Crowdsourcing Cancer Challenge

An international consortium — including University of Toronto professors Quaid Morris and Paul Boutros – has launched a global crowdsourcing competition to find new tools for researching cancer.

Research groups from Canada, the U.S. and the U.K. have joined forces to create a public contest in software development in order to address one of the greatest challenges in cancer biology. The contestants will analyze vast amounts of DNA sequence data to identify genetically distinct groups of cells within tumours that are often the reason why therapy fails.

Called the ICGC-TCGA DREAM Somatic Mutation Calling Heterogeneity (SMC-Het) Challenge, the competition is now open and runs until May 2016.

Cancer remains a leading cause of death throughout the world because of its ability to evade even the best available treatments. Recent advances in DNA sequencing reveal that tumour cells don’t all share the same DNA – rather some have acquired unique mutations that cause them to respond differently to therapy and spread around the body at different rates. To find an effective treatment, we first need an understanding of the many different types of cancer cells present in each patient.

“We know that cancers are made up of many different populations of cells, known as ‘subclones’, and understanding the relationships between these subclones is critical in developing successful long-term treatments,” said Dr. David Wedge, Staff Scientist at the Wellcome Trust Sanger Institute in Cambridge, U.K. and one of the co-organizers.

The new ease of collecting patients’ genomic information also poses a challenge of how best to analyze the growing mountain of data.

“There’s been an explosion of computational methods for teasing cancer subclones apart, each using different strategies and making different assumptions. We wanted to organize this competition as a way of providing independent assessment of which approaches work the best,” said Morris, who is also a professor in U of T’s Donnelly Centre and the Department of Molecular Genetics.

The competition merges the efforts of the world’s largest cancer genome sequencing consortia with DREAM Challenges and Sage Bionetworks - the world-leading organizations in community-based open science that championed crowdsourcing as way of finding answers to some of the biggest questions in biomedicine.

“Crowdsourcing has provided a way for researchers across the world to collaborate on solving problems together. As a whole, the SMC Challenges have provided focus on specific problems that are understudied, and provided an easy way for scientists from other fields, and citizen scientists, to get involved in solving key problems in cancer genomics,” said Boutros, a scientist at the Ontario Institute for Cancer Research and a professor in the Department of Pharmacology and Toxicology and Department of Medical Biophysics.

The organizers have created a set of 50 realistic tumours with distinctive life histories and evolutions. Contestants will create tools for detecting cancer subclones in the cloud, donated by Google and using the Google Compute Engine that will be run in Galaxy, a widely used open-source platform in biomedical research. This will allow researchers to access the data after the contest, creating a new library of algorithms that can be used in future studies and compared in an objective way.

“The only way cancer researchers can make informed decisions about the tools that they use is if the quality of algorithms is assessed objectively and independently,” said Amit Deshwar, a graduate student in Morris' group.

All participants who present a final algorithm will be co-authors on a Challenge overview paper, which will be submitted to Nature Biotechnology, a leading journal in the field and the official partner of the Challenge. Top performers will also receive travel awards and speaking invitations at the 2016 DREAM Conference, the 2016 Sage Congress or a similar event. Finally, the overall winning algorithms will be applied to real cancer data, including a subset of up to 2,500 whole genome tumour sequences.

“That’s a great rewards to think of: You can run your method on real patient data and ultimately make a contribution which could transform our understanding of cancer and how we treat it,” says Morris.