Interview: James Fleming, CIO, Francis Crick Institute

Supporting transformational change

The Crick began operating in its new, purpose-built facility near London’s King’s Cross Station in early 2017. Today, the centre houses more than 100 research groups. Fleming, who reports to the institute’s director of research infrastructure, says much of the work during his six years at the organisation has been focused on helping to establish its infrastructure.

“That first phase was frenetic,” he says. “It wasn’t even an enterprise transformation because there wasn’t much to transform. It was about building everything from scratch. We were in startup mode for the first couple of years.”

I go to sleep at night thinking that I’m helping cure cancer with computers

Further challenges came in 2020 when the infrastructure startup process rapidly segued into the coronavirus pandemic. Fleming and his colleagues repurposed the entire institute as a Covid-19 test pipeline for acute trusts in London. The organisation tested employees and patients during that time. He describes the pace of the transformation effort as frantic.

“We had the pipeline up and running in nine days,” he says. “We then built a staff testing solution on top of that pipeline in a further nine days that we also extended out to the Wellcome Trust and the Sanger Genome Campus in Kingston, which was vital because they were coordinating all of the genomic variant analysis for the country. Our work kept those operations open so they could keep tracking and tracing the different variants.”

Fleming reflects on this tumultuous period of change by suggesting there’s hardly been time to draw breath. The Crick has continued to recruit researchers and scale up its scientific activities. The organisation’s size has almost doubled since 2017. Fleming says the increase in scale has been accompanied by “a complete revolution” in how science is performed.

“The seeds of change were there six years ago when I joined, and the Crick was quite forward-thinking in its approach to data science. The institute invested heavily in first-generation, high-performance computing. We’ve just gone to our second generation now, after a major refresh investment,” he says.

“It’s fair to say that data science methods have moved from being a small, niche activity in bioinformatics to now being part of the standard toolkit of every lab in the building. So, there’s been this revolution in that period between wet lab biology and computational biology that we’ve had to keep pace with and help enable as well.”

Creating a flexible platform

Fleming joined the Crick as director of IT. He says one of the major changes during his time with the institute was when scientific computing was brought together with the rest of the IT department. This integration process led to the creation of the CIO role in 2021.

In his role overseeing technology, Fleming works with multiple hardware and software providers. Nvidia supplies graphics processing units (GPUs), Dell supplies central processing units (CPUs), and Lenovo and IBM supply storage. Crick also runs an open source software stack on top of its hardware.

Snowflake is another key provider. In 2022, Fleming explained to Computer Weekly how the supplier’s Data Cloud provides precise access controls, management reporting, authentication, billing control and data sharing capabilities. The institute has used Snowflake to build trusted research environments (TREs), which support groundbreaking research across various areas. Fleming says that effort continues to be scaled up.

“The Crick was quite forward-thinking in its approach to data science. The institute invested heavily in first-generation, high-performance computing. We’ve just gone to our second generation now, after a major refresh investment”

James Fleming, Francis Crick Institute

“We use Snowflake as an adjunct to all the high-performance compute research we undertake,” he says. The platform ensures clinically identifiable data stays secure. The TREs support multiple projects, from the institute’s original pilot project for Snowflake, which looked at the effect of long Covid on rare cancers, to investigations into the epidemiology of coronavirus. Fleming says Snowflake assists cross-departmental research processes.

“That is an elegant solution to the problem of ‘how do I work with my dataset?’, which is combined slices of multiple people’s datasets,” he says. “But everyone is assured. You’re not squirming the data away into servers, and it’s never seen again.”

Today, the technology supports a dozen concurrent projects with different TREs. Fleming describes Snowflake as a “business as usual” platform for clinical research. He says the next stage will consider how the technology can support new areas.

“The TREs are effectively just a configuration of Snowflake,” he says. “The flexibility that approach affords you when you get to grips with the architecture of the platform’s core components means you can do lots and lots of very, very cool things.”

Exploring new fields

Fleming recognises the challenges he’s faced during the past couple of years have intensified with the rapid pace of technological development. The IT industry and the general public have become excited by the potential of generative artificial intelligence (GenAI).

The institute has long used emerging technologies, such as artificial intelligence and machine learning, to help researchers unlock discoveries. However, Fleming says the increased opportunity to use GenAI hasn’t led to a big shift in working at the Crick.

Provability is an essential element of the scientific research process. He says researchers can’t simply use publicly available large language models to answer life’s big questions: “You’ve got to be rigorous with the provenance of what you’re working with, the data you’re using, how you’ve improved the model’s performance over time, what that analysis has done for your mechanistic understanding of the disease, and so on.”

The key to successful scientific research is using tried-and-trusted techniques to deliver rigorously tested and provable results. Rather than jumping on the GenAI bandwagon, Fleming says the next big challenge will be using technology to help researchers ensure humans are at the heart of scientific analyses. He gives an example.

“One of our group leaders did landmark work published last autumn that created a new mechanism for lung cancer development related to air pollutants,” he says. “The obvious question they want to explore is, ‘OK, now we’ve made that connection at a local lab level, what does it mean at a population-wide level?’”

Fleming says other researchers are investigating the effect of microplastics in the environment on neurodegeneration. Exploring those research questions requires new data types in fresh combinations.

“You need to work geospatially,” he says. “You need to work with environmental data sources. So, how do we bring that data and connect it to a clinical lab?”

The answer, suggests Fleming, will be for his team to work out how the institute can use non-traditional data sources that would normally be considered part of environmental sciences rather than the life sciences world.

“How do you join those worlds with platforms, perhaps using technologies like the internet of things and sensors and geospatial data-mapping tools, to everything from the Earth as an ecosystem to imaging an individual protein? I don’t know how we’ll solve that question yet, but that’s probably the next unanswered question we need to address.”

Delivering pioneering solutions

Fleming says the digital organisation he’d like to create in 24 months would stay true to the long-term principles that have governed his work as CIO at the Crick. “When we put together the strategy for the department, the mission statement said we wanted to enable world-class science – and that’s still the case,” he says.

Increasingly, Fleming says scientists from across the organisation come to his department with problems that need to be solved. His team can help these experts if they’ve built strong hardware and software foundations. Fleming sees the foundations as a conduit that takes scientists from hypotheses to solutions.

“They’re only going to get to answering the questions they want to answer once we build the bridge from one place to the other. Across the organisation as a whole, I hope we’ve built that structure already – and I hope we continue to hone our approach,” he says.

“We need to provide an IT organisation that is dynamic and very problem-solving orientated, with deep technology expertise to reconstitute scalable, high-quality, quite niche solutions in the first instance, but then potentially generally applicable, platforms very, very quickly.”