OSTHUS Insights

In 2013, a request for proposal to develop the Allotrope Data Format (ADF) was submitted by a consortium of top 20 big pharmaceutical and biotech companies, members of the Allotrope Foundation, seeking to pursue machine learning and artificial intelligence and enable the modern digital laboratory. The foundation requested a suite of specifications and products to amplify the data lifecycle in a highly regulated environment, using automation to generate and reuse scientific data of the highest possible quality and compliance.

In other words, 80 percent of the work data scientists perform today according to this Forbes Magazine report can be automated. In conjunction, data accuracy and transparency can be amplified to new heights which greatly reduces auditing and compliance costs wherever applicable.

„Currently, data of sufficient quality and quantity is not readily available to modern analytics platforms. But if you want to do machine learning, you need to provide the machine with data so that it can learn from it,” Chief Innovation Officer at OSTHUS and Leading Architect of the Allotrope Framework, Wolfgang Colsman, said. “High quality results depend on high quality data.”

The Allotrope Framework contract was awarded to OSTHUS in 2013. For the past five years, innovators at OSTHUS have developed technologies to meet the pharmaceutical companies’ specifications for a standard framework to acquire, share, and reuse data.

“In 2013, everyone said: this can not be done,” Colsman said, in reference to being selected to fulfill the Allotrope request for proposal. “Now that we proved this is possible, the Allotrope Framework has become a commodity that’s creating its own industry agnostic market.”

The success of the Allotrope Framework initiative, according to Colsman, means that every record of every data point derived from any instrument or data source will be accessible to humans and machines. Product releases include a software library for reading and writing Allotrope data files to streamline data capture and aggregation for the purpose of automation and machine learning.

Further amplifying the value of the ADF format, the Allotrope Ontologies developed by OSTHUS give machines a domain-specific vocabulary for identifying and classifying what is being stored. Colsman says this is the first time in history that the industry has attempted to standardize how semantic data is modeled. As a result of the project’s success, machines now have the ability to use data that comes pre-loaded with complex layers of human context.