The challenge is to construct a predictive model that can transform the L1000 transcriptomic data, given as input, to the corresponding P100 phosphoproteomics data as the output. Submitted models will be evaluated based on their ability to successfully predict future P100 data from L1000 data profiled under the same exact conditions. Below are two data files provided as the training set. They contain matching experiments where L1000 and P100 experiments were applied under very similar conditions.
The data files contain 91 unique conditions. The columns have a one-to-one correspondence in both files. Timings are 3, 6, and 24 hours with a preference for the 3 hours time point and for the P100 and 6 hours for the L1000. Concentrations are binned to the closest match: 0.1, 1.0, 10.0, 100.0 (muM) according the minimal fold difference.