Fi Global Insights is part of the Informa Markets Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them. Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.

Multi-layer analysis of microbes and gene candidates speeds up product development

Article-Multi-layer analysis of microbes and gene candidates speeds up product development

© iStock/Sam Edwards child drinking milk, dairy, iStock Sam Edwards, iStock-170510614- RS.jpg
Biotech startup Evogene’s computational predictive biology platform is accelerating and derisking product discovery and development in the fields of probiotics and animal-free dairy proteins.

In April, Israeli biotech firm Evogene and The Kitchen FoodTech Hub, the foodtech incubator and investment arm of Israeli dairy giant Strauss Group, jointly announced the establishment of the company, Finally Foods.

They described the venture as “an AI-driven company specialising in molecular farming for the food sector”, with the goal of “providing sustainable alternative sources to animal-based proteins”. The company is modifying plants as bioreactors to produce these proteins, and, by leveraging Evogene’s GeneRator AI technology, is aiming for short R&D cycles and rapid time to market.

In an interview with Fi Global Insights, Evogene’s executive vice president of business development, Eyal Ronen, explained why he thinks plant-based genomics hold the key to viable production of alternative proteins.

“When cultivating meat in a bioreactor or fermenter, all kinds of growth factors and regulators are needed to increase cell division. These are currently super expensive, plus there are issues with scale up and contamination. Using molecular farming these elements can be grown in plants at a fraction of the cost and many of the challenges around bioreactors can be overcome,” he said. 

Dairy protein from potato

In the case of Finally Foods, the protein is casein and the plant is potato. In simple terms, a gene is inserted into the potato tuber that changes its DNA to produce casein when grown.

Where Evogene’s GeneRator AI technology comes into play is in identifying and selecting functionally relevant gene candidates for incorporation into the potato tuber, as well as accelerating the development of the plasmid.

“The plasmid is the delivery vehicle that transfers the genetic information into the genomics of the plant. So designing the plasmid, trying to target where to insert it, trying to measure the inferences  for the rest of the genome - all of those things could be done computationally. And you can go for prediction before you start the wet process, so that is shortening the time between discovery and commercialisation,” said Ronen.

GeneRator AI is one of three software engines developed by Evogene for facilitating the discovery and development of life science-based products.

Screening microbes

The company’s microbe-focused MicroBoost AI tech-engine is being leveraged by Verb Biotics, a microbiome health ingredient company that is collaborating with Evogene on the advancement of new probotic strains.

The collaboration is focusing on identifying and enhancing the currently unknown genetic pathways in microbes that support the production of novel metabolites.

“Verb Biotics was looking for a certain molecule, but the bacteria it found it in cannot be consumed by humans. So we are looking for similar materials that are suitable for human consumption and are producing that molecule,” explained Ronen.

The third tech-engine is ChemPass AI, a platform for the discovery and development of small molecules in synthesised chemistry. Ronen said that this cross-sector reach is one of the major differences between Evogene and others in this space.

“We cover small molecules in chemistry, we cover the microbiome and we cover genetic information. I’m pretty sure that none of our competitors span all of those areas,” he said.

Another point of difference is the way in which Evogene screens candidates, Ronen explained.

“Our main competitor has developed high throughput analysis, which means they have developed a programme that can screen thousands of molecules very quickly, but that is a different approach to ours. Our screening takes a little more time but results in a more accurate outcome because it is intelligence driven. It uses multi-layer analysis to narrow and refine the search. It results in fewer candidates, but with a higher probability of commercialisation as it eliminates ‘false development’.”

Mining big data

Evogene’s database is built from data libraries the company has purchased, publicly available data and data generated internally, from trials and research carried out by its own laboratories or subsidiaries. The datasets it works from are vast - Ronen said the ChemPass AI database contains over 50 billion molecules, whilst the MicroBoost AI platform works through several hundred thousand microbes.

Working with ‘big data’ maximises the field of observation, said Ronen.

“Normally, when companies are developing a solution, they will approach a certain family of molecules, microbes or ingredients, which, based on their experience and on literature, they believe should hold a solution. But from a research perspective, that narrows the field of observation. We are screening a broader spectrum which means we have a great chance of finding a solution.”

Screening this volume of data with regular computational power is not feasible, which is where AI comes in. With Evogene’s platform analysis takes up to ten days - slower than rival platforms but far quicker than with regular computational power, which would take 10-20 years to screen the same amount of data.

The initial results are sets of molecules, microbes or genetic sequences. The second phase is optimisation, in which the software makes recommendations to enhance the candidates. This phase usually whittles the field down to three to five strong candidates.

“When they come out of that optimisation process, the chance of these candidates becoming commercial products is much higher than it would be with a regular R&D process,” said Ronen.