IDG-DREAM Drug Kinase Binding Prediction Challenge is open
FIMM researchers are organizing the IDG-DREAM Drug Kinase Binding Prediction Challenge that benchmarks machine learning models in the effort to accelerate mapping of drug-target space by prioritizing most potent compound-target interactions for further experimental evaluation.
DREAM is launching the IDG-DREAM Drug Kinase Binding Prediction Challenge that will run through the fall of 2018. This Challenge seeks to evaluate the power of statistical and machine learning predictive models as systematic and cost-effective means for guiding efforts to map drug-target space by prioritizing most potent compound-target interactions for further experimental evaluation. The Challenge will focus on kinase inhibitors, due to their clinical importance, toward extending the druggability of the human kinome space.
The Challenge will be implemented in a screening-based, pre-competitive drug discovery project in collaboration with the NIH-funded Illuminating the Druggable Genome (IDG) Kinase-DRGC consortium, using their unpublished target selectivity dataset to evaluate the model predictions, with the aim to establish kinome-wide target profiles of small-molecule agents. Teams with new approaches to compound-target interaction prediction are especially encouraged to participate. All the models, new bioactivity data, and benchmarking results will be made publicly available.
The experimental-computational approach is based on the work that was published in 2017, where the FIMM & HIIT PhD student Anna Cichonska was the first author of the study.
“Our pilot results demonstrated the potential of machine learning methods for filling the gaps in existing drug-target interaction maps. This challenge will continue the development and benchmarking of the models on much larger scale”, says HIIT/FIMM-EMBL PhD student Anna Cichonska.
The collection and standardization of the target profiling data for model training is based on the crowd-sourcing DrugTargetCommons platform implemented and housed at FIMM.
“This Challenge will also provide a great test-bench for the operation and wide usability of the DrugTargetCommons platform to provide the training data for participating teams from all around the world”, says FIMM-EMBL PhD student Balaguru Ravikumar who has been responsible for preparing, compelling and testing the training data retrieval pipeline in DrugTargetCommons.
Registration for the IDG-DREAM Drug Kinase Binding Prediction Challenge is now open. The round 1 leaderboard closes in November 2018.To learn more, interested teams are encouraged to register for the Challenge webinar on October 10th.
“Even though FIMM-affiliated teams cannot participate in the Challenge, as FIMM is one of the organizers, we hope FIMM researchers will widely advertise and promote the Challenge among their colleagues and using social media”, says FIMM-EMBL Group Leader Tero Aittokallio.
This challenge is jointly organized by:
- Sage Bionetworks
- DREAM Challenges
- Illuminating the Druggable Genome
- Institute for Molecular Medicine Finland, HiLIFE, University of Helsinki
- UNC Eshelman School of Pharmacy
- Icahn School of Medicine, Mount Sinai
- UCPH Biotech Research & Innovation Centre BRIC
Anna Cichonska, FIMM & HIIT, Aalto University, firstname.lastname@example.org
Balaguru Ravikumar, FIMM-EMBL PhD Student, email@example.com
Tero Aittokallio, PhD, Prof., FIMM-EMBL Group Leader, firstname.lastname@example.org