Peptone was founded in 2018 with the goal of using experimentally derived biophysical data and molecular dynamics simulations to model the behaviour of disordered targets so enabling investigation of new drug candidates. By bringing these proprietary approaches together, Peptone is exploring the world of intrinsically disordered proteins (IDPs) – proteins without a fixed structure that play a significant role in health and disease – opening up the possibility of novel therapeutics against an entire class of high value and previously undruggable targets.
UK-headquartered Peptone is at the forefront of this research, in partnership with NVIDIA and Verne Global datacentres. Peptone is a computational molecular physics company that has used artificial intelligence methods to analyse the anomalous behaviour of proteins and has developed the Protein Engineering Operating System (PeOS).
Project Background
Protein structures and their various conformations are dynamic and complex. Proteins such as receptors, antibodies, growth factors, and hormones have a vast number of potential variations in structure and their molecular behaviour can be varied. Proteins can aggregate, or stick together, altering their characteristics, and one protein’s interaction with another can change its shape and functionality. Understanding how proteins act, knowing their size, and visualising how they look is a fundamental key to creating protein-based drugs and vaccines.
With intrinsically disordered proteins accounting for one-third of all human proteins, many of which play pivotal roles across cell growth and signalling, IDPs represent an abundant pool of novel and validated drug targets across a wide range of therapeutic areas and clinical indications.
Project Approach
Peptone's team of physicists, structural biologists, computer engineers and mathematicians, led by physicist and founder Dr Kamil Tamiola, has spent the last three years developing the foundations of Protein Engineering Operating System (PeOS). In collaboration with NVIDIA, Peptone developed PeOS to handle massively parallel molecular simulations. The core functionality of the platform is an automated search for non-obvious protein variants with desirable therapeutic properties and cost-effective manufacturability. To build its repository of proteins, Peptone must capture, store and analyse huge amounts of data, orchestrated and supervised by reinforcement learning algorithms. Reinforcement learning is the training of machine learning models to make a sequence of decisions. The agent learns to achieve a goal in an uncertain, potentially complex environment.
To achieve the scale of compute required to develop the PeOS, Peptone sought to use GPU-accelerated computing in the form of a NVIDIA DGX A100 supercomputer located at Verne datacentre in Keflavik, Iceland, as it forms part of the NVIDIA DGX-Ready Data Centre Program and is optimised to maintain and support high density and HPC applications running on NVIDIA infrastructures. Additionally, it was important to Peptone that the datacenter is powered by 100 percent renewable energy in line with its commitment to sustainability without compromising security. Ultimately, the PeOS will handle a variety of mundane and complex tasks that are related to anomaly detection, risk of failure calculation and suggest the most plausible routes of failure resolution, including choosing the right commercial partner to generate lab-quality molecule at a pace and cost not attainable before in the Big Pharma sector.
Project Results and Next Steps
The Peptone team intend to build on the success of PeOS in order to continue crating small-molecule therapeutics that selectively target IDPs, using techniques including bespoke mass spectroscopy and generative AI.
The Scan Partnership
Peptone approached the Scan AI team, as a leader in the UK AI market, to help it find a solution to delivering significant but cost-effective GPU-accelerated compute resource. Taking advantage of its long-standing relationship with NVIDIA, Scan AI helped Peptone unlock the potential of its research by running its deep neural network models on an NVIDIA DGX A100 system. Working together with Scan, the next task was to find a datacentre facility equipped to host and support this highly specialised equipment. In addition, with privacy and security top-of-mind for Peptone’s pharmaceutical customers, there was a reluctance to place this valuable, proprietary data in the cloud. One of Scan’s datacentre partners, Verne Global, was identified as a provider that could fulfil all these criteria.
Project Wins
Reduced time to identification of drug bonding sites within IDPs using GPU-accelerated compute
Success in using renewable energy to power AI through partnership with Verne
Dr. Kamil Tamiola
Founder & CEO, Peptone
"Scan and Verne Global are the ideal partners for our hybrid supercomputing cloud approach due to its foundation in sustainability, global connectivity and reassuring ability to keep our world-leading practical research completely secure."
Dominic Ward
CEO, Verne Global
"Verne Global as datacentre partner, is powered by 100 percent renewable geothermal and hydroelectric energy sources, is optimised for the secure and scalable high-intensity computing required by Peptone to successfully meet the growing demands of the protein therapeutics market and accelerate the time to market of protein-based drugs."
Speak to an expert
You've seen how Scan helped Peptone accelerate protein-based drug discovery through GPU-accelerated computing infrastructure. Contact our expert AI team to discuss your computational physics project requirements.
phone_iphone Phone: 01204 474210
mail Email: [email protected]
Related content
Peptone
Learn more about the company and the ground breaking research they are undertaking.
Read more >
Sustainable Datacentres
Learn more about our partnership with Verne datacentres to promote renewable energy sources.
Read more >