Unlocking nature’s secrets? How Nuritas harnesses AI to uncover hidden health-boosting ingredients

13 Jun 2023 --- Biotech company Nuritas speeds up ingredient discovery in nutrition through AI, using techniques from the pharmaceutical industry where innovation and investment rates are much higher. Through its bioactive peptide finder, Magnifier NπΦ, the company can boast a clinical trial success rate of 80%.

Development of an ingredient goes through three phases, from discovery to preclinical biology testing and clinical trials.

“People focus on AI being something that it’s not – it’s math. It’s not a decision-making tool on its own. It needs scientific rigor and input to make it valid,” Dr. Andrew Franklyn-Miller, chief medical and innovation officer at Nuritas, tells NutritionInsight.

He adds that in nutrition, many ingredients used are hundreds of years old.

“Partly because the discovery process is so long and success rates are so low, investment in those areas is pretty patchy and certainly risky. Also, often with ingredients, the patentability of that ingredient doesn’t balance against the potential cost.”

The company offers two main products identified through its AI platform, PeptiStrong for muscle health and PeptiYouth with anti-aging properties.

Franklyn-Miller highlights that investment in nutrition is patchy and risky.Ingredient discovery and testing
Franklyn-Miller explains in a first discovery round, Nuritas’ Magnifier platform might find 700 potential peptides with a particular benefit which, after biologic confirmation rounds, might get down to 30 or 60.

“We’ve demonstrated 69% success at discovery, where we look for a target or a phenotype, which compares very well to pharma. That window is about 12 months, much shorter than the pharma lines. Then we go into some preclinical biology work to look at absorption in the gut and maybe some cellular changes, and then into a clinical trial.”

“Our clinical trials success to date has been 80%. It’s been so high because of the success of Magnifier and the front end in being able to hone down on target,” he emphasizes.

“Clinical trials allow our customers in a B2B sense to put claims on the packaging, which allows them to differentiate using our peptides to both improve their product but also, in some cases, to reduce waste in terms of being able to get more out of potentially a whey protein than otherwise.”

One of the challenges as a company is to remain focused, notes Franklyn-Miller. The company’s AI platform could also be used in multiple other areas and synthetic proteins or peptides.

“It’s easy to start to broaden out, but ingredients need a pipeline. If we broaden too wide, we cannot deliver on that pipeline and look for successive generations with broader benefits, reduced volume, or dosing with different clinical claims.”

Magnifier NπΦ peptide finder
Franklyn-Miller explains that Nuritas’ AI platform is a combination of complex data in terms of training sets from biological assays on the company’s mass spectrometry.

“We’ve taken a lot of natural plant sources and broken them down using proprietary enzymes to break down those peptides. Then we test in the laboratory what they do regarding up or down-regulating a receptor.”

Development of an ingredient goes through three phases, from discovery, to preclinical biology testing and clinical trials.Along with natural language processing – a combination of technologies that enable computers to process human language – of data scraping and curation, the company employs different models depending on what the company looks for, such as a specific phenotype or receptor.

“We start to weave in some knowledge graphs to unlock much of the data we had in those early archives in our peptide vault. The world has billions of peptides; we have the largest library with over six million from plant sources, which gives us this very secure dataset to work with.”

“Our modeling is very much geared off that proprietary peptide vault, which we’ve mapped out to be sure of what we’re doing.”

He continues that Nuritas is continuously working to improve its generative modeling to make the research more specific. The team also builds in high throughput assays in the biology lab to produce content more quickly.

Data quality limits speed
Franklyn-Miller notes that data quality requires a stronger focus. One of the significant limitations of natural learning models used in AI is that these go too fast and don’t address the quality of the data they’re working from.

“For us, quality assurance means we must be absolutely certain that these natural ingredients have both the benefits that they are expected to but also work in the same way with the same safety efficacy.”

“Without that quality of data, you could miss a huge treasure trove of information about pathways which up or down-regulate particularly vital functions that otherwise we would miss.”

Nuritas’ AI platform combines complex data regarding training sets from biological assays on the company’s mass spectrometry.He adds that biological confirmation of in-silico data limits AI’s expansion but is crucial for data accuracy. An in-silico study is performed via simulation on a computer to predict how a compound will react. Nuritas performs these biological tests in-house, but these take time.

“Mass spectrometry is where we look for the actual peptides within a substance. This is time-consuming because you’re drying the peptides and injecting them into a gas to identify them in a chamber. That process must be very accurate and done in triplicate, to lab grade standards.”

Health benefits from nutrition
Nuritas aims to improve the health of billions, highlights Franklyn-Miller. As consumers don’t like taking medicine, the company sees benefits in preventative care but wants to ensure this is safe.

“If we can give [consumers] medicinal-like benefits at much lower therapeutic levels, but across multiple areas of health, in blood pressure and sugar management, neurocognition, skin health and anti-inflammation, then we have the beginnings of being able to improve people’s lives as a byproduct of what they eat.”

In two clinical trials, Nuritas’ PeptiStrong performed better than traditional animal proteins in muscle protein synthesis. At the same time, it also improved strength recovery and restoration, reduced fatigue and positively modulated alterations in markers related to muscle homeostasis.

“Currently, our product focuses very much on muscle health,” continues Franklyn-Miller. “That has both an athletic and an aging population target because we know that muscles are the largest mass organ in the body.”

Though it takes time, biological confirmation of simulation data is crucial for accuracy. Wearable tech in clinical trials
Franklyn-Miller states that Nuritas is working to incorporate wearable technology in its clinical trials.

“I think the consumer is well informed now and wants to see the benefit. Feeling a benefit is no longer good enough – you need to be able to see it on your work band or your continuous glucose monitor. I think that’s something clinical trials have lacked in the past.”

The company has two products going into clinical trials in the next quarter, which will use wearable technology. One looks at sleep, anxiety and calm and the other at glucose control in the immediate aftermath of eating to try and reduce that big sugar spike and fall.

“We know with continuous glucose monitors the focus is on trying to reduce those spikes in the insulin demands,” adds Franklyn-Miller.

“One of the options for us in the future is to work with more partners with particular targeting needs; therefore, we can co-discover some of those early phases. If they’re looking for something particular such as the taste or stability of a product, or they want to work specifically in certain areas, like sugar metabolism, bone health, neurocognitive health or skin health.”

Nuritas hopes to have 24 ingredients across the next five years, addressing multiple areas of health.

By Jolanda van Hal

To contact our editorial team please email us at editorial@cnsmedia.com