DP Technology Unveils Hermite®, a Next-Gen AI-Powered Drug Design Platform Embodying the AI for Science Paradigm

At 14:00 on October 29, 2022, DP Technology's new generation drug computing platform Hermite based on the AI for Science research paradigm^®Officially Released to the Public!The platform integratesArtificial Intelligence、Physical ModelingAndHigh-Performance Computing, providing a one-stop computational solution for preclinical drug research and development,Including important functions such as protein structure prediction, drug-target binding mode prediction, hit compound screening, lead compound optimization, and molecular property prediction.

Attended this Hermite^®The participants in the year-end press conference included: E Weinan, academician of the Chinese Academy of Sciences (CAS), director of the Beijing Institute of Artificial Intelligence for Science, and chief scientific advisor of DP Technology; Sun Weijie, founder and CEO of DP Technology; Zhang Linfeng, founder and chief scientist of DP Technology; Xu Bo, chief product officer of DP Technology; Xu Tao, head of DP Technology's drug design platform; Zhu Weiliang, director of the Center for Drug Discovery and Design at the Shanghai Institute of Materia Medica, CAS; Wang Renxiao, distinguished researcher at the School of Pharmacy, Fudan University; Tang Yun, professor and doctoral supervisor at the School of Pharmacy, East China University of Science and Technology; as well as renowned experts and scholars from leading companies in the industry such as Volcano Engine, Hengrui Medicine, EndPoint Pharma, Lizhu Medicine, Hongbo Medicine, BeyondSpring Pharmaceuticals, XtalPi, and Insilico Medicine.

At the press conference,E Weinan, Academician of the Chinese Academy of Sciences, President of Beijing Institute of Artificial Intelligence, Chief Scientific Advisor of DP TechnologyDelivered the opening remarks, pointing out that "the emergence of the AI for Science research paradigm represents an important historical opportunity for technological innovation." This technology not only expands the capability boundaries of data-driven and physics model-driven approaches but also holds the potential to promote their organic integration, providing a theoretical foundation for further solving practical problems. E Weinan emphasized that "achieving the innovation-implementation closed loop cannot rely on wishful thinking, will not happen overnight, and will not be a one-time solution." To achieve the innovation-implementation closed loop, it is necessary to address the real needs of the industry, tackle the genuine scientific and engineering issues behind those needs, and foster continuous communication and collaboration among people from all fields. Hermite^®This is a milestone practice and representative product of AI for Science, from algorithm innovation to application implementation.

E Weinan, Academician of the Chinese Academy of Sciences, President of Beijing Institute of Artificial Intelligence, Chief Scientific Advisor of DP Technology

"AI technology can better solve the 'curse of dimensionality' currently faced in drug development."Founder and CEO of DP Technology, Wei Jie SunIn the welcome speech, it was expressed that, "Choosing Hermite^®"It is about choosing to align with a new era." In the past, the calculation of compound properties was done by computing their abstract interaction forces, which was fast but had significant errors. Calculations based on the first principles of quantum mechanics, analyzing the electronic structure and density around atoms, yield more accurate results, but they require calculations of multi-dimensional functions, resulting in relatively slower speeds. The ideal computation should combine precise algorithms with efficient computational power. DP Technology successfully overcame this challenge using the Deep Potential method, establishing a molecular simulation framework that is both fast and accurate. This method won the highest international award in high-performance computing applications, the "Gordon Bell Prize," known as the "Nobel Prize of computational applications," in 2020. It was also recognized as one of China's top ten scientific advances that year.

Founder and CEO of DP Technology, Sun Weijie

He divided the algorithms into three levels: L1 to L3. L1 involves imitating reality, which refers to the replication and extrapolation of experimental results; L2 is about predicting reality, where predictions of phenomena approach experimental accuracy with an expected margin of error; L3 focuses on searching reality, requiring accurate prediction of phenomena and the ability to directly search for optimal results from the space. Currently, for experimental disciplines such as chemistry, biology, materials, and pharmaceuticals, significantly reducing experimental errors and defining precision boundaries can largely replace existing experimental models. Since its establishment, DP Technology has been pursuing the mass production of a series of L2-level algorithms in every aspect, while making the realization of L3-level algorithms a key task post-2030. The widespread promotion of L2-level algorithms and the active development of L3-level algorithms represent the future direction of experimental scientific algorithms.

Hermite^®Aiming to become the eyes, hands, and brain of clinical drug development scientists, accelerating drug discovery, reducing R&D costs, and discovering disruptive innovative molecular entities.Founder and Chief Scientist of DP Technology, Linfeng ZhangIn the review of DP Technology's product development journey, it was shared that the team decided in July 2021 to develop Uni-Fold, a protein structure prediction tool based on deep learning models. It was clear that only by achieving high-quality results within a short period could they ensure the tool’s pioneering role and competitiveness in the market. Under the intense development process of continuously facing and solving problems, the product originally scheduled for completion in May 2022 had already gone through two iterations. "The development of cutting-edge algorithms is not as simple as making demos or publishing papers; quality software isn’t just developed—it’s used and iterated," emphasized Zhang Linfeng, Hermite.^®The release and launch is just the beginning; the real barrier of industrial software lies in continuous optimization driven by needs and feedback.

Founder and Chief Scientist of DP Technology, Zhang Linfeng

DP Technology adheres to the continuous cycle of learning, exploring, innovating, and solving, aiming to optimize its products under current conditions and continuously iterate to meet customer needs.Chief Product Officer of DP Technology, Xu BoFocused on讲述Hermite^®Before its market launch, the product has undergone iterative optimizations, including visualized 3D atomic mapping, an adjustable operational layout, displayable and editable pharmacophore representations, a purchasable library of 12 million physical molecules, and deployment options that are either privately local or cloud-based for immediate use, all of which have been updated to the latest version in this release.

Hermite^®The specific function modules launched at this release event areXu Tao, Product Director of DP Technology's Drug Design PlatformThe main explanation was conducted.

First, based on the deep learning method, Uni-Fold can quickly and accurately predict the three-dimensional structure of proteins dynamically. It is also capable of rapidly predicting large symmetric protein polymers. Through open-source code, parameters, and datasets, Uni-Fold reduces training time by at least 50% compared to previous platforms and improves complex protein prediction accuracy by at least 15%.

Secondly, Uni-Docking is a GPU-accelerated molecular docking engine. While maintaining screening accuracy, it pushes the virtual screening of multi-million molecule databases into an accessible, usable, and reliable era. When combined with MM GB/PBSA in a virtual screening workflow, the accuracy is further enhanced.

Again, by simulating the "induced fit" effect generated by the binding of drug molecules to targets, Uni-IFD can accurately predict the binding mode of drugs and targets. Meanwhile, Uni-Aquasite can calculate the free energy of each water molecule to identify water molecules that significantly impact drug-target binding.

Moreover, Uni-FEP, based on free energy perturbation theory, molecular dynamics, and enhanced sampling algorithms, efficiently evaluates the binding affinity between proteins and ligands with chemical accuracy. Meanwhile, Uni-FEP features a user-friendly interface and workflow, and its integration analysis of multi-dimensional and repeated calculations makes it an Assay computational tool for medicinal chemistry scientists.

In the subsequent based on “CADD Pioneer Interactive"and"Future Prospects of Drug Calculation" Roundtable Discussion on the Theme ofJian Ma, Co-founder and CEO of XtalPiUnder the current national celebration, what we need to do is not only to develop First-in-Class drugs but also to abandon the mindset of seeking quick success and instant benefits. We should utilize our self-developed Best-in-Class tool platforms to create our Best-in-Class drugs. As the saying goes, "A craftsman who wishes to do his work well must first sharpen his tools." Solving the issue of being restricted by imported products in terms of equipment platforms and possessing a drug discovery platform independently developed by the Chinese can fundamentally help cultivate domestic talent in medicinal chemistry and assist AI pharmaceutical enterprises. At the same time, it can reduce the cost of drug research, development, and manufacturing, promoting the improvement and development of China's medical security system.

Promote the deep integration of AI with life sciences, drug research and development, and other fields. The rich research tools and methods brought by AI for Science are leading to a significant leap in new paradigms within scientific research. DP Technology always believes: Achieving partners, leading innovation. Without each link in the industry chain being closely connected, without cooperation among industry collaborators, it would be impossible to understand needs, improve products, and continuously innovate.

In the future, DP Technology will continue to iterate and optimize Hermite.^®To ensure a good user experience, while continuous exploration will be carried out in the direction of molecular generation and large molecule compounds.We look forward to deeper cooperation with more leading pharmaceutical companies in the industry, expanding industry-university-research collaboration, and building a more comprehensive industry ecosystem. We and DP Technology share the same expectation!