How quantum computing - artificial intelligence can change the pharmaceutical industry
Pharmaceutical companies spend 15% of their sales on R&D, a figure that represents more than 20% of total R&D spending across all industries worldwide.
Fortunately, high investments are often coupled with innovation.
Pharmaceutical companies are constantly seeking new ways to improve the R&D process, from early adopters of digital tools for computational chemistry; artificial intelligence is currently accelerating its penetration...
The next technology in line, however, is quantum computing.
How is this technology being applied to the pharmaceutical sector? What is the current state of marketability? And how can pharmaceutical companies get ahead of the curve? This industry report from McKinsey is worth reference.
01A natural candidate for quantum computing
A core aspect of drug development is the identification and development of molecules that help cure diseases, which also means that pharmaceuticals are natural candidates for applying quantum computing.
Whereas molecules are systems based on quantum physics, QC can predict and model their structure, properties and reactivity, and even interactions at the atomic level, more effectively than traditional computing.
While the technology behind quantum computing is difficult to understand intuitively, its impact is much easier to grasp: it can handle certain types of computational tasks much faster than today's traditional computers.
As a result, once fully developed, QC could add value across the entire drug value chain.
02The main value is in drug development
Quantum computers can increase the range of computations applicable to biological mechanisms, shorten screening times, and clean up some of the "dead ends" of research, resulting in significant cost savings in the drug discovery phase.
Figure: Major application segments of QC in biopharmaceuticals
For example, predicting molecular properties with high accuracy to make current CADD/AIDD tools more effective; parallel screening of computational libraries against multiple possible structures of a target to increase the chances of identifying the best drug candidate.
Figure: Specific areas where QC-enhanced CADD can improve
In the long term, QC can discover new structure-property relationships using machine learning (ML) algorithms to improve hypothesis generation and validation. Once sufficient maturity is reached, new libraries of drug candidates, including small molecules, peptides and antibodies, can be created, enabling a more automated approach to drug discovery.
Target Identification and Validation
During target identification, QC can be used to predict the 3D structure of proteins and obtain high-quality data; even the AlphaFold developed by DeepMind, which does not address the challenges of protein complex formation, protein-protein interactions and protein-ligand interactions, may be improved with QC's feature of allowing explicit processing of electrons. In addition, QC will achieve more powerful computational efficiency than Google's AI model, which requires more than 120 high-end computers for several weeks.
Molecular generation and validation
The bottleneck of existing computers is the sequential approach, insufficient arithmetic power, and pharmaceutical companies can only use CADD on small to medium-sized drug candidates. with a sufficiently powerful QC, pharmaceutical companies can extend all use cases to selected biologics, such as semi-synthetic biologics or fusion proteins, and perform computer search and validation experiments at higher throughput. This use case will go beyond protein identification to cover almost the entire known biological community.
Pilot compound optimization
QC allows for enhanced absorption, distribution and metabolism, more accurate organ system activity and toxicity prediction, and safety issues such as dose and solubility optimization - three key parameters for improving R&D productivity.
Data Linkage and Generation
Creating logical connections between data points through effective semantic management is one of the core technologies in drug discovery. The industry is currently conducting research on "topological data analysis", which aims to identify "gaps" and "connections" in large data sets. In addition, QC can "deepfake" missing data points throughout the research process, i.e., use ML algorithms to generate a kind of fake data. This is particularly useful in rare diseases where data is scarce and then supplemented by manual datasets.
Clinical Trials
Clinical trials can be optimized through patient identification, stratification, and population pharmacogenetic models. In trial planning and execution, QC can optimize trial sites and enhance causality analysis of side effects to improve active safety monitoring.
03Current industrialization trends
Quantum computing technology is becoming increasingly mature, and the trend of its application in the pharmaceutical industry is as follows.
The initial application period will be from 2020-2030, when quantum computing will be used in some innovative drug companies and individual aspects of pharmaceuticals; after 2030, pharmaceutical companies will be more firmly in the field as they play a greater value through quality control.
So when will pharmaceutical companies start to reap the benefits of QC?
It depends on the technology starting point (current level of digitization) and business focus, i.e., the number of small active pharmaceutical ingredients (APIs) in the portfolio.
Pharmaceutical companies with a strong footprint in CADD and a focus on small molecules in their R&D will be the first to take advantage of emerging QC technologies.
Over the next five to ten years, the McKinsey team predicts that the first QC tools deployed by pharmaceutical companies will rely on hybrid approaches that use classical algorithms and QC subroutines to create additional value. Examples include the Variable Quantile Quotient Solver, or VQE (a computational API and target receptor).
04A guide for pharmaceutical companies to act
AI pharmaceutical companies have already built a good technical foundation, such as CADD, AI, ML, etc.; some of them are already using quantum chemistry simulations, so the barrier to entry is low.
Scientists don't need to change the way they develop drugs, they just need to use more powerful tools.
Whether you wait and see, or go all in, the following strategic guidelines will help companies make informed decisions based on their circumstances.
First, assess the size of the opportunity. At the current rate of development, each drug company should find out how much exposure it has and the size of its QC opportunity.
The McKinsey team asked three major questions to aid in the determination.
Will QC disrupt the company's field and restructure the competitive landscape?
Is there an area in the company's value chain where QC's value is reflected? When will this happen?
Can the company devote resources to investigating the QC opportunity?
Second, building partnerships. Individual companies have begun to collaborate in the area of quantum computing, such as the QuPharm consortium, consisting of GlaxoSmithKline, Takeda, Pfizer, Merck, and AbbVie, whose members have worked together to develop more than 20 cases of QC applications in the pharmaceutical industry.
Other cross-industry QC research consortia, such as NEASQC and QED-C, include a small number of pharmaceutical companies among their members.
Third, develop talent early. The digital talent gap is already a reality, and QC may only exacerbate that gap. And Boehringer Ingelheim, Amgen, and Roche have already established QC teams comprised of pharmacology scientists, physicists, and computational scientists.
Fourth, ensure internal collaboration. Separating internal research, technology, and business functions, communication, and cross-functional collaboration in action will be the new hallmark of pharmaceutical companies that leverage QC.
Quantum computing may be the key to exponentially more effective discovery of drug treatments and cures, and to creating hundreds of billions of dollars of value for the pharmaceutical industry.
The McKinsey team predicts that: by 2030, global pharmaceutical companies will spend billions of dollars on QC in R&D. Pharmaceutical companies are advised to assess the opportunities for themselves and start laying the groundwork for themselves in the new competitive track.
Appendix: Some of the companies applying AI+Quantum technology
1.Jingtai Technology (XtalPi)
Founded in 2014, Jingtai Technology is a quantum physics-based AI drug discovery and development company with a mission to revolutionize drug discovery and development by increasing speed, scale, novelty and success rates. Headquartered in Boston, it returned to China in 2015 to launch full-scale research operations.
Its intelligent digital drug discovery and development platform, ID4, combines quantum mechanics, artificial intelligence and high-performance cloud computing algorithms to allow prediction of the physical and chemical properties of small molecule drug candidates, as well as their crystal structures, with high accuracy.
Currently, Epistar has raised a cumulative total of $786.4 million with investors including Sequoia China, Tencent and Google, making it one of the most well-funded computational drug discovery startups in the market. It has also conducted several research collaborations with pharmaceutical companies, including Pfizer.
2. Aqemia
Incubated by ÉcoleNormale Supérieure and based in France. Its CEO and co-founder, Dr. Maximilien Levesque, has been working on algorithms for eight years, and the company combines two technologies, quantum computing and artificial intelligence, to design small molecule drugs from scratch.
Using quantum-inspired statistical mechanics algorithms - structure-based design of lead-like molecules - it can accurately predict the affinity between a compound and a therapeutic target and do so 10,000 times faster than its competitors; in addition, its AI platform gets feedback from affinity predictors and generates compounds with higher accuracy.
In February 2022, Aqemia announced a pilot study with Johnson & Johnson to leverage its quantum physics-driven drug discovery technology to predict the potency of small molecules for a given target based on Janssen's selection of physics-based computational datasets. in June, a molecular design and drug discovery collaboration with Sanofi in oncology was launched, the latest development following a collaboration agreed upon by the two companies in late 2020.
3. Hafnium Labs
Hafnium Labs was founded in Denmark in 2018 as an early-stage startup. The researchers have developed two software packages, Q-props, which is used to simulate the physical properties of pure components and mixtures with high accuracy, and Epsilon, which focuses on the simulation of electrolytes.
Both software products combine the latest technologies in quantum chemistry, artificial intelligence (AI) and cloud computing to achieve high accuracy predictions.
Through its cloud-based computing capabilities, the company can accurately predict chemistry to accelerate drug discovery, new material development and other areas. In addition, it has a pay-per-use business model, which the company says is more affordable than a license-based model.
To date, Hafnium Labs has raised a total of $1.8 million through grants from the Danish Innovation Fund, IBM and the Climate-KIC Acceleration Initiative (EIT).
4. Kuano
founded in early 2020, is a UK-based startup currently developing novel AI and quantum solutions for designing enzymes to address key questions of specificity, efficacy and resistance. It has raised $1.4 million in seed round funding.
Kuano's research platform combines quantum simulation with quantum-inspired artificial intelligence and chemistry using structural data from target enzymes or catalytic sites.
It has in-house programs in epigenetics, protein degradation, immunometabolism and infectious diseases, as well as working with partners to develop next-generation inhibitors of clinically and commercially validated enzyme targets.
5.Menten AI
A Canadian startup founded in 2018 to develop a protein design software platform powered by machine learning and quantum computing. The company uses quantum-optimized proprietary algorithms because of their ability to not only improve the accuracy of drug discovery, but also reduce costs and development time.
MentenAI created protein design algorithms for current quantum computers and became the first team to apply quantum computers to design protein molecules.
The company claims that they can complete a design cycle from target selection to in vivo efficacy in less than six months. The primary development pipeline is peptide therapeutics for indications with high unmet medical need.
To date, nearly $4 million has been raised. And has established partnerships with D-Wave supercomputers, IBM-Q, and recently announced a collaboration with Xanadu.
