The Connectivity Map Inference Challenge is launched today.
Problem Description
You are given a large training dataset (100,000 samples) of 12,320 gene expression levels that include 970 landmark and 11,350 non-landmark genes. You are also provided with a smaller offline-testing dataset of 1,000 samples where the landmark and non-landmark genes are in separate files (for compatibility with the scoring scripts). The challenge is to build a model that uses the 970 landmark genes to predict the expression levels of the inferred genes for a distinct set of 1,650 separately measured samples.
Ref: http://www.lincscloud.org/contest/
https://community.topcoder.com/longcontest/?module=ViewProblemStatement&rd=16753&compid=52659