How a digital backbone can transform industrial fermentation

Joseph Walker

Fermentation — one of humanity’s oldest scientific and industrial processes — is used today to create not just wine and cheese but also complex products such as renewable fuels, antimalarial medication, synthetic meat, flavor and fragrance molecules, and even spider silk. Yet while the industrial applications of fermentation continue to grow, underlying challenges to scaling up this fundamental process still remain.

And at the core of this lies a common challenge: a reliance on manual data collection and paper forms and notebooks, ad hoc process and parameter documentation, Excel spreadsheets with multiple versions that are difficult to track, inflexible ELNs and LIMs, and disparate data files from multiple instruments that are difficult to impossible to format and contextualize for statistical analysis.

Soaked in wine, drowning in data

Take, for instance, one of the most ancient applications of the fermentation process: wine making. It’s a wonder any bottle of wine costs less than $100 considering the amount of manual labor and headache involved in the process.

Most of the year is “easy” — grapes are grown, watered, protected from pests, and pruned. The magic of wine making really happens during four critical months: August through November, when grapes are harvested, crushed, and fermented to create wine. During this time, vintners collect hundreds to thousands of data points, moving from one fermentation tank to the next and walking down rows of hundreds of barrels, collecting measurements with various instruments to track the progress of the ferment. And it doesn’t end there — the contents of each fermentation tank are tasted and logged to compare notes across different tanks.

Most, if not all of these data are collected manually — jotted down in notebooks and paper forms as the vintner passes from tank to tank and barrel to barrel — only to later be re-entered into Excel spreadsheets or other digital forms for reporting and data analysis. It’s a tedious job, extremely time consuming, and prone to human error. But, it is critical for producing the next bottle of award-winning wine.

Fermenting disease treatments

With the successful production of human insulin by E. coli bacteria in the 1980s, it became clear that the millennia-old process used to produce wine (and other fermented foods and beverages) could have an even bigger impact on human health and well-being. Now, hundreds of biopharmaceuticals are produced via fermentation, and although the output is different, the process is essentially the same as that for winemaking (with a few key differences, of course): grow large amounts of bacteria or yeast in huge fermentation tanks, and then harvest and package the result.

When it comes to biopharmaceuticals production, the process must be carefully monitored — the final product eventually ends up in the bodies of humans to treat them for diseases, after all. Tanks must be sterilized, buffers must be synthesized, instruments and equipment must be set up and tore down, samples must be collected at regular intervals, and chromatography results must be collected and carefully analyzed. All of these processes, and every tiny detail associated with them, are recorded. It’s a ton of data, and it’s easy to get lost in a sea of paper forms, instrument log sheets, and batch records.

And, once all the data are collected, all of those forms must be carefully reviewed and validated by a human being, not a machine, in quality control.

Are you beginning to see the theme? Is your brain hurting yet?

Fermenting better products

The story is essentially the same within the walls of biotechnology companies. Instead of producing biomedicines for humans, companies in this space produce a range of products, from flavors and fragrances to biodegradable plastics. Just as with the production of biopharmaceuticals, hundreds of data points must be collected: bacterial strain characteristics, media ingredients, instrument logs, and the results of downstream analytics, such as analytical chemistry, proteomics, or metabolomics.

In all cases, the process starts small: benchtop fermentations are used to identify the best bacteria or yeast strains for producing the molecule of interest, whether that be an enzyme used to treat a disease or a biopolymer. Once the candidate strains are chosen, they are passed along for large-scale fermentations to optimize growth conditions and ensure the right molecules are being produced in sufficient quantities.

This means that multiple different teams, from strain engineering to fermentation to analytical chemistry, interact with the same strains. Each team also collects its own mountain of data, and to make the right production and manufacturing decisions, all data collected from all teams must be analyzed together and in context. Unfortunately, much of R&D and transition from process development to manufacturing is slowed by data preparation, not by data analysis.

Joe headshot

Riffyn Scientist Joe Walker has a decade of hands-on experience in industrial fermentation. Read more about Joe here.

Redesigning how science is done

I’ve personally experienced each of the scenarios described above, which comprise essentially a decade of my professional life. It was during my last position at a biotech company that, immersed in a sea of data awaiting analysis, I learned the importance of replication for statistical rigor, and how fundamental repeatability and reproducibility are when it comes to qualifying assays. I learned how frustrating it can be to debug code, and how rewarding it is is to drive decisions through elegant, impactful data visualization. Most importantly, I learned how to ask the right questions and the skills required to answer them.

That is why I now work at Riffyn. All of the headache I experienced wine-making, in biopharma, and in biotech could have been avoided had I had a tool like Riffyn Nexus at my disposal. With Riffyn Nexus — a cloud-based process data system — there are no more error-prone spreadsheets or paper notebooks. All data are stored digitally, automatically formatted and contextualized, and prepared for statistical analysis in seconds. The software’s UI enables users to see their experimental flows so they can quickly and easily trace samples through an entire process or set of linked processes. Sample naming is standardized, and all versions of an experimental design are accessible. You can even upload videos demonstrating specific experimental processes, bringing “lab notebooks” to a whole new level. It’s simple to share experiments (and therefore, results) with collaborators, and because everyone has access to the exact same data, there is no confusion around which data is the “most recent” or whether the spreadsheet you have is the same version as the one your colleague has.

As a scientist at Riffyn, I have the opportunity each and every day to utilize the skills and experience I’ve acquired over the last decade to tackle challenging problems and help our partners do better science in the fermentation space and beyond. When I see our customers achieve groundbreaking results, like reducing the time needed to prepare data for analysis 20-fold, reduce animal use by 80%, or bring products to market two times quicker with half the personnel effort, I know my work is making a difference.

I know the pain — and I know how good it can be when you have a better way. As science and particularly white biotechnology evolve faster than ever, the need for new tools and ways of thinking are inevitable. Riffyn Nexus isn’t just software. It’s a paradigm, and I truly believe it’s essential for the future of fermentation. What organisms will we leverage? What materials will we make? I’m excited to see where it takes us and to help solve new and existing challenges along the way.