Differential Bio emerges from stealth to de-risk bioprocess scaleup

Scaling bioprocesses is “notoriously slow and prohibitively expensive,” says Differential Bio CEO Christian Spier. “We’re here to change that.”

Emerging from stealth this week with €2 million ($2.2 million) in pre-seed funding, the Munich-based startup has developed a platform integrating AI, microbiology, and lab automation to help firms cut costs and de-risk the biomanufacturing scaleup process.

The round was led by Ananda Impact Ventures and ReGen Ventures with participation from Carbon13, Climate Capital, Better Ventures, CDTM Ventures, and angel investors, and will help Differential Bio expand its platform and team and attract new clients in food, cosmetics, and specialty chemicals.

“Biomanufacturing is constrained by a scaling crisis. By leveraging machine learning on extensive phenotypic and bioprocess data, Differential has the potential to redefine biomanufacturing at scale by providing a step function reduction in cost and time to commercialization.” Tom McQuillen, partner, ReGen Ventures

De-risking the scaleup process

Three core innovations underpin the platform, said Spier:

• Advanced microbiology to miniaturize fermentation processes
• Robotics to automate lab workflows and generate high-quality data
• AI algorithms to simulate and optimize bioprocesses virtually

In combination, they minimize reliance on costly and resource-intensive physical experiments and deliver precise insights into factors such as biomass growth rates, metabolite production, and cell viability, he explained.

“The outcome? Faster optimization cycles, significant cost reductions through automation, and enhanced profitability thanks to higher yields and improved process efficiency.”

Probiotics case study

As an example of its virtual platform in action, Differential Bio worked with a client to transition a microbial strain from animal-based to plant-based growth media, achieving a 4x increase in biomass yield, boosting production efficiency, and delivering a 16% reduction in production costs.

“This project showcased the platform’s ability to handle multi-objective optimizations,” CSO Dong Zhao explained. “We proved that sustainable biomanufacturing can enhance performance and cut costs simultaneously while achieving sustainability objectives.”

Self-driving lab

Speaking to AgFunderNews at the Future Food-Tech summit in San Francisco, Spier said: “The promise of data generation at scale and AI is huge. But to go from lab-scale to pilot to commercial scale, the cost per experiment increases exponentially. So you have to go to the micro scale to generate data and then bridge the gap to large-scale.

“So we have a fully automated lab with liquid handling robots, and we use that rich data to train our models,” added Spier, a bioinformatician who is based in Munich but has lived and studied in the US.

“We’re simultaneously looking at optimizing the media, the strain, the mechanical bioprocess, the downstream processing, and scale up.”

When it comes to microbial fermentation, he said, finding cheaper alternatives to purified sugars such as glucose that will work as feedstocks for microorganisms is becoming increasingly important, and Differential Bio’s tech could help companies identify cheaper carbon sources that will work with their strains.

“These side streams and waste streams [that could serve as alternative carbon sources] are complex and heterogeneous, but the potential is huge. We generate the phenotypic experimental data that the industry is lacking in house and map that so we can establish the connection between the microorganism and feedstock.

“And then on the downstream processing side, which can also have a huge impact on profitability, we can work on process stability and for instance figure out how many cells survive after freeze drying [an important metric for probiotic bacteria, for example] and use that as training data for our model.”

In short, Differential Bio can help firms de-risk their bioprocess at a small scale before they spend exponentially more money testing it at a larger scale, he said.

According to Spier, Differential Bio has been working with multiple microbial strains including E. coli, yeast and lactobacillus and is now “onboarding customers in the mostly probiotic spaces, where we see a lot of demand, especially when it comes to next-generation probiotics.”

Business model

Initially, said Spier, “We are effectively doing the work for our clients, from data generation to modeling to optimization, and working directly with their proprietary strains, for example. We do that to show the breadth of our approach in a short period of time and generate case studies proving our optimization capabilities, next to expanding the capabilities of our technology (more strains, more bioprocesses, reduce data requirement over time).”

Over time, however, the business model will evolve, he said. “We are launching the platform making it available to partners so that they can generate and learn from data in-house, without sharing sensitive data with us. We have a very easy to use web platform, a virtual experimentation platform, where you can drag and drop, adjust parameters and conduct virtual experiments. You don’t have to know any data science or have programming skills.”

In general, said Spier: “Huge corporates have their own processes around data sharing, which can prolong decision cycles, but in general, there’s an understanding that the more you provide, information-wise, the higher the likelihood of success. There are also ways to work together if you don’t want to share data with us, though. So we can work with a closely related strain that is publicly available, for example.

“We can also offer pre-trained models that you can use to fine tune to your strain and apply to your use case, and we never see your data.”

‘We need data’

Stepping back, he said, there are multiple enabling technologies that can improve the economics of biomanufacturing, from new tools to domesticate a broader array of microbial hosts (Wild Microbes), to cheaper and more sustainable feedstocks (Hyfé), cell-free production (Debut), continuous processing (Pow.bio, Cauldron), novel bioreactor designs (Biosphere), AI/ML to help optimize cell lines and bioprocesses (Triplebar), and more productive cells (Enduro Genetics).

“I think non-model organisms [beyond the standard workhorses of biomanufacturing] have huge potential, including hosts that use carbon dioxide as feedstock [via gas fermentation].

“But with all of these things, we need data. How do you do data driven optimization if you don’t have enough data? So that’s where we can help.”