C4XD Builds Platform for Advanced Drug Discovery with Devoteam on AWS

C4XD, a drug discovery company founded in 2008 as a spinout from the University of Manchester, uses advanced technologies to develop medicines for immune-inflammatory diseases such as rheumatoid arthritis. It needed to update its cloud infrastructure on Amazon Web Services (AWS) so it could better analyze large genetics datasets without bottlenecks or cost inefficiencies, and asked for help from AWS Partner Devoteam. Working with Devoteam, C4XD designed a new infrastructure and migrated 13 TB of data to a more modern solution on AWS in just 13 weeks. This has reduced data storage costs by around 75 percent, eliminated bottlenecks in data analysis, and expanded C4XD’s capabilities so it can more easily work with different datasets and partners, opening new possibilities for future drug discoveries.

Using a unique combination of scientific expertise and data analysis, C4XD specializes in researching and developing small-molecule drugs to treat immune-inflammatory diseases. These pharmaceuticals have low molecular weights that can easily enter human cells and act on other molecules that contribute to disease. C4XD recently changed its research strategy to use precise analyses of large volumes of genetic data to target patients most likely to respond to different kinds of drugs for different diseases. And it aimed to build a new platform called PatientSeek to support that approach.

To enable that new strategy, C4XD needed to update the platform that it had built on AWS. It had adopted that platform after it migrated from its on-premises systems several years earlier, but it had not optimized its infrastructure to make the most of the cloud-native capabilities on AWS. As a result, it couldn’t easily scale compute resources to meet changing needs and had to pay for some services even when it wasn’t using them. It had recognized for several years that it needed to upgrade its infrastructure, but worried that such a project would be too difficult.

“Our previous platform just had a lot of outdated solutions, and a very fragile and complex infrastructure,” says Neil Humphryes-Kirilov, associate director of human genomics at C4XD, adding that the platform would require rebuilding from the ground up.

Humphryes-Kirilov’s team had contacted Devoteam on a recommendation from a software developer at C4XD who had worked with the consultancy on an earlier successful project. Devoteam, since acquired by Devoteam, is a UK-based firm that specializes in cloud migration and modernization on AWS. It quickly understood what C4XD needed to achieve its goals. “It was always this enormous task that nobody could do,” says Humphryes-Kirilov. “It was going to be too complicated. And then Devoteam came in and said, ‘This is what we need to do.’ They got right to the core of the issues.”

Before starting any development work, Devoteam organized a series of workshops over 6 weeks to clarify C4XD’s needs and goals. These workshops, which involved the entire genomics team, covered everything from compute and storage needs to how users would interact with the new platform. Even after the workshops were completed and development began on the new infrastructure on AWS, Devoteam consulted the genomics team frequently to make sure the new platform would operate as expected. And it worked to address unexpected quirks in the platform’s behavior, such as problems that arose when upper-case letters were used when requesting specific compute resources.

Devoteam also incorporated some of the recommendations it had come up with during a previous AWS Well-Architected review it had conducted for C4XD. One of the areas where Devoteam saw the greatest room for improvement was in C4XD’s use of data storage. “The majority of their bill was spent on storage that was just sitting there and really doing nothing,” says Ross Purdon, head of modernization at Devoteam. The solution was to move most of C4XD’s data—including about 5 TB of data in the network file system and 8 TB in the MySQL database—from Amazon Elastic File System (Amazon EFS) into Amazon Simple Storage Service (Amazon S3). C4XD can then use Amazon Athena, a serverless, interactive analytics service built on open-source frameworks, to analyze that data whenever needed.

Another change involved updating C4XD’s Taxonomy3 processing engine, which powers the PatientSeek platform, by using Amazon EMR to easily run big-data workloads. This integrates with Amazon SageMaker, which helps to build, train, and deploy machine learning models for C4XD’s data analyses. In addition, Devoteam used Amazon SageMaker to support the creation of different data silos for different users and datasets.

Developed over the course of just 13 weeks, the project also strengthened security. “We have a nice, granular layer of security, and data implemented and classified across all the different projects, so we know who has access and who doesn’t,” Purdon says. “We also have audit tracking as well.”

Moving 13 TB of data to Amazon S3 and the new PatientSeek platform has helped C4XD to reduce its data storage costs by about 75 percent. And all that data was successfully migrated in around 300 hours—just under 2 weeks—with zero downtime for the company. “C4XD are a very nice customer to work with,” says Purdon. “They also have a new DevOps guy who joined for this project, who’s got a lot of experience and exposure to AWS. And they’re even looking at bringing more of their on-premises infrastructure into AWS as well.”

Along with gaining an improved platform with a more user-friendly interface, C4XD has also gained the ability to run analyses on data vendors’ trusted research environments. In such an environment, a partner like C4XD isn’t allowed to download vendor data but can use its own algorithms within the vendor’s system and, with the vendor’s approval, export the results of that analysis.

With more data vendors moving to such environments, this was essential for C4XD’s new PatientSeek strategy and future needs, explains Humphryes-Kirilov. “There’s absolutely no way that we’d have been able to do that before,” he says. “But we’re planning in the near future to access data in that way. Having that portability is really a massive advantage.”

C4XD is a UK-based drug discovery company that combines scientific expertise with advanced technologies to research and develop new candidates for small-molecule drugs that target immune-inflammation diseases such as rheumatoid arthritis. Founded in 2008, it began as a spinout from the University of Manchester.