’re trying to show that you don’t need to do a traditional animal test, for example, those models may not be the best ones. It may be better to scale back and use something far more simple that you can actually explain, talk through, rationalize. But if you’re in the business of coming up with a new lead drug, for example, and you’re going to screen tens of thousands of structures using more of a black box model, that’s very, very quick, it’s absolutely acceptable because you’re going to follow that up with subsequent testing capacity.

Jesse Harris   28:31

Fascinating, there’s a lot to it. I’m actually really curious about what this looks like practically from what you’re describing. It definitely doesn’t sound like you will just walk up to a computer and throw up a structure, press enter and spit out an answer. Let’s say that somebody comes to you with something like this, like surfactant molecules, a new version or something like this, what does that kind of look like practically for either you or your team to then turn that into this passes, or this fails, or these are the problems you should be concerned about.

Steve Gutsell   28:56

So first thing, especially with surfactants, is to characterize that the heck out of it, because as I mentioned, surfactants, particularly commercial grade surfactants, are quite complex mixtures. So they’re complicated chemicals, and then they’re complicated mixtures of individual chemicals. And it really is important to understand that because, for example, if you take the chain length of the surfactant, you can vary from say a C8 to a C18–20, the properties of those two different ends of the spectrum are very, very different.

So understanding the relative proportions of those chain lengths, for example, will be critical to understanding the properties of that particular surfactant. So that’s the first thing, to really understand what it is that you’re dealing with and also accepting that there’s a bit of variability involved as well. So it’s again, the nature of commercial-grade substances versus kind of pure lab-grade samples; so understanding what the variability will be over the lifetime of the product.

Then we would look at its fundamental physical chemical properties. So things like its partition coefficient, something really simple like the optimal water partition coefficient is a really good surrogate for lots of things in biology, partitioning into membranes between environmental compartments, those kinds of things. For surfactants, particularly, octanol water is not the best system. Surfactants, by their nature, like to sit on surfaces or interfaces between things, and they will quickly turn a practical octanol water experiment into a foaming mess.

And so some of the cool models I mentioned at the start, it’s now possible to measure specifically partitioning between water and a biological membrane. You can do this on HPLC columns. It’s really neat stuff. You can also do it in the computer. You can model a phospholipid bilayer, and the interaction of surfactants with that. And that interaction with membranes is a critical determinant in how the toxicity of surfactants manifests itself in the environment. Surfactants are produced in really large amounts; generally during use they end up going down the drain. And whilst the vast majority of what goes down the drain biodegrades, so effectively gets digested by bugs and even anaerobic degradation. Some may end up into the aquatic environment and we need to understand whether that’s going to have an impact. Understanding how things interact with membranes in aquatic organisms is the fundamental mechanism behind the QSARs, and actually the membrane water partitioning models are a much better surrogate than the octanol water partition coefficients.

So another thing that we would investigate is something called the molecular initiating event. This is the concept of the first thing that a chemical does when it interacts with a biological system that could lead to some toxicology. So my skin allergy example is chemical reactivity; could it react with a protein sidechain? But also you’d look at, could it bind to receptors or interact with enzymes? And these types of models are all very tractable from a computational chemistry point of view.

So we’ve worked closely actually with some academic groups, Cambridge University in the UK, to build a suite. We now have models for over a hundred protein receptors, and we can screen new chemicals against those, and it gives an early indication of this chemical could bind this particular receptor, which could be linked to the particular toxicological outcomes. So, characterize it, study the physicochemical properties and the impact those could have, and then into things directly related to toxicity—molecular initiating events, or direct predictions of toxic effects.

Charis Lam   32:45

Mm hmm. You touched upon this briefly when you were talking about the aquatic toxicity and environmental effects, but we were interested in the role that computational chemistry plays in areas other than what people would normally think about, which I think is human toxicity. So are there other areas and what is the role played there?

Steve Gutsell   33:05

Yeah. So environmental safety is a big role for computational chemistry, in there as well, both on the exposure side, but also on the effects side. If you think about risk in terms of toxicity, or decay, toxicity is a function of two things. It’s a function of the hasn’t—so can the chemical cause a nasty effect, whether that’s in humans, that analogy, or whatever happens to be, or in an aquatic species, it tends to be more crude measures than that. And the other part of the equation is the exposure—how much of the chemical will the organism in the environment actually be exposed to? And the properties of the chemical impact on both of those; so that physicochemical properties impact where a chemical will end up in the environment—will it partition into the air, stay in the water, could it biodegrade—all of these things have an impact.

And in terms of the effects, the hazards, QSARs for predicting toxicity to aquatic species were actually some of the first to be developed. And they’ve got a history of around 100 years. The US EPA, in particular, have been big advocates of the use of QSAR to predict, for example, fish toxicity. So a huge role for computational chemistry in the environment.

Jesse Harris   34:26

Now regulatory requirements obviously play a really big role in how all of this works, how toxicology is studied and assessed. How have you seen changes on the regulatory side affect your work throughout the time that you’ve been working at Unilever? You’ve been there now for 18 years, so I imagine there’s been some regulatory changes over that time.

Steve Gutsell   34:49

Yeah, absolutely. So some regimes are very progressive and lead the way. So people like the US EPA, Health Canada, Environment and Climate Change Canada, are very early adopters of computational methods and indeed, they’re developers of these methods. A lot of the QSARs for fish toxicity that I mentioned originated from the US EPA. Other regimes are willing to talk and they are willing to change, so we, as Unilever, we actively go out and advocate for these changes—moving away from animal testing to the alternatives. And if you take somewhere like China, where the regulations can be quite strict, the Chinese regulators are actually willing to talk to industry, to learn, and to adapt, which is really refreshing. And then at the other end of the spectrum, there are some regulators who are a little bit stuck in the Dark Ages. There have been changes, but it’s really variable depending on the geography and the sector.

Charis Lam   35:52

So you touched upon Unilever’s advocacy in the area of moving away from animal testing, can you explain a bit more what your company’s position is on that?

Steve Gutsell   36:00

Yeah, it’s really simple. We believe that there are alternative, better ways of ensuring that people in the environment are safe from the use of chemicals without using lots and lots of animal tests.

The science has moved on a huge amount, so not just the chemistry and the computational methods, but the in-vitro world, the systems biology, the machine learning, it’s all so much further on than it was when I started, and it continues to evolve. And this is why, as Unilever, we go out and we share our science, we publish, and we present our results all the time.

We also collaborate with lots of different groups around the world to try to advance the science. So that’s essentially wherever the best science is, it’s happening. So academics, industry, NGOs, and regulators. In the past, we’ve seen that regulations traditionally have lagged behind the latest science. So scientific developments, maybe that you can do something in a new way, and it takes a number of years for the regulations to catch up. But we feel that that trend has got to stop, and together with our collaborators, we’re really on a mission to make that happen. So rather than just complying with requests for the new animal test data, we’re pushing back and we’re pushing the alternatives to force the discussion directly with the regulators where we believe that that is scientifically the right thing to do.

Jesse Harris   37:23

Well, it’s very admirable for every host to be pushing for that, and I’m very interested to hear how those affect things over time. But thank you so much for your time with us today, Steve, it’s been lovely chatting with you.

Steve Gutsell   37:36

Excellent, thanks very much for the invitation. It was great talking with you.

Charis Lam   37:39

Yes. Thank you.

Jesse Harris   37:40

Great talking.

Charis Lam   37:41

It’s exciting to hear about all the progress in this field. These tools are only getting stronger, which means better predictions.

Jesse Harris   37:49

If you want to learn more about computational chemistry software, check out the Percepta Platform from ACD/Labs. We’ll leave a link in our show notes.

Charis Lam   37:57

That’s all for this episode. In our next episode, we’ll hear the case for CASE—computer assisted structure elucidation.

Jesse Harris   38:05

Remember to subscribe to the analytical wavelength so you never miss an episode.

Charis Lam   38:09

The Analytical Wavelength is brought to you by ACD/Labs. We create software to help scientists make the most of the analytical data by predicting molecular properties, and by organizing and analyzing the experimental results. To learn more, please visit us at

Enjoying the show?

Suscribe to the podcast using your favourite service.