Periodic Reporting for period 1 - KvasirAccelerator (The first utility-oriented, agnostic software platform for quantum-centric drug discovery)
Reporting period: 2023-08-01 to 2024-07-31
Pharma companies are striving to meet the ever-increasing complexity of drug R&D, while not wasting money upfront (~€2.5Bn average development costs, 42-50% of which is spent on drug discovery/preclinical development) and increasing trial success rates (currently <12%). Quantum Computing (QC) is getting closer to its commercial utility, with chemistry and pharma seen as two of its most promising and disruptive areas of application.
Currently, computational problems are holding back pharma R&D due to: 1) lack of context-specific, large training datasets for machine learning (ML) frameworks; 2) inaccurate simulations of intricate molecular/biochemical systems using even the world’s largest supercomputers; 3) too much complexity for targeted users to comprehend or extract knowledge.
Kvantify aims to develop and deploy the first custom-made, user-friendly combined QC-HPC software platform. The goal is to reduce wet-lab and animal testing, and ultimately reduce time to market for new drugs. Our platform will provide chemistry and pharma (including biotech) industry with a versatile, high-performance, cost-efficient software capable of harnessing the power of QC and high-performance computing (HPC) to address complex computational problems with unprecedented speed, accuracy, and explainability. Our solution is based on quantum mechanics and classical mechanics methods to simulate physical systems to boost the interpretability, efficiency, and speed in chemical and pharmaceutical R&D processes.
Our goal is to develop a problem- and customer-centric, hardware-agnostic software tool. It will address the needs of customers who currently rely on unintuitive, slow, and laborious simulation routines, which require extensive knowledge and computational expertise to use, often generating inaccurate results that are not translatable to the wet lab, leading to increased failure rates and spending by pharma players during drug development.
Currently, computational problems are holding back pharma R&D due to: 1) lack of context-specific, large training datasets for machine learning (ML) frameworks; 2) inaccurate simulations of intricate molecular/biochemical systems using even the world’s largest supercomputers; 3) too much complexity for targeted users to comprehend or extract knowledge.
Kvantify aims to develop and deploy the first custom-made, user-friendly combined QC-HPC software platform. The goal is to reduce wet-lab and animal testing, and ultimately reduce time to market for new drugs. Our platform will provide chemistry and pharma (including biotech) industry with a versatile, high-performance, cost-efficient software capable of harnessing the power of QC and high-performance computing (HPC) to address complex computational problems with unprecedented speed, accuracy, and explainability. Our solution is based on quantum mechanics and classical mechanics methods to simulate physical systems to boost the interpretability, efficiency, and speed in chemical and pharmaceutical R&D processes.
Our goal is to develop a problem- and customer-centric, hardware-agnostic software tool. It will address the needs of customers who currently rely on unintuitive, slow, and laborious simulation routines, which require extensive knowledge and computational expertise to use, often generating inaccurate results that are not translatable to the wet lab, leading to increased failure rates and spending by pharma players during drug development.
We have established a core platform for computational drug discovery, where the Kvantify workflow system is the main component. With the core platform it is possible to automate the use of cloud resources and control the overall flow of complex computational drug discovery pipelines. The platform has been used to perform computations for customers in drug discovery, and for benchmarking of Kvantify's own pipelines and products.
We have implemented classical computing tools to provide stable state-of-the-art accurate and reliable benchmarks for quantum computing simulations. Specifically, a Nudged Elastic band (NEB) tool has been developed, which identifies the saddle points and minimum energy paths between known reactants and products. This NEB tool has been coupled with a projective embedding quantum chemical methodology into a classical projective framework, which allows us to accurately evaluate potential energy surfaces for chemical reactions.
We have implemented quantum computing tools, especially the FAST-VQE algorithm. This algorithm is a novel and exceptionally adept algorithm for Noisy Intermediate-Scale Quantum (NISQ) devices, delivering performance and cost-effective electronic structure calculations of small molecules. We have included FAST-VQE in the classical projective embedding framework, thus expanding the usage of quantum computing into the realm of realistic applications.
We have developed an algorithm for calculating kinetics rates, which gave rise to the launch of our first product called koffee. With koffee unbinding kinetics are calculated in minutes.
We have implemented advanced molecular docking tools. Our development efforts are centered around fast Machine Learning-based methods for both the initial pose generation and scoring.
We have implemented classical computing tools to provide stable state-of-the-art accurate and reliable benchmarks for quantum computing simulations. Specifically, a Nudged Elastic band (NEB) tool has been developed, which identifies the saddle points and minimum energy paths between known reactants and products. This NEB tool has been coupled with a projective embedding quantum chemical methodology into a classical projective framework, which allows us to accurately evaluate potential energy surfaces for chemical reactions.
We have implemented quantum computing tools, especially the FAST-VQE algorithm. This algorithm is a novel and exceptionally adept algorithm for Noisy Intermediate-Scale Quantum (NISQ) devices, delivering performance and cost-effective electronic structure calculations of small molecules. We have included FAST-VQE in the classical projective embedding framework, thus expanding the usage of quantum computing into the realm of realistic applications.
We have developed an algorithm for calculating kinetics rates, which gave rise to the launch of our first product called koffee. With koffee unbinding kinetics are calculated in minutes.
We have implemented advanced molecular docking tools. Our development efforts are centered around fast Machine Learning-based methods for both the initial pose generation and scoring.
Based on our focus on developing modules to extract reaction-rate binding constants we have launched our first product – koffee – which calculates kinetics rates in minutes. This innovative product aims to complement and reduce costly and time-consuming laboratory experiments by calculating unbinding kinetics, which is a critical parameter in screening and selecting new drug candidates. Kvantify koffee is the first commercial product designed to simulate unbinding kinetics. The currently launched version of koffee adds value using classical computers and is poised to offer even greater value as quantum-computer hardware matures. We are in the process of patenting the algorithm that drives koffee.
In quantum computing, we've developed the FAST-VQE algorithm, a novel and efficient solution for quantum chemistry calculations, enabling electronic structure analysis on today's Noisy Intermediate-Scale Quantum (NISQ) devices. We have successfully run this quantum-classical hybrid algorithm on Amazon Braket and demonstrated the feasibility of conducting accurate electronic structure computations on the quantum computing hardware that is currently available.
We have developed a projective embedding framework, and by including the FAST-VQE algorithm in this embedding, we have pushed the use of quantum computing into the realm of realistic applications thanks to the quantum-classical hybrid approach embraced. One example is the calculations of the energy profile of an enzymatic reaction, made in collaboration with Novonesis. Results are published in: Ettenhuber, P., Hansen, M. B., Shaik, I., Rasmussen, S. E., Poier, P. P., Madsen, N. K., Majland, M., Jensen, F., Olsen, L., and Zinner, N. T. 2024. ‘Calculating the energy profile of an enzymatic reaction on a quantum computer’. arXiv.org: 2408.11091.
In quantum computing, we've developed the FAST-VQE algorithm, a novel and efficient solution for quantum chemistry calculations, enabling electronic structure analysis on today's Noisy Intermediate-Scale Quantum (NISQ) devices. We have successfully run this quantum-classical hybrid algorithm on Amazon Braket and demonstrated the feasibility of conducting accurate electronic structure computations on the quantum computing hardware that is currently available.
We have developed a projective embedding framework, and by including the FAST-VQE algorithm in this embedding, we have pushed the use of quantum computing into the realm of realistic applications thanks to the quantum-classical hybrid approach embraced. One example is the calculations of the energy profile of an enzymatic reaction, made in collaboration with Novonesis. Results are published in: Ettenhuber, P., Hansen, M. B., Shaik, I., Rasmussen, S. E., Poier, P. P., Madsen, N. K., Majland, M., Jensen, F., Olsen, L., and Zinner, N. T. 2024. ‘Calculating the energy profile of an enzymatic reaction on a quantum computer’. arXiv.org: 2408.11091.