Removing the computational barriers to healthcare advancement

Last week I joined a distinguished panel at the 10^th Edition of Pharma Integrates, the virtual pharmaceutical industry event hosted here in the UK Innovation Corridor. I was fortunate to share the platform with Kevin Cox from Biorelate, Nick Pappas, UKRI-STFC, Andrew Pierce, Crescendo Biologics and someone well known to Kao Data, Craig Rhodes, who champions AI for healthcare and life sciences at NVIDIA.

Our discussion focussed on the need for a faster, and more fluid research environment, underpinned by high performance computing (HPC) and artificial intelligence (AI), and more specifically, how AI-enabled research is driving the adoption of advanced computing and delivering a host of benefits around the globe.

To put it simply, I was in a room full of rock stars, with Kao Data representing the data centre industry that provides the stage and backdrop for their performances. The combination of our panellist’s expertise and imagination has already changed the world for the better, yet they all still require a dedicated IT-environment in which to model, train and hone their craft.

A shining example of rock star computing is NVIDIA’s Cambridge-1; the UK’s fastest and most powerful supercomputer. Located at our Harlow campus between London and Cambridge, this system was founded by a host of private and public sector leaders within the realms of pharmaceutical and biomedical research and is designed as a high performance platform to test and develop lifesaving science.

In just a short space of time, Cambridge-1 has become a centre of excellence in the field of healthcare. Its capabilities were brought together in just a matter of weeks, not the years it usually takes to design and build supercomputers, and all amid a global pandemic. This is one area in which our collective response has shone a light on both the power, the potential, and the reach of bio-computational healthcare, and it’s paramount that we learn from it - translating the findings to help prevent future outbreaks. One might argue that only via our collective experience have we now begun to understand what it means to be responsible for ‘mission-critical' compute.

Cambridge-1 has marked a step-change for the industry, and it demonstrates how accessible supercomputers have become. Where once supercomputing was the realm of the public sector, of universities and research bodies, today NVIDIA has given AI Start-up organisations ease of access to these resources via their industry-leading DGX range. Peptone, who are building the world’s first Protein Engineering Operating System, P^eOS™, is the first start-up to use Cambridge-1’s supercomputing capabilities and combine generative AI with state-of-the-art computational molecular physics. Indeed, the system has in many respects become a community resource for pioneers in healthcare.

Throughout the discussion, what was obvious was the enormity of the prospect. Today’s pharma-industry has become increasingly data centric; from its generation and capture, to its use and ownership. 90% percent of data in the world today was only generated in the last two years, and in the pharmaceutical and biomedical sectors alone, the amount of data involved is vast and ever-growing. The European Bioinformatics Institute (EMBL-EBI), for example, who’s HPC and storage infrastructure is located at our Harlow campus, receives over 80 million access requests for data daily, from users across the globe.

On our campus it’s easy to walk across the car park and assume all is quiet. If we could only see those 80 million daily access requests as lines of connectivity into our facility it would be a staggering image above our heads. One can only begin to imagine how that will change in future.

Programmes such as AlphaFold and RoseTTAFold are also helping to transform biology by determining a protein’s 3D shape, while elsewhere increasing volumes of HPC are being deployed for large-scale bioinformatics, pipelining and pathology analysis, and generating new data.

The consensus appears to be that, looking forward, the industry needs to create frameworks and structures to allow such data to be efficiently implemented into HPC analysis, and with it, to facilitate the downstream process. Gaining access to that category of data is another step-change, which opens valuable gateways and increases the number of successful outcomes.

One must also remember that the skills and expertise of people is crucial. As data and AI applications explode in number and diversity, so too does the need for trained data scientists, and it is at the interface between data and biology where the requirement for experienced data scientists intensifies. The skills shortage is a challenge affecting many industries, and in AI, it cannot be solved quickly.

At Kao Data we believe that we are in a very exciting position: working with global technology vendors, private and public organisations, and start-up businesses in some of the fastest growing and meaningful industries. What we do makes an impact on what our customers achieve, so we need our teams to have the skills, the high standards of technical excellence and the capabilities to exceed expectations throughout the lifecycle.

As a high performance data centre operator, it is essential that we continue to collaborate with our customers and gain a deeper understanding of how they use data, so that we can remove the computational barriers to healthcare advancement. By doing so we can and will create both processes and systems to support their objectives.

Further, while this ground-breaking work continues, our role at Kao Data is to build a deeper understanding of how HPC interacts with end-users’ data, and to develop a dedicated, high performance environment that evolves with the rock stars requirements. AI has indeed taken the world stage, creating a new era for healthcare.

Watch the Pharma Integrates panel session, ‘Underpinning the Future of Healthcare with High Performance Computing and Artificial Intelligence’, here.