Speed and accuracy are critical in drug development, with only 14% of candidates receiving US Food and Drug Administration approval1. Therefore, any efficiency improvements are of tremendous interest to a drug industry operating under escalating costs and discovery pressures. This is one of the reasons why Nuritas was founded in 2014 by mathematician and computational biologist, Nora Khaldi: not only to deliver incremental discovery gains, but also to create a disruptive opportunity for real change.
Nuritas combines the power of artificial intelligence (AI) with deep learning and ‘omics’ analyses to discover novel, therapeutic and commercial food-derived peptides. By integrating its AI predictive platform with in-house laboratories, Nuritas can rapidly discover and preclinically validate peptide therapeutics within months instead of years. Notably, peptide therapeutics carry the significant potential to provide the safety and efficacy required to better advance through the clinical trial process2.
From consumer to pharma
Nature by its sheer breadth and depth carries a widely recognized inexhaustible reserve of bioactive opportunities. However; it is this same vastness that has, until now, made nature relatively impenetrable to the 20th-century drug discovery process. This is the prevailing benefit of the Nuritas platform—the ability to use AI and machine learning to open new smarter, faster and more accurate pathways to discovery.
Nuritas’s discoveries have applications in many sectors including pharmaceuticals, health and wellness, dermatology, and animal and plant health. November 2018 witnessed the launch of the first product stemming from Nuritas’s technology, in collaboration with BASF. A global first, PeptAIde incorporates an AI-discovered bioactive serving as a compelling indicator of the real commercial impact Nuritas can create. “Delivering novel AI discoveries via consumer products enabled a quicker route to market versus traditional pharma and allowed us to validate our platforms for innovation and commercial applicability. To our knowledge, we are the first AI discovery company with a commercially validated platform,” said Chantelle Kiernan, director of life science at Nuritas.
The Nuritas discovery platforms are agile, flexible and disease agnostic, making them highly attractive to the pharmaceutical industry. Nuritas has proved its ability across a range of disease areas, with active early-stage discovery programs in diabetes, non-alcoholic steatohepatitis (NASH), inflammation and infection areas.
The Nuritas discovery engine
Nuritas’s discovery approach combines predictions generated by machine learning with in vitro laboratory testing all done in house (Fig. 1). This iterative feedback loop has been ongoing for 5 years, supplying the knowledge to discern what makes a peptide prediction effective in vivo. This has created unparalleled technology, applicable to any disease that captures the core information maximizing peptide activity, binding, stability and bioavailability.
Three data entries are used by the technology: unstructured and structured literature sets of publications and information on disease biology, which Nuritas manually curates; the company’s proprietary peptide library amassed through years of research on safe global natural sources, including foods, algae, animals, fungi and bacteria; and the phenotypic screening in Nuritas’s integrated laboratories of novel peptides in different types of human cell.
Using these complex entry points, Nuritas begins its prediction process for a selected target or disease area. The transferable technology can be applied across areas that include rare diseases, orphan diseases and unmet medical needs for which no targets have been identified. The predictions are synthesized and tested in vitro in house, with active peptides tested in vivo by partners or contract research organizations. At each stage of the validation process, results are immediately fed back into the predictor for full optimization of the predictive feedback loop. Lead peptides progress to optimization, including chemical modification to improve stability, delivery and availability.
The company delivers an accuracy rate of up to 60%; meaning that 60% of peptide predictions have successfully exhibited positive activity in in vitro and/or in vivo studies of biological relevance. This demonstrates the unique accuracy of the Nuritas platform, which, driven by machine learning, consistently improves.
“Our AI platform can get from concept to candidate or lead for optimization in as little as a year. This speed, compared to current norms where getting from initial ideation to the early animal-testing phase typically takes 4 to 6 years, really demonstrates how we can help the industry achieve its goals in a much better way. And of course, it also promises to reduce the time it takes to bring a product to market and most importantly to the patient,” said Paul Rohricht, CBO at Nuritas.
Working with Nuritas
Nuritas works with industry to collaboratively develop better products through its Discovery on Demand model, starting with the identification, by industry, of a receptor, target or phenotype for which modulation with a peptide may be useful. Nuritas then deploys its discovery engine to identify candidate molecules that are tested directly in Nuritas’s integrated wet laboratories. If successful in in vitro and/or in vivo models, patents are filed and transferred by license agreement to the industry partner for further preclinical testing and clinical evaluation.
Nuritas has also developed several preclinical assets that are available for licensing. These include proprietary peptides for type 2 diabetes, NASH, hypertension and antimicrobial applications. The licensing model is flexible and can be structured to meet the partner’s needs. Existing collaborations within its consumer division include multinational companies such as BASF, Nestle and PharmaVite (part of Otsuka Pharmaceuticals).
Wong, C. H., Siah, K. W. & Lo, A. W. Biostatistics 20, 273–286 (2018).
Fosgerau, K. & Hoffmann, T. Drug Discov. Today 20, 122–128 (2015).