AstraZeneca chief scientist tells of the potential of CRISPR and the Functional Genomics Centre with CRUK
AstraZeneca’s chief scientist has told the Cambridge Independent that the company’s new collaboration with Cancer Research UK (CRUK) will be of “massive benefit” to both parties.
Dr Ultan McDermott was speaking after the announcement of a new collaboration between the two to create a Functional Genomics Centre within the Milner Institute on Cambridge Biomedical Campus.
As reported earlier this month, the multi-million pound joint venture will be a centre of excellence in genetic screening, cancer modelling and big data processing, using CRISPR gene editing technology to create new biological models more reflective of human disease that will help advance cancer drugs.
Speaking to the Cambridge Independent at the CRUK Cambridge Institute, opposite the Capella building that will house the Functional Genomics Centre, Dr McDermott said: “CRISPR is a really good tool in a research setting to turn genes on or off. In a cancer cell, you might want to turn a gene off to try to kill a cancer cell. We can do that across all 20,000 genes.
“So for example in lung cancer cells, you could turn every single gene off and find one which kills the cells that we hadn’t even thought of as a target.
“Running these screens is challenging and there are undoubtedly a lot of investigators at CRUK who would love to run these screens, but it’s difficult to do.
“The ability for those 4,000 scientists and doctors across CRUK to propose a screen, have it run here at very high quality and then to get that data back for them to work on for the next couple of years to identify new ways to kill cancer cells or overcome drug resistance, which is a big problem, is going to be a massive benefit.”
AstraZeneca’s £500million-plus new global headquarters and R&D centre, due to open in 2020, lies just minutes from the CRUK Cambridge Institute and the Capella building where its scientists will work side by side with those from Cancer Research UK.
“The collaboration is about bringing expertise. From AstraZeneca’s point of view, we want to work and collaborate with them. Undoubtedly part the reason of AstraZeneca’s move from Manchester to Cambridge is about the proximity. The opportunity is huge,” said Dr McDermott, who worked at the Wellcome Sanger Institute in Hinxton for nine years before joining AstraZeneca in 2017 - a move that he said reflected the biopharmaceutical company’s commitment to CRISPR technology.
While the gene-editing technology has long-term potential as a therapeutic tool, Dr McDermott believes it will be of more use in the research setting to develop new drugs.
“I think it’s a while off using CRISPR as a drug in patients,” he said. “I think the reality is it’s a really good tool to identify new targets. The reason the success rate [of new drug candidates] is low is partly because we invent drugs that don’t work as well as we think they should.
“These genetic screens are a really accurate way of identifying novel vulnerabilities - the Achilles heels of many cancers.
“The screens take three or four months. It’s still many times faster than many other types of screens and importantly the data is very clean.
“In the past, you would run chemical screens, where instead of knocking out genes you would put chemicals on and see what the effect was. The problem was the chemicals would have lots of off-target effects.
“With CRISPR, while you don’t have zero, you have many times fewer off-target effects.”
These off-target effects are mutations caused elsewhere in the genome when carrying out CRISPR gene-editing on a target site.
It remains a significant barrier to the use of CRISPR techniques as a therapy, as it could disrupt the function of otherwise normal genes.
But last month researchers at the Wellcome Sanger Institute announced that they had developed a machine learning tool, called FORECasT, that predicted the exact outcomes of CRISPR-Cas9 gene editing using a huge quantity of sequence data.
That tool is now openly available for use by all researchers using gene editing to study health and disease.
Dr McDermott explained that the issue of off-target effects when using CRISPR can be overcome in a research setting by the use of multiple guide RNAs - a protein that acts like a homing device to a specific section of DNA in the genome.
“You use a guide-RNA as a ‘bullet’ to target a gene,” he said. “Typically each gene will have six guide-RNAs so what you look for is for each to have the same effect and then you know it’s not an off-target effect.
“If you break up the process into running the screen, sequencing it and analysing it, the three are all equally important. The computational side is critical.
“We will be building a single pipeline, which will deliver a data package that is useful to biologists. Our computational approach is you take a huge amount of data and put them in different pathways that might be druggable.
“It will be work in progress. Working with CRUK, we’ll ask what is the best way to analyse the data so that when we deliver it back to the scientists on the bench they can most rapidly translate that into something that is eventually going to be a drug.
“What you get out at the end is 20,000 genes ranked by how well they kill a cancer cell and what you want to do is find out whether they point to one or two important pathways so we can drug those pathways.”
Machine learning will play two critical roles, he suggested. Firstly, it will help design the most useful libraries.
“The rulebook for how you design a guide to go to this bit ofDNA and nowhere else is still not complete,” he said.
“But there have been hundreds of thousands of guides now designed and we know whether they work, so machine learning and artificial intelligence will be a really useful tool for designing the best libraries possible. In this centre we will very quickly have a huge amount of data.”
Next, AI can be applied to the harmonised dataset that the Functional Genomics Centre will develop.
“Once you run the screens and you get all that information about which genes do what, over time we will have one of the largest CRISPR data sets on the planet to apply artificial intelligence to,” said Dr McDermott.
“That takes time to build up because AI needs truly huge amounts of data. But the advantage we have is that if you were running these screens in 12 different laboratories, using 12 different protocols, the data is all different. It’s very hard to combine it all together.
“What AI needs is a single, high quality combined dataset to work. Because this is all being run in one centre, through one computational pipeline, whether it’s an AZ or CRUK project, it will all be harmonised data.
“So the ability to combine that data has huge potential.
“And because we are designing this centre from scratch, we can make sure the data is AI compatible, which is the other big challenge.
“Some people have big datasets but because they were gathered pre-AI, they aren’t in a format that makes AI easy. It’s a big headache for datasets worldwide.
“We can make sure everything we build is compatible with AI.”
AstraZeneca and Cancer Research UK are investing 50-50 in the new centre, which will begin preliminary research in January, with laboratory work expected to begin in the second half of 2019.
CRUK will be able to use its data across its network.
“Each data package goes back to the investigator,” said Dr McDermott.
“The actual libraries and tools developed within this centre are free to distribute among the entire CRUK research community, which is a huge plus, and speaks to this as a collaborative endeavour.
“There is lots of potential, but the hard work starts now…”
More by this authorPaul Brackley