Optimizing biologics manufacturing with data automation
Biologics are incredibly challenging to manufacture, especially at scale. Access to automated data can transform the industry.
The biologics revolution
Biologics are reshaping the frontiers of medicine, unlocking novel treatments for diseases that have long eluded traditional pharmaceutical approaches. These therapies, derived from living cells and biological systems, represent the vanguard in treating a vast range of conditions, from autoimmune disorders to deadly cancers to rare genetic diseases. They include mRNA vaccines, monoclonal antibodies, gene therapies, and immunotherapies – and leverage the way our genes, cells, and immune systems function, often enabling much more precisely targeted attacks on antigens and diseased cells than is possible with traditional drugs. 
Biomanufacturing challenges
Manufacturing biologics is complex. Conventional drugs, like ibuprofen or metformin, are typically well-characterized “small molecules”, synthesized from chemical reactions with predictable, consistent outcomes. Generics are possible because the process is easily repeatable, and it is straightforward to test the end product for quality. In stark contrast, biologics come from living organisms and are produced through complex biological mechanisms, resulting in an inherently nondeterministic process that is far more difficult to control, scale, and verify.

Biologics manufacturing generally begins with the cultivation of living cells, such as bacteria, yeast, plant, or animal cells. Chinese hamster ovary cells are commonly used, for example, due to their favorable safety profile, rapid doubling time, and ability to tolerate environmental changes. In the case of autologous cell therapies, however, the master cells are taken directly from the patient being treated, resulting in an even more delicate and time-sensitive process. The master cells are then transduced or transfected to engineer a desired change in their nucleic acids, allowing them to then serve as factories for producing specific DNA, RNA, antibodies, or proteins as they proliferate. The cell cultures grow in containers such as bioreactors that enable precise environmental control and monitoring of parameters including pH, temperature, oxygen levels, agitation, cell density, nutrient supply, and metabolic waste. When ready for harvesting, the cells are separated from their growth medium. The therapeutic protein is extracted and purified through mechanisms like centrifugation, filtration, and chromatography. 

Each step of this undertaking has its challenges, based on the sensitive and complex character of biological systems and the nondeterministic nature of their growth. Not only can slight fluctuations in heat, light, or pH profoundly impact the quality or yield of the product, but even in the most carefully controlled environment, some degree of variability in the product is inevitable – an issue that compounds at scale. And since biologics’ end products are difficult to characterize (in contrast to conventional drugs), process control and environmental monitoring play a much more crucial role in regulatory approval and product quality.
The critical role of data
As a result, in addition to the physical and logistical complexities they must juggle, biologics manufacturers also face the challenge of continuously gathering, monitoring, and analyzing huge amounts of process-critical data. This data – including bioprocess data, omics data, analytical data, and quality data – comes from a wide variety of disparate sources, ranging from lab equipment to handwritten observations to third-party systems. It can be structured or unstructured; digital or analog; text, image, or video; collected at one point in time, temporally, or continuously. Many of the collection processes are at least somewhat manual, time-consuming, and error-prone. Managing and making use of these large, heterogeneous, noisy data sets in a timely fashion is extremely difficult, made all the more so by the stringent security, integrity, and regulatory requirements drug manufacturers must contend with.

We built Fathom to help pharmaceutical companies solve this very problem. By combining revolutionary advances in machine learning and artificial intelligence with deep domain expertise in the data challenges life sciences manufacturers face, Fathom empowers its customers to automate their data, transforming it from an unwieldy problem into a powerful tool. Access to automated data not only allows biologics manufacturers to better monitor, understand, and optimize product quality and yield, it also enables a host of other transformative benefits, from accelerated batch releases to streamlined regulatory approvals to faster development of new therapies.