Challenge phases

This challenge is comprised of a single phase:

Identifying functional enhancers for each annotated cell type

Here the goal is to define novel machine learning algorithms to predict functional enhancers specific to individual cell types, using multi-modal genomic and cross-species profiling of the motor cortex. The data sets provided for this task are a diverse collection of multi-omics profiles from four species (human, macaque, marmoset and mouse) that can be integrated as well as incorporated with biological priors.

Teams will optimize models that provide a ranked list of genomic regions (enhancers) for each cell annotation. The predictions will be evaluated against several hundred experimentally validated enhancers.

Submission File

For each peak called and reported in Mouse_atac.h5ad, you should report the rank of top 10,000 peaks for each unique annotation in subclass_Bakken_2022. Thus, each element of the submission will correspond to a peak / subclass_Bakken_2022 pair and contain the corresponding rank as determined by your method.

Your submission should contain a header and have the following format:


Each team should create a Team directory at Dropbox and place their submission file within.