AWS Architecture Monthly

Streamline Innovation in Biopharma with AWS
Evidation Health Enables Next Generation Clinical Studies on AWS
PwC: Building an Efficient, Intelligent, Serverless Pipeline for the Pharma Industry
Bristol-Myers Squibb: Enabling DevOps with AWS Service Catalog

Ask an Expert

What are the general architecture pattern trends for biopharma in the cloud?
 
Biopharma has long been on the cutting edge of cloud adoption—some of the earliest customers of AWS were in this space. Thirteen years ago, I was in biopharma myself. Personally, I used Amazon Elastic Compute Cloud (Amazon EC2) and Amazon Simple Storage Service (Amazon S3) to obtain resources that were difficult to get. Storage had to be forecast at least a year in advance—so if unplanned projects came up, we would scramble to buy desktop hard drives. Compute resources took at least 3 months to be ordered, delivered, and installed, which limited the ability of researchers to experiment with new ideas. Today, we’re seeing cloud adoption across almost all efforts in biopharma. For example, development of drugs and therapeutics is data intensive—especially with the emerging use of artificial intelligence (AI)/machine learning (ML). While customers are now able to access all of the compute and storage resources they need, now they’re looking for expanded access to their critical data. Patient data, trial data, data from earlier research, “real-world” data, and more are required for development, manufacturing and distribution, commercialization, and monitoring effectiveness and safety. As a result, biopharma has been adopting a more decentralized approach to data sharing and consumption. This includes data mesh architectures and methods for fully federated analytics and compute. These architectural patterns give biopharma organizations access to necessary data without having to exhaustively centralize it first. This helps them to more easily overcome internal and external organizational boundaries.
 
When putting together an AWS architecture to solve business problems specifically for biopharma customers, what are some of the considerations?

Helping our customers deliver safe and effective products while maintaining compliance with government regulations and industry standards is critical. They need to closely adhere to regulations and standards from the moment a researcher records their idea in a lab notebook to establishing the specifics of manufacturing and distribution. For obvious reasons, governments around the world regulate the safety and effectiveness of everything that can be prescribed to patients. This means our biopharma customers need to submit products for approval, retain complete records of the product's development, and enforce rigid controls to ensure consistency and safety. It’s a balance to ensure that these considerations do not become constraints. They should be implemented as a complementary, functional layer within every biopharma customer's architecture.

The next thing to consider is the diversity of workloads. There is no common architectural approach for the process of developing and delivering a drug or therapeutic in biopharma (also known as the “value chain”). The value chain spans fundamental scientific research to post-market surveys of therapeutics for safety and effectiveness with things like manufacturing along the way. The architectural requirements for each step are very different. In early research, tightly coupled high performance computing (HPC) is active for workloads like computational chemistry or molecular modeling. This has a different implementation from the more familiar analytics we support on Amazon EMR used during the commercialization of drugs and therapeutics.

Whatever architecture we help customers implement, we need to create something adaptable and extensible and not just another “silo.” The architecture should address their specific requirements but fit with all the other architectures being used to support workloads at other points of the value chain.
 
Do you see different trends in biopharma in the cloud versus on premises?
 
Not especially—as I noted earlier, biopharma has seen significant adoption of the cloud already. This experience has fed into overall design considerations. Now, most new on-premises deployments are capable of being hybridized or at the very least, extendable into the cloud. For the foreseeable future, most new workloads will continue to be deployed into the cloud. Those workloads with extremely low-latency requirements, like those seen in robotic laboratories or in manufacturing, will require on-premises infrastructure. We’ve already seen the trend for on-premises infrastructure to be cloud-connected, and I expect that to evolve into hybridized implementations like AWS Outposts.
 
What’s your outlook for biopharma, and what role will the cloud play in future development efforts?
 
My outlook for biopharma is overwhelmingly positive—I've always felt it is one of the most scientifically exciting domains I could work in (the other being space 🙂). Even with the current wide adoption of cloud, we’ve only just scratched the surface of what can be accomplished within biopharma. With emerging technologies and developments for AI and ML, the cloud is becoming more essential due to the requirements for extreme scale, flexibility, and data. These technologies are already beginning to be applied to difficult problems in biopharma like computational chemistry, computational biology, toxicology and so many more. Beyond that, the cloud can accelerate every point along the value chain. It can deliver therapeutics more quickly and precisely and ultimately help close the loop between biopharma and patients to accelerate early research and development.
 
Anything else you’d like to add?

COVID-19 has been a devastating worldwide pandemic. But the rapid development, approval, and ongoing distribution of the vaccine is one of humanity’s greatest scientific and industrial achievements. One of reasons it was so quickly developed is the cloud. Early in the pandemic, all observational and laboratory data regarding COVID-19 was exchanged and analyzed using the cloud. As the vaccines were being developed, all of the modeling, toxicology, and construction of clinical trials was performed using the cloud. Even prior to obtaining approval, the supply chain for the vaccine was established, the manufacturing capability was expanded, and the distribution network was built out using the cloud. Going forward, I know we will continue to apply what we have learned from this pandemic. And we will apply the core mission of our biopharma customers: to improve and help save the lives of all patients, everywhere.

patrick-combes
 
 
 
 
Patrick Combes
Tech Director
Healthcare and Life Sciences, AWS
Patrick Combes is the Worldwide Tech Leader for Healthcare and Life Sciences at Amazon Web Services. He began his career in scientific computing at Fermi National Accelerator Laboratory (FNAL) where he developed embedded systems for gathering and processing instrumentation data in real-time. In 2005, he joined Rosetta Biosoftware (Merck) to build and integrate products for genomics and proteomics into HPC environments. He was part of the Amalga Life Science Platform within the Health Solutions Group (HSG) at Microsoft and then led the industry solution architecture team for EMC as the Principal SA for Life Sciences and HPC. Patrick has a B.S. in Computer & Electrical Engineering from the University of Illinois at Champaign-Urbana with additional graduate work focused on programming language design.

Previous Issues


     Click images below to view or download past issues 


AWS Architecture Monthly provides new and curated content about architecting in the AWS Cloud. Our goal is to provide you with the best new technical content from AWS, from in-depth tutorials and whitepapers to customer videos and trending articles. We also interview industry experts who provide unique perspectives about the month’s theme and its related AWS services and solutions.

To get issues of AWS Architecture Monthly:

1) Open and download a PDF from the carousel above for past issues
2) Kindle Newsstand: Free subscription, available in the USA, the UK, France, and Germany
3) Flipboard: Personalized mobile magazine app that you can also read on your PC