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.
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
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:
chr,start,end,subclass_Bakken_2022,rank chr1,1,500,"Pvalb",1 chr1,501,1000,"Pvalb",2 ... chr1,1,500,"Astro",1000 chr1,501,1000,"Astro",10000
Each team should create a Team directory at Dropbox and place their submission file within.