CHICAGO (GenomeWeb) – Bayer is in the process of deploying an internal data-sharing platform that the company intends to become a single point of reference for all patient genotype and phenotype data in its biomarker-driven translational research.
Called PORTIN, which stands for Portal for Translational Data Integration, the system allows researchers working on Bayer-funded projects to search across studies, biomarkers, cohorts, and samples in an effort to shorten trial recruitment time, save money, and facilitate data reuse.
PORTIN is intended to address the common problem of data inaccessibility in research environments, according to Christiane Unger, an IT business partner in clinical sciences at Bayer. In a presentation earlier this year, Unger alluded to an oft-cited 2016 survey from CrowdFlower that found that data scientists spent 80 percent of their work time collecting and organizing data rather than analyzing it.
"We have a clinical database, we have a database for exploratory biomarkers, and we have a database for our sample data," but those resources were essentially siloed, Unger told GenomeWeb. "We were looking for a solution supporting cross-study queries."
Bayer first thought about such a platform as early as 2013, according to Unger, who joined the German multinational drug company two years ago. But as recently as 2015, Bayer was unable to find a suitable product on the commercial market, so it sought partners to build a custom system.
The firm settled on PointCross Life Sciences, a privately held Bay Area firm with a platform to help pharma and diagnostics companies mine their own clinical datasets. Together, PointCross and Bayer developed PORTIN, which has been designed to comb databases in search of ready-to-analyze study data to answer trial-related questions and to re-analyze earlier trial findings. PORTIN also is meant to encourage collaboration among researchers for Bayer-funded projects.
"We need to understand which biomarker and sample data belong to the data from the clinical database," Unger explained. "We'd also like to understand where the other [Bayer] clinical trials show the same results or not."
Bayer conducted a proof-of-concept study on PORTIN in 2016, testing it with Phase III studies and 15 data domains in a single therapeutic area, oncology. The system officially went live in 2017 after expanding to eight studies, 2,900 subjects, and 12,000 samples across 25 data domains in both oncology and cardiovascular disease, according to Unger.
"The interest increased after the successful proof of concept," Unger said. "When we started the project, there were a lot of concerns and doubts. With the ongoing progress we could show, the interest became greater."
Bayer released PORTIN 1.0, its first "technical productive" version, near the end of last year. Unger said that the qualifier was necessary because of the relatively small reach of the platform at the time. "That's not the data volume that you expect if you want to work with such a portal," she said.
For 2018, Bayer scaled up and explored a "bulk approach," featuring 100 studies in three therapeutic areas, Unger said. Full business adoption should come next year, she said, though researchers in any Bayer therapeutic area may request access now. Unger also expects to finalize a formal operational structure and governance processes for PORTIN in the coming months.
As the portal rollout accelerates, Bayer is having to educate all of its researchers on how to capture data with an eye toward future reuse.
"The quality of the data [may be] fine for the original use case, but with all these new initiatives where we want to see data as an asset, we have to reach a different mindset that the data I am developing is not just important for me today for my purpose," Unger said. "There will be other use cases later on from others. Therefore, I have to be more careful with how I collect the data."
Meanwhile, the bioinformatics team within the pharma giant is considering further ways to expand PORTIN's reach and functionality, though given the proprietary nature of early-stage R&D, it is unlikely the portal would be offered industry-wide.
Unger said that she is looking at ways to bring in phenotypic data not just from research labs but from clinical settings as well. Bayer also is paying attention to privacy issues that heighten as more people gain access to databases. Beyond that, the company is not tipping its hand.
"We are open for additional amendments or extensions of PORTIN. Whether it means we will make the change within the portal itself or just [create] connections to others, that's not clear yet," Unger said.