If you’re familiar with the Crick, you’ll probably know that we have a lot of cancer researchers here – nearly half of our 1,200 scientists conduct research relevant to cancer. What you might not know is that a large proportion of them don’t wear white coats, and may never lay eyes on a tumour. In fact, many of them rarely set foot in a lab at all.
But these people – our computer scientists – are key players in the fight against cancer, and are making discoveries that will benefit patients in the near future.
“Nowadays, cancer research would simply not be possible without data science,” says Francesca Ciccarelli, a computer scientist who heads up the Cancer Systems Biology Lab.
The explosion of DNA sequencing tools in the last two decades has enabled scientists like Francesca to collect vast amounts of data about the genomes of both normal and cancerous cells to try and understand what causes cancer and predict how the disease will progress.
Given that there are 3 billion letters in the human genome, there’s an awful lot of data to crunch. Hence the need for high-processing-power computers and computational experts who know how to write code that can detect patterns in large datasets.
Francesca’s interdisciplinary team analyses data from patient tumour samples to investigate how multiple gene mutations acting together drive cells out of control. Over the years, they’ve built up a catalogue of known and possible cancer genes that researchers worldwide can use to help them understand more about the disease.
“The idea is that if we can figure out cancer’s ‘rulebook’, we can start to predict its next steps, and get one step ahead of the disease,” she says.
Although still in its relative infancy compared to many biological approaches, computational oncology has already led to improvements in patient healthcare.
“Data science has given us major insights into the genes and processes involved in cancer,” explains group leader Peter Van Loo. “This has led to more accurate diagnostic and prognostic tests, and many new targeted treatment approaches for specific cancers.”
For instance, researchers in Peter’s lab recently discovered a genetic mutation that enables clinicians to correctly distinguish between benign and cancerous bone tumours and decide on the best course of treatment for the patient.
“Every tumour is unique,” adds Paul Bates, whose team uses computer simulations to study how different cancer cell populations emerge, develop and are maintained. “So we need to take this into account when deciding how best to treat someone.”
The emphasis is to move away from a ‘one-size fits all’ approach to cancer treatment, and instead make it much more tailored to the individual.
Like many of our computational groups at the Crick, Paul collaborates extensively with ‘wet-lab’ scientists who provide him with biological data that feeds into his computer models, and can also test the predictions that the models spit out.
“We’re very intertwined with experimentalists here; I like to think we’re greater together than the sum of our parts,” he says.
One group that Paul interacts with regularly specialises in how cancer cells spread to other parts of the body – a major obstacle in treating the disease that is further complicated by the fact that different cells within a single tumour may differ in sensitivity to different drugs.
Computer simulation of a cancer cell migrating toward a blood vessel.
- Paul Bates' labAccording to Paul, experimentalists alone may never hit the “sweet spot” of knowing what ratios of drugs and in what order to administer them to have the maximum impact on slowing or stopping cancer from spreading.
“Using computer models we can apply ‘virtual drugs’ to different cancers and run hundreds of simulations to make predictions about the best way to slow metastasis,” says Paul. “We give our predictions back to the biologists who can then see if the predictions are right.”
Given the speed with which computer science has become an integral part of cancer research, and had an impact on patient’s lives, it seems fair to assume that the years ahead will be equally, if not more, fruitful.
“In the next ten years, I expect that we will be able to develop detailed maps of how normal and diseased tissues develop and evolve, which over the following decade will lead to new diagnostics and treatments for many cancers and other diseases,” says Peter.
For Francesca, the excitement is around studying the interactions between cancer and healthy cells in the so-called ‘tumour microenvironment’. Her team plans to identify populations of healthy cells and see how their genes and proteins change in response to coming into contact with cancerous cells. Looking at not just genes but protein products will throw up even more data. “There’s lots of room for developing innovative ways to process all these data, and find hidden connections – that’s the exciting part!”
Paul is hopeful that computational oncology will continue to allow us to understand tumour diversity to a deeper level than experimental observations alone. “Right now, we can build up a picture of tumour heterogeneity by taking multiple biopsies from patients, but this is rather invasive,” he says. “What we’re aiming towards is building computer simulations from the data gathered in a single biopsy and using it to figure out the best treatment regime for that person.”
They might be tackling different research questions, but Peter, Francesca and Paul are all in agreement that collaborations between lab-coat-wearing and keyboard-wielding scientists are key to progress in understanding and tackling cancer.
