Notes for the definition of this webapp
-
Real datasets and synthetic: Loom (DCP)
-
Ground truth: what are the columns?
cell_id, rank (discrete), (pseudotime(real),branch,....)
Folder structure real
/datasets/real/dataset_name/input.loom /datasets/real/dataset_name/truth.?
Folder structure synthetic
As above but everything is under:
/datasets/syn
Output Json for scoreboard (Charlotte can tell us more on the requirements)
Folder structure
/results/dataset_name/method_name/hca_score_metric-name.json
Then in the file we have the fields we want to show in the board:
{key:value} where key is the metric-name and value is the value of that metric
We may have special cases where you have more elements, for example:
{'spearman':0.8, 'kendeall':0.75}
{'stability mean':1.0, 'stability std': 0.2}