Editors' ChoiceComputational Biology

A Model Challenge

See allHide authors and affiliations

Science Translational Medicine  12 Sep 2012:
Vol. 4, Issue 151, pp. 151ec162
DOI: 10.1126/scitranslmed.3004863

What’s first on the list in Robert Fulghum’s book, All I Really Need to Know I Learned in Kindergarten? “Share everything.” Second? “Play fair.” Designers of the open-science Sage/DREAM Breast Cancer Prognosis Challenge learned these lessons well, and there is still time for other computational modelers to join in the show-and-tell. This open computational challenge to identify predictors of breast cancer progression is accepting submissions of models until 15 October 2012.

Breast cancer is the second leading cause of cancer death among women in the United States. Despite the fact that billions of dollars are spent each year on research and treatment, biomedical scientists have an incomplete understanding of prognosis and survival rates, which vary greatly among patients. The goal of the Challenge is to use crowdsourcing to mold a computational model that accurately predicts breast cancer survival. Challenge participants are invited to use genomic and clinical information from 2000 women diagnosed with breast cancer (the Metabric data set) accessed via Sage Bionetworks’ Synapse platform for data sharing and analysis; run computations on their own standardized virtual machine donated by Google; and submit developed models back to Synapse, where they are immediately scored and posted to a real-time leaderboard. Standardized and shared computational infrastructure enables participants to use code submitted by others in their own model building.

Challenge models are scored throughout the Challenge by assessing the concordance index between the predicted survival and the true survival information. The best performers from the first phase of the Challenge will be announced at the DREAM 7 Conference in San Francisco this November. To determine the overall Challenge winner, the Avon Foundation has funded the creation of a never-before-released validation data set to be used for the final assessment. The Challenge winner will be invited to publish an article about the winning model here in Science Translational Medicine and to present at Sage’s 4th Annual Congress in April 2013.

Sage’s Breast Cancer Challenge hopes to provide incentives for researchers to share data and work collaboratively to build disease models. To encourage collaboration in the Challenge community, Sage Bionetworks announced an award of $500 to the first participant who used code submitted by other participants to improve on the current leading model—as long as he or she properly credits his or her collaborator. And in less than a day after the announcement, the first set of $500 prizes was awarded for code sharing—one to the participant whose code was used and one to the participant who used that in combination with his or her own—strengthening the notion that open collaboration has the potential to accelerate disease model development.  

[Sage/DREAM Breast Cancer Prognosis Challenge] [Real-time leader board] [Synapse]

Stay Connected to Science Translational Medicine

Navigate This Article