The days of bridging are gone. Now it is full integration.

The era of merely bridging the gap between drug discovery and data science is over. We are now entering a phase of full integration, where these two fields are becoming inseparable.

The Historical Divide

When I started in the science business, I was lucky to be one of the rare people who could translate between languages: the language of biologists (cells, molecules and patients) and the language of the computer geeks (linux, bash, python, R etc).? Two worlds, almost always from two different corporate silos, often led by people with incompatible views of the world.? The data-people performed ‘services’ for the biologists or chemists (who always called the shots), often transactional in nature and often getting involved far too late.?

The change should have come before: Omics, imaging, wearables and other high-content and high-throughput technologies.? The opportunity to radically leverage these technological advances has been around for a while now.? But Pharma has not been equipped to fully reap the benefits of this data revolution yet.? Yes - we have big departments: Careers and fancy groups have been built. They are often called things like ‘Research Informatics’, ‘Data and Analytics’, ‘Digital Products’ or similar. ? Bridging to biological groups is done by employing translators called Business Analysts, Product Owners, Liaison leads, Value-stream leads etc. ? ? Often, leaders on the biology side of the bridge, have a feeling that they need and want more from the ‘analytics-people’, but not always getting it and not really the right thing.??The bridge is narrow and crossing is troublesome for both sides.

The Slow Shift Towards Integration

I believe it has to do with the slow percolation of true data-literacy among the pharma ‘biology business leaders’.? There has been, and still are, sentiments that data science is a speciality field that is and should be different from biological sciences.?

For younger scientists, this is changing: molecular biologists, chemists and pharmacists have for several years now been trained in data science at university: Coding and data analysis in R or python are no longer esoteric.? Unfortunately, I have seen that this reverts when they enter the pharma industry: the culture and senior colleagues do not expect them to analyse their own data in R. Computers are locked down and cannot be tinkered with, some have to go through a maze to get access to a programmatic analysis environment.? That is something taken care of by people in that other geeky department down the road. In the worst case data literate young biologists forget their skills and revert to Excel (oh no!). It is almost in the walls that advanced data analysis is someoneelse's business.??

This must change.

The Role of Data Literacy

“Everyone needs to be a data scientist. We need to increase our data literacy”.

Aviv Regev, the head of Genentech Research and Early Development, recently emphasized the importance of data literacy at Roche. This top-down endorsement is crucial for driving change across the organization.? Messaging on data-literacy and vision like this is now percolated to the top of the world’s largest pharmaceutical companies. ? (Check out Aviv here and here)

It is not about bridging, it is about integration. Data science and data literacy everywhere, most importantly in our minds already when we start thinking about biological questions and designing experiments.??

Adding to the movement of this data-and-biology integration, there is another recent positive development in the local Danish environment: Novo Nordisk Foundation, LEO foundation and Lundbeck Foundation have joined forces to finance a cross university Center for Pharmaceutical Data Science Education, which will provide research based state-of-the-art education both future and current drug experts. This addresses the roots of the challenge.

The Future: Full Integration?

The hype and promise of data and AI in pharma have been around for a while - but only now there is movement towards integration and embedding rather than separation.? We are not there yet, overall data literacy is not there in pharma - and the biggies like Roche and Novo feel there is a lot of value in various forms of specialised data science centers and groups - they need bridging. However, on the long playing field it is about full integration. And that needs education, culture and habits.

In my own small world, we started the Roche #RNAHub with a key strategic goal of high data literacy: eventually, >50% should have a programmatic general-purpose data analysis environment as their main tool.? A big ambition and experiment that is only possible by our peculiar cross-functional unit inside big pharma that is otherwise dominated by somewhat byzantine point-and-click systems, spreadsheets (and specialised detached analysis teams).? In a later post, I will tell you about how that goes - stay tuned!