The potential to transform bioprocess development is found at the crossroads of biology and advanced technology. As the biotech industry continues to evolve, harnessing data-driven innovations is key to improving biologics manufacturing by helping researchers make informed decisions and accelerate the discovery of new therapies.
With more accessible technological tools, the focus is now on creating reliable datasets to unlock the full potential of these innovations in biological research.
This concept is discussed in greater detail with Jeremy Agresti, the founder and Chief Technology Officer of Triplebar Bio, a company revolutionizing biologics development through high-throughput screening and AI-driven bioprocess innovation. In an episode of the Smart Biotech Scientist Podcast, hosted by David Brühlmann, founder of Brühlmann Consulting, they discuss the transformative fusion of AI and biotechnology.
Jeremy’s work is changing how biologists approach biologics manufacturing, especially with proprietary hyper-throughput screening platforms. In this article, we explore Agresti's insights regarding bioprocess development, the science behind his work, and his journey in biotech.
High Throughput Screening and Mutagenesis Techniques
Creating Diversity Through Organisms
At Triplebar Bio, one of the core techniques used to drive bioprocess development is high throughput screening. This method involves creating a diverse population of organisms, such as 10 million microbes, which can be used to discover novel traits or improve biological processes.
Agresti explains that these organisms are generated through techniques like whole genome mutagenesis. Rather than engineering individual changes one by one, his team uses random processes to create mutations, a strategy with two primary benefits:
- Speed: Millions of mutations can be generated quickly.
- Flexibility: Solutions can come from anywhere within the population.
Exposing a population of organisms to environmental stressors, such as UV light, can cause many mutations, making it possible to screen for desirable characteristics quickly and cost-effectively.
The Problem with Scale-Up in Biotech
Rethinking Scale-Up Challenges
Regarding the scale-up process in biotechnology, Agresti offers a controversial perspective. He suggests that most problems in scaling up production processes are not inherent to the scale-up itself but are rather a result of suboptimal inputs. Specifically, Agresti argues that if better cell lines and strains were available, many of the issues faced during scale-up could be avoided altogether.
His view is grounded in the belief that engineering strains specifically for manufacturing would significantly reduce the challenges encountered at large scales. By addressing these problems at smaller scales, biotech companies could better prepare for the demands of scaling up production without encountering delays or inefficiencies.
Early Problem-Solving for Better Results
Agresti emphasizes the importance of solving problems when there is more design freedom. Early in the development process, researchers have greater flexibility to modify and engineer biological traits in an organism before moving to the complexities of large-scale production, where manipulating environmental factors like temperature and oxygen transfer becomes more challenging.
Jeremy Agresti's Professional Journey
Early Career and Influences
Jeremy Agresti's journey into bioprocess development began with a foundation in biochemistry and genetics. During his undergraduate studies, Agresti was drawn to evolutionary biology and population genetics, particularly in natural populations of fish and aquatic species. This early interest in science was coupled with a fascination for the tools and technologies used to answer complex biological questions.
His path took a significant turn when he joined a lab at Cambridge University led by Greg Winter, who had invented the groundbreaking technology of phage display. Phage display became a pivotal tool for antibody discovery and development. Agresti was part of a subgroup within Winter's team, working on antibody-target interactions and applying similar evolutionary principles to other biological functions, such as enzyme and catalyst development.
The Concept of Miniaturization and Emulsions
During his time in Cambridge, Agresti was exposed to miniaturization. His work focused on developing biological systems at smaller scales, specifically using emulsions. Emulsions are systems where two immiscible liquids, like water and oil, are combined to create microdroplets that can be used as reaction vessels for biological processes.
Agresti and his colleagues explored how these droplets could be manipulated to perform various biological functions, similar to how test tubes are used in traditional lab settings. This led to innovations in microfluidic technologies, which allowed them to create controlled environments at the cellular scale.
Advancements with Microfluidics and Controlled Emulsions
After completing his postdoctoral fellowship, Agresti transitioned to a more practical application of his research by joining Dave Waits' lab at Harvard University, where he worked on advancing the manipulation of emulsions using microfluidics.
This field involves using patterned channels on a glass substrate to control the formation and manipulation of droplets at very precise sizes. The controlled emulsions allowed Agresti and his team to replicate biological functions within microdroplets, a concept inspired by the ideas of high-throughput biology from his time in Cambridge.
At this stage, Agresti and his team at Harvard began developing tools to manipulate and control these microdroplets, adding the necessary components for biological assays, mixing, and measurements. This work laid the foundation for future applications in genomics and biotechnology, including single-cell genomics and large-scale screening.
Application of Microfluidics in Biotechnology
Agresti's transition to the biotech industry came when he joined Amyris, a company initially focused on biofuels but later pivoted to synthetic biology.
At Amyris, Agresti applied his knowledge of microfluidics and emulsions to engineer organisms that could produce renewable hydrocarbons, such as diesel fuel replacements. This experience gave him a deep understanding of how large teams can be mobilized to solve complex problems with innovative technologies.
While at Amyris, Agresti was inspired by the company's problem-solving approach and ability to attract top-tier talent. He learned the value of working with a motivated, skilled team to tackle ambitious goals, an experience that shaped his leadership style and future endeavors.
Learning at Bio-Rad
After his time at Amyris, Agresti worked at Bio-Rad, a company known for providing tools for biological research.
At Bio-Rad, Agresti gained invaluable insight into product development. He learned to translate customer needs into product specifications and structure research and development efforts to meet market demands. This experience solidified his understanding of bringing innovative products to market, which became crucial when it was time to found his own company.
Founding Triplebar Bio
Agresti's journey culminated in the founding of Triplebar Bio, where he could combine all the knowledge he had accumulated about high-throughput screening, microfluidics, and bioprocess development. The company focuses on engineering organisms for biologics production to streamline the process and remove the inefficiencies associated with traditional scale-up challenges.
By leveraging his expertise in evolutionary biology, microfluidics, and product development, Agresti and his team at Triplebar Bio aim to unlock the full potential of biological systems and advance the future of biologics and synthetic biology.
Product-Focused Innovation in Biotech
Jeremy Agresti's company, Triplebar Bio, utilizes cutting-edge technologies to unlock biology's full potential. At the core of the company's mission is the application of innovative solutions to meet product needs in food and pharmaceuticals. The company leverages its technology to address these challenges by identifying areas that require product innovation.
Triplebar Bio applies technology to problems where traditional scientific methods struggle to provide solutions. One key area the company specializes in is high-yield protein production in organisms like yeast, especially proteins that are difficult or expensive to produce using conventional methods, such as animal agriculture.
The Screening Process: A Hybrid Approach
A significant part of the company's success lies in its unique approach to screening. The screening process combines physical laboratory experiments and advanced computational tools to optimize results.
Here's how it works:
- Diversity Creation: The company creates a diverse population of organisms—up to millions of variants—using techniques like whole genome mutagenesis. This allows the introduction of random mutations into the organisms, speeding up the discovery process.
- Screening for High Performers: These organisms are then tested in micro bioreactors, where their properties (e.g., protein production) are measured. The most promising candidates are identified as "high producers."
- Computational Design: After identifying high-performing organisms, the company applies computational tools to analyze the genetic makeup of these variants. By examining millions of sequences, they can pinpoint the key genetic mutations contributing to higher protein production. The data gathered is used to further optimize future iterations of the organisms.
This hybrid approach—lab-based screening with computational analysis—allows the company to progress significantly in areas like enzyme design, as demonstrated in various scientific publications.
I think that most problems in scale up are not really problems with scale up. I think they're problems with the input [...] If we had better cell lines and strains, I don't think many of the problems of scale up would exist. Usually we think of scale up as being this difficult thing and it's very expensive and of course there's a lot of science and knowledge that's behind it.
Scaling from Discovery to Production
One key challenge in bioprocess development is ensuring that discoveries made at small scales apply to larger production scales. Jeremy Agresti challenges the conventional wisdom that scaling from small-scale experiments to large-scale production often leads to decreased predictability.
In Triplebar Bio's case, the company has found that the transition from miniaturized screening to larger-scale production is surprisingly predictive.
This is largely due to the following factors:
- Biocompatible Interface for Organisms: When working at the microscale, organisms do not perceive the small scale as a constraint. They grow and behave like they were in larger tanks, making smaller-scale results reliable when scaled up.
- High-Throughput Screening: At the microscale, the company can screen millions of variants in a single experiment, testing various conditions simultaneously. This high throughput enables the identification of robust solutions critical for ensuring scalability to larger production systems.
- Screening Under Realistic Conditions: The company can select candidates that will perform well in large-scale bioreactors by screening at a miniature scale while simulating the conditions of large-scale production systems—such as oxygen limitations and pH variations.
One of Triplebar Bio's key advantages is its ability to perform detailed testing at small scales under diverse conditions and still predict performance at larger scales.
Microbioreactor Technology: Flexibility in Screening
The company's use of micro bioreactors is crucial to its success. Microbioreactors allow for the flexible screening of organisms under various conditions.
These small, controlled environments enable the team to test organisms in different types of growth conditions, including:
- Batch Conditions: The organisms can grow without continuous nutrient input, mimicking a typical batch fermentation process.
- Perfusion and Feeding: Sometimes, the team utilizes perfusion systems or continuous feeding to simulate different growth phases and nutrient availability.
This flexibility in screening allows Triplebar Bio to "ask" different questions of the organisms, ensuring that the selected candidates are optimized for their intended use.
Adaptability Across Cell Types
Triplebar Bio's technology is adaptable to various cell types, including bacteria, yeast, and mammalian cells. The company works with multiple cell types in biologics production, including mammalian cells and cells used in cultivated meat applications. While mammalian cells are more complex and delicate than microbes, the company's system successfully accommodates them, optimizing traits across diverse cell types.
However, the company has found that single-cell organisms, like yeast and certain bacteria, are more effective in expressing proteins than filamentous fungi, commonly used in specific bioprocessing applications. This is primarily because single-cell organisms are easier to manipulate and evolve for optimal production traits.
The Intersection of AI and Biology in Bioprocessing
Introduction to AI and Biology
Artificial Intelligence (AI) is increasingly making its mark in biotechnology, particularly in protein prediction and biologics manufacturing. The use of AI in predicting protein structures and understanding the language of biology has opened new possibilities for bioprocessing.
The sequence of amino acids in proteins and nucleotides in genetic material shares similarities with the sequences of letters and words in the language. This is why AI tools used in natural language processing also apply to biological data.
Developing Bioprocesses for New Molecules
Developing a new molecule typically begins by identifying partners with complementary expertise. Biotech companies, especially those working on developing dairy proteins, may have strong capabilities in selling and formulating their products but may lack microbial development teams. In such cases, they seek partners to help overcome these gaps and bring the molecule to fruition.
Once a partnership is established, the initial step is conducting a techno-economic assessment. This analysis helps determine the appropriate scale and efficiency needed for production to ensure that the costs align with market realities. For example, proteins sold at low market prices may not be suitable candidates for precision fermentation, which has its own production costs. However, proteins with higher market prices are more likely to meet the economic criteria.
Once the economic feasibility is determined, the next step is to analyze the protein's characteristics. Key considerations include the importance of post-translational modifications (PTMs) to the protein's function and whether these modifications can be achieved in a heterologous system (e.g., microbial fermentation).
If the modifications are mammalian-specific, the development path may need to be adjusted. From there, the gene encoding the protein is transferred into an expression system to evaluate its performance under evolutionary pressure, which allows for developing a robust production plan.
The Role of Contract Development and Manufacturing Organizations (CDMOs)
After the development work is completed, the next stage involves scaling up production. Biotech companies typically work with Contract Development and Manufacturing Organizations (CDMOs) to produce the molecule at scale.
These organizations have the facilities and expertise to handle large-scale production, although sometimes a partner company with its own strain or production capability may be involved. For example, a partner may opt to build or access production space if they have confidence in the strain's ability to perform at scale.
A key challenge in this area is the limitation of large-scale production facilities and a shortage of available inputs. However, if a company has a diverse portfolio of vetted products and suitable organisms to produce them at scale, financing the construction of new production facilities becomes much more feasible.
High-Quality Data and Its Importance for AI Predictions
AI's effectiveness in biotechnology heavily relies on the quality and quantity of the data it is trained on. For AI to be effective, it requires large, well-annotated datasets.
In the field of biology, high-quality data is often scarce, though there are certain datasets, such as those for proteins and protein structures, where AI has made significant progress. AI has already proven effective in predicting protein structures, offering optimism that similar approaches can be applied to other biological challenges.
To address the challenge of data availability, some companies have developed screening tools that help associate genotype information with functional outcomes. These tools allow next-generation sequencing to count various species (or types of genetic information) as they pass through the system. Doing so creates a dataset that links genetic sequences to their corresponding functions. This data can then be used to train AI models and generate predictions about the behavior of genetic sequences.
A preprint published in collaboration with Google DeepMind demonstrated the success of this approach for enzyme development. Using AI to predict how genetic sequences will function can revolutionize areas like enzyme development, pharmaceutical discovery, and therapeutic design.
The language of biology, you know, it's a sequence of amino acids or a sequence of nucleotides. Language is a sequence of letters and words. Those things are at some level the same. And so the same kinds of tools can be used to make predictions out of those data sets.
High-throughput, Noninvasive Methods
One key advantage of the tools being developed is their ability to conduct high-throughput screening noninvasively without requiring large quantities of material. Through miniaturization, large datasets can be generated rapidly and inexpensively, significantly reducing these experiments' cost and material requirements. This allows for the screening of thousands or even millions of genetic sequences, generating valuable data that would be otherwise difficult or costly to obtain.
In traditional biotechnological methods, such as using liquid handling robots with high-density well plates, screening tests are limited to a few thousand per day. Furthermore, these methods can be costly, requiring large materials and specialized equipment. In contrast, the developed miniaturized systems can generate data much more efficiently, driving down the per-test cost to mere pennies, making it possible to scale up genetic studies to millions of samples.
Overcoming the Data and Cost Limitations
The traditional high-throughput methods for genetic screening are expensive, with the cost of sequencing a single genotype variant reaching $100 or more. Scaling this to millions of sequences would lead to costs in the range of $100 million. These limitations make it difficult to generate the large, annotated datasets necessary for AI-driven discovery.
However, with the advent of miniaturized screening tools, the cost of sequencing and labeling each genotype can be significantly reduced. This breakthrough makes large-scale studies more feasible and opens up the potential for AI applications to become more widely accessible, especially in biotechnology.
The Future of AI in Biotechnology
Looking to the future, AI and biotechnology are poised to make remarkable strides, especially in predicting protein function and designing genetic sequences. Currently, AI is effective in predicting protein shapes but still has limitations in predicting interactions or functions. The frontier lies in associating genetic sequences with functional data—predicting how an antibody binds to its target and how it triggers a receptor or activates T cells.
AI has the potential to revolutionize the design of genomic sequences. With enough data, designing genomes that carry out specific functions may soon be possible to enhance biotechnological applications or better understand diseases. This shift could lead to profound advances in disease diagnosis and personalized medicine, where AI models could predict the genetic makeup responsible for specific functional outcomes.
The next few years hold tremendous potential for breakthroughs in AI and biotechnology. While much progress has been made in using AI to predict protein structures and streamline bioprocessing, there is still much to be done in scaling up data collection and refining AI models for broader applications in drug development, genetic research, and disease treatment. As these datasets grow and evolve, AI will play an increasingly pivotal role in unlocking new possibilities for biotechnological innovations.
Final Remarks
In this insightful conversation, Jeremy Agresti shared his perspective on how artificial intelligence revolutionizes biologics manufacturing, emphasizing the importance of high-quality, large-scale datasets.
He discussed how miniaturization transforms bioprocess development by enabling the testing of numerous candidates at a fraction of the cost, essential for making meaningful predictions using AI.
With the evolving capabilities of AI, particularly in predicting protein structures and potential functions, the future of biotechnology looks promising. As data quality and quantity continue to improve, these advances could unlock breakthroughs in understanding complex biological systems and facilitate faster, more efficient development of therapeutics.
The key takeaway from the discussion is the crucial role that comprehensive datasets will play in driving innovation and ensuring the full potential of AI in biotechnology is realized.
About Jeremy Agresti
Jeremy Agresti is the Founder and CTO Triplebar Bio, Inc. leading the R&D team bringing a critical innovation to the bioeconomy with a proprietary Hyper-throughputTM Screening Platform to accelerate and innovate food & pharma product discovery. Jeremy is a pioneer in the space of microfluidics with more than 25 patents and 7,500 citations from his work. Licenses from his work have been taken to start companies like 10X Genomics. Prior to founding Triplebar, Jeremy led the R&D team at BioRad developing tools for various applications and prior to BioRad, Jeremy was at Amyris working on synthetic biology applications to biofuels and other products.
Jeremy is also a co-founder of Slingshot Bioscience which uses a proprietary synthetic cell system called FlowCytes™ for scaling manufacturing. Jeremy holds a Ph.D. from MRC Laboratory of Molecular Biology Cambridge UK followed by a post-doc at Harvard University focused on pioneering microfluidics to engineer biology. Jeremy tends his urban orchard of 30+ fruit trees and loves to care for his vegetable plots. One of his favorite gardening activities is sharing tomato seedlings with the Triplebar team. He and his family love to go on food adventures around the Bay Area, LA, and anywhere they travel.
Connect with Jeremy Agresti on LinkedIn.
David Brühlmann is a strategic advisor who helps C-level biotech leaders reduce development and manufacturing costs to make life-saving therapies accessible to more patients worldwide.
He is also a biotech technology innovation coach, technology transfer leader, and host of the Smart Biotech Scientist podcast—the go-to podcast for biotech scientists who want to master biopharma CMC development and biomanufacturing.
Hear It From The Horse’s Mouth
Want to listen to the full interview? Go to Smart Biotech Scientist Podcast.
Want to hear more? Do visit the podcast page and check out other episodes.
Do you wish to simplify your biologics drug development project? Contact Us
