DREAM CHALLENGES



The DREAM (Dialogue for Reverse Engineering Assessment and Methods) project was founded in 2006 to assess model predictions and pathway inference algorithms in systems biology. Today DREAM aims to foster collaboration amongst researchers, to evaluate computational biology methodologies, and to multiply the impact of data by making it available to a vast community. DREAM achieves its goals through the organization of Scientific Challenges and conferences.

The DREAM Challenges are open problems presented to the systems biology community of potential solvers. Participants submit their predictions, which are evaluated and scored, and eventually discussed in the annual DREAM conference. Eventually data, predictions, gold standards and source code are openly available. In the past 7 years, DREAM has run 28 successful Challenges, enabled the publication of over 60 DREAM Challenge-related papers, and aggregated a "crowd" of thousands of "DREAMERs."

In 2013, Sage Bionetworks (www.sagebase.org) and DREAM partnered to co-lead a new generation of Challenges that leverage collaborative data hosting and analysis tools available on Synapse (www.synapse.org) such as real-time leaderboards and shared project spaces.

Announcing the DREAM9 Challenges

On June 2, 2014, we launched the DREAM9 "Challenge season" which consists of three new Challenges described below. These challenges will be discussed at the RECOMB/ISCB Systems and Regulatory Genomics/DREAM Conference to be held in San Diego, CA November 10-14, 2014.

1. Alzheimer's Disease Big Data DREAM Challenge #1 The goal of this Challenge is to identify accurate predictive Alzheimer's disease biomarkers that can be used by the scientific, industry and regulatory communities to improve Alzheimer's diagnosis and treatment. Participants will work with genetics data, clinical data and imaging data to create predictive models of cognitive scores, predict discordance between cognitive ability and amyloid load and/or predict diagnostic groups.

2. The DREAM AML Outcome Prediction Challenge The goal of this Challenge is to develop the best predictive models of clinical outcome in AML. Participants are given the clinical correlates, genetics and cyto-genetics data and the expression level of 231 proteins probed by RPPA analysis of a cohort of AML patients. Challenge participants are asked to predict which AML patients will be primarily resistant to therapy and which patients will have complete remission as well as to predict remission duration and overall survival.

3. The Broad-DREAM Gene Essentiality Prediction Challenge The goal of this Challenge is to infer genes that are essential to cancer cell viability using gene expression and/or gene copy number features. This Challenge, which will leverage the data from the NCI funded project Achilles, is a natural progression of the NCI-DREAM7 Drug Sensitivity predictions challenge.

For more details about how to register to participate in these Challenges, a description of these and other ongoing Challenges, the incentives for participation (including publication opportunities and podium presentation at conference) and the collaborators and sponsors that enabled the Challenges please go to www.synapse.org/dream. Every participant can contribute to the solution of these important Translational Systems Biology challenges.

Contacts: Thea Norman (thea.norman@sagebase.org) and Gustavo Stolovitzky (gustavo@us.ibm.com)