Rene - Hi! Welcome to QuBites, your bite-sized pieces of quantum computing. My name is Rene from Valorem Reply and today we're going to talk about quantum computing for life sciences and pharma. And for this, I'm very honored to have a special expert guest today, Ivan Ostojic. Hi Ivan and welcome to the show. How are you today?

Ivan – Hi! I'm very good, thank you for having me. It's a real pleasure to talk to you.

Rene - Perfect, can you tell us a little bit about yourself and your background as it relates to quantum computing.

Ivan – Yes, my name is Ivan. I'm a partner in McKinsey in Zurich and my background is that I lead our quantum group within McKinsey globally. I have kind of dual background, life science, biophysics where I also did a lot of analytics and then the other one is technology innovation management. So, I'm looking at a lot of emerging technologies within McKinsey and I fell in love with quantum. I think the power of technology to transform many industries is immense and it can help us absolve some burning problems for humanity.

Rene - Absolutely, and that's what we're actually going to talk about and so let's dive into today's topic. So, like you actually work at McKinsey on life science, pharmaceutical and also work with other chemical industries and so on, right? So, what are some of the challenges these industries are currently facing, when it comes to computation and computational power and how can quantum computing help to overcome these challenges?

Ivan – Yeah, so at the moment if you look at the pharma value chain, I think there are essentially one huge barrier to entry to this industry, which is around R&D and this R&D at the moment is kind of undergoing disruption because a lot of knowledge that was implicit in scientists, you know, picking the right molecule and so forth, it's becoming explicit to use through use of artificial intelligence Find new molecular targets etc. However, there is a computational challenge in that piece, especially around chemistry because you know taking human out of equation and doing kind of, in-silico modeling of chemistry has progressed but not to the extent that we can automate this. So that's one thing where really quantum has a huge promise. The second one is as machines get stronger and we're probably talking or more powerful, we're probably talking some decades ahead, but actually we have a chance to move pharma from real kind of experimentation to real in-silico. A little bit of what happened to automotive industry like some decades ago, you know, when you wanted to do a kind of a crash test for a car, you had to do everything manually and now you can simulate. I think we can simulate a lot of things in human, whether molecules are toxic, we see it already with AI but I think, quantum computing would bring that on a whole different level. So, to summarize, I think in R&D, it's the chemistry and we can go much more into details, and I think it's a lot around biology and interaction between molecules and the body that we will be able to simulate to understand the effects. And then just to wrap up this whole thing, although, I focus on R&D because that's kind of imminent opportunity for quantum, if you think of complex optimizations from production to logistics, supply chain and even, you know, through some commercial interactions, helping doctors choose the right drug for right patient and optimize the treatments, all of these are areas where quantum computing algorithms can play a role especially optimization ones. So, I think I think you know the effect would be end-to-end, but I think the first immediate or imminent opportunity is in R&D.

Rene - So, this is amazing so basically you can apply quantum computing to pharmaceutical especially for the chemical processing side. Could you also say, since you know chemistry, if you're actually going down how, you know, all these smallest particles interact with each other like atoms and you know, subatomic particles you're actually dealing with quantum mechanics, right? And so, if you're kind of simulating this quantum mechanics, of course, it's better to simulate those on a quantum computer right, because you're closer to the real thing. Is that fair to say?

Ivan - It's fair to say. So, if I may just come back, it's not really how we make molecules that is important part but if you understand deeply pharma there is something like… Let me just explain for your audience who is not expert in pharma on a high level. So first you need to find your target, which is the molecule in your body that you are targeting with the drug, where the biological effect will happen. Once you find a target, you need to find the optimal lead and at the moment it's a very, kind of manual process, where you are screening like tens of thousands of different chemicals in order to find an optimal hit, hit means this molecule is binding to this molecule in your body that will have biological effect and then once you find a hit then you need to, that's called lead molecule, you need to optimize that lead molecule using different chemical methods and that process is very manual today. And I think now coming back to your follow-up question, if you use quantum computing, first of all, at the moment we are limited to the libraries of chemicals that we already have. With quantum computer, we can kind of extrapolate, we don't need machine learning to learn on past molecules, we can actually literally simulate you know molecular structures. So finding hits will be easier and be much more assisted. And then what is known as I explained lead optimizations, optimizing that molecule to be soluble, to get into the body, to be less toxic, to have higher affinity, all of that I believe can be done much better by use of sort of quantum computing.

Rene - Makes a lot of sense and regarding the impact you're already seeing today. So, can you share a little bit, you know, what are some of the impact we're already seeing today with things like quantum inspired computing and do you have an example maybe?

Ivan – Yes, I have two examples, and so just not to overhype technology right, I'm very cautious about that, so that you manage expectations appropriately. I mean, at the moment, you know, there is no application in pharmaceutical on, let's say, especially in chemistry, on a real quantum hardware. However, you are absolutely right, quantum inspired algorithms, which means, as most of your audience will know, but I’ll just say its quantum inspired, so these algorithms are running on a conventional machine or in, you know, certain hardware configuration of conventional machines or we have seen that couple of pharmaceutical companies have found fairly promising applications that actually generated some real impact. So let me give you two example one is around protein-protein interactions and protein optimizations using, you know, combination of quantum annealers and high-performing computing sequence that can optimize certain interactions or help scientists understand. And the other is quantum-inspired algorithms in the, what is known as, cat screening. So essentially, chemical compound screening and modeling to match ligands meaning the small chemical molecules with target proteins, these are like biological molecules in your body. There actually people have made some breakthroughs generating a whole workflow to improve that screening using quantum inspired algorithms. And apparently, there are proofs this is coming from some of our clients, so but there are proofs actually that this produces superior results, the quantum inspired algorithms than conventional, let's say, artificial intelligence or machine learning algorithms. So very interesting and exciting that we already see some early proof points.

Rene – Agreed, and maybe to also add another example, I was just talking with a friend actually for another QuBites episode about quantum machine learning right and he was basically saying the same thing is like even with these small quantum computers you can build like hybrid AI models, like neural networks and at a higher layer you can basically inject a quantum layer and even with, like you know, when you already have the features extracted, you can basically use that quantum layer to give you better accuracy for image recognition and so on right, because it's more in a probabilistic nature. It's amazing.

Ivan - I do believe that strongly that the first application will be exactly like that. You will you will have to devise the complex workflows which would contain you know high performance computing sequences of algorithms that would, as you said, extract feature prepare them then you'll run certain things on a platform or quantum inspired algorithms to get either higher accuracy or to solve certain things that you couldn't solve classically and then once the, you know, the problem is cracked you can you can pull it back into normal high performance computing sequence. I think that's the probably first you know first areas of application and then depending what we have fault tolerant quantum computer will we probably be able to do much more on chemistry side.

Rene – Gotcha. That makes a lot of sense and IBM just announced they built I think they're leading the game now. I think a 127 qubit computer, they just built that one, and they're planning to do a 433 qubit quantum computer as a next step. They're approaching pretty fast, right. So, since you're also very well-versed in the whole quantum computing world, let's talk about the general health of the ecosystem and your future perspective because you talk a lot with clients in the industry, so I really appreciate your thoughts on that. So, where are you seeing the biggest blockers at the moment for photo adoption in general but especially in these industries and what's going to happen in the next couple of years in-terms of use cases and you know, how to accelerate growth in the ecosystem?

Ivan - Yeah, so I think, I mean, couple, let's call it couple of blockers. First of all, I don't know if it's a blocker but it's an opportunity. I think we need this, kind of, more integrated working across the whole ecosystem. So what I mean by that, I call it mission teams, you know, if we would have, let's call it like, if we talk about life science pharmaceutical companies working with specific, you know, quantum hardware companies and then software communities end to end, in order to define valuable problems to solve with quantum computing on a specific level such that quantum scientists can convert that into algorithms and then you know, try to execute with expectation to learn. I don't think we can have huge expectation at this moment, and I think that's number one and I know that there are pharma groups like Q Farm and so forth already working on it, but I think this would really accelerate all learning and understanding what's missing and so forth. Number two, I think you know at least in Europe where both of us are, we have gap in funding that we need to close in order to further accelerate but maybe that's not the biggest problem. The third biggest problem I'll mention is talent and I think talent both on corporate side but also on quantum application side. So what I mean by corporate side, the main breakthrough in artificial intelligence happened when we built a kind of a pool of people who we call translators, so the people who actually quite understood artificial intelligence or machine learning very well but they knew the industry in depth and I think we need that pool for quantum in order to help pharma cos, you know, identify the use cases and define you know parameters of solution space in order so that quantum science application scientist and then you know, even already in our modelling, we see a huge gap in talent on a quantum side because I mean you know master programs and so forth are just emerging and writing quantum code or further as you know yourself and your audience it's not the trivial thing. So, I think these are the main three things. If you ask me about sort of hardware milestones and development, I think, a, we should be cautious not to overhype technology because if we raise expectation and don't deliver that's a huge let's say, it will delay us in terms of corporates, who will say oh this is overhyped, and adoption will slow down. And the second thing is you know we should be patient with this hardware milestones. Although you know we see two bitcoins and different things increasing, I mean the jury is out and there is a big debate among um we have our McKinsey technology council where we have a lot of competing companies in quantum that are discussing and there is a huge, let's say, debate are we able to do something in this uh what people kind of label as risk error or not so. Meaning are we going to get some useful use cases until we have higher order of error correction or not and I think while that jury might still be out. We still have quantum simulators, we can develop algorithms, we can understand if certain things will work and so forth, so instead of, you know, overhyping and going in that direction, I will rather more focus on fundamental groundwork, like some companies are doing to actually, you know, develop algorithms and prove that certain cases can be used. So, I think my sequence to elevate a lot of words that I used up would be, let's define useful problems. let's build communities that will work on this and then actually let's first work on algorithm development in a simulated environment to understand how much capacity we'll need, what are prerequisites and can we theoretically solve those problems. I think that would be actually pretty good proof points that technology will work already, and I think that's probably what we should focus at the moment in order to drive this.

Rene - Fully agreed, I mean, like, what you're saying, it definitely makes a lot of sense. Also, the point you made about talent and, you know, that there's a lack of skilled workforce but not just, like you said, not just like quantum physicist with a PhD but the whole stack, right, you need also these people in between that are domain experts in both sides a little bit. But like you said, translators is a perfect description of that and that's what we're also trying to do with the show, right, making it more approachable and trying to provide a way that you can understand the impact, like, you know our audience can understand the impact and we can share what's already possible today and build this community a little bit. So, thank you so much for joining us today and sharing your insights. That was fantastic, very much appreciated.

Ivan - Thank you for having me, I enjoy your shows and I like really this short QuBites, you know, talks. So, if we can bring a little bit like in simple language into the discussion. I really enjoyed the conversation and I hope you found it useful.

Rene - It was awesome and thanks everyone for joining us for another episode of QuBites, your bite-sized pieces of quantum computing. Watch our blog, follow our social media channels to hear all about the next episodes and also to find out the previous episodes of course. Well thanks again, take care and see you soon! Bye-bye.

QuBites 4.4 - Enterprise Quantum Computing and QML