show image

Sooraj Shah

Contributing Editor

Sooraj Shah is Contributing Editor of New Statesman Tech with a focus on C-level IT leader interviews. He is also a freelance technology journalist.

The Crick’s CIO Alison Davis on using machine learning to speed up scientific research

The Francis Crick Institute is an independent charity that focuses on understanding the fundamental biology underlying health and disease. Its aim is to better understand why disease develops and translate discoveries into new ways to prevent, diagnose and treat illnesses such as cancer, heart disease, stroke, infections and neurodegenerative diseases.

Many may not be familiar with the organisation, perhaps because the institute was renamed in 2015 in a merger. Its new founding partners are the Medical Research Council (MRC), Cancer Research UK, Wellcome, UCL, Imperial College London and King’s College London. However, the Crick as it is known in short, was formed from two former institutes, one of which, the National Institute for Medical Research, dates back to 1913.

The advancements in science and technology have gone hand-in-hand in that time, and so the Crick has required a state-of-the-art IT infrastructure to enable world-leading scientific research to ensure it becomes the world-class organisation it strives to be.

Alison Davis is the woman tasked with managing this ambition. As CIO, Davis has already worked to ensure that the company’s IT was in order post-merger, and again when her team moved to the Crick laboratory opposite St Pancras station at the end of 2016, bringing together the scientists and the IT team.

There are updates to the existing technology too; the Crick started with 2.3 Petabytes of storage, and it has just updated that to 10 Petabytes of ‘high availability’ storage.

Now, her work is focusing on getting the most out of the technology that has already been implemented.

“We’ve done a lot of work on migration, and now it’s really about trying to optimise the services and work closely with the scientists,” she says.

Davis’ role is slightly different to many CIOs in that she has to ensure that her team provides the backbone and underpinning technology that enables scientific researchers to innovate and produce high quality research.

“We’ve got bioinformaticians in the labs, who are doing things that push the envelope and we’re trying to respond behind that to ensure we’ve got the platform for them to do it on,” she states.

She says that as an organisation, the Crick has two modes of IT; one which is focused on the systems and records such as finance and HR – the classic IT environment, and then another mode in which her team needs to be more agile, flexible and provide platforms that allow sandboxing to test things like machine learning.

Scientific use of machine learning

While Davis’ IT team doesn’t use some of the emerging technologies – the scientists do. For example, they’re looking into the potential of machine learning to analyse images that would otherwise take up a huge amount of time for people to process.

“The goal is ultimately to speed up research time, because if machine learning can help you look at images more effectively, that will reduce the time it would take a highly qualified scientist to do it and leave them much freer to look at the overall problem itself,” she says.

Davis, who was a chemist earlier in her career, suggests that one of the key similarities between science and IT is about asking the right questions.

“Often a new innovation or insight makes people think ‘that’s so obvious why didn’t we think of that before?’ – and a lot of that is about having asked the right questions and then making it easy for them to be answered – it’s about asking them in the right way,” she states.

“And if you’re not bogged down by having to do a lot of the mechanics of the experimenting to ask the question or prove the hypothesis that you came up with, then that should make science quicker,” she adds.

A big part of making machine learning effective is ensuring that the right training data sets are available and that there is no bias involved, Davis explains.

“You can’t just point to machine learning apps and say ‘go and do it’, because you need to train the application and you have to make sure the way you’re training is balanced,” states Davis.

The adoption of machine learning is not linked to commercial outcomes for the Crick, unlike some other organisations – but it is still important to get right.

“We’re a charity that is trying to extend the sum of human knowledge about human health and disease, and so most of the outcomes of our work are hypotheses that will be tested. In this sense the risk of adoption of new technologies, such as machine learning, is not the same as it is for commercial organisations; however, if experiments get faulty results and conclusions this creates a risk in terms of slowing down the overall progress of the scientific endeavour and understanding,” she states.

But this will not deter the scientists or the IT team from ensuring that the Crick can use the most up-to-date technologies available. After all, new technology can play a huge part in helping to better understand human health and disease, and this is the organisation’s ultimate aim.