Dear organizer,
The submitting example in the MCP-counter gave 2 data sets (ds1, ds2) and totally 5 samples (2 samples in ds1, 3 samples in ds2).
However, our method is highly dependent on the sample-based information to predict cell type proportion. Thus, few samples will cause a serious problem.
Could you please provide a rough number of sample or a range of the number of samples in the validation phase (eg. larger than 30 )?
Best,
Created by wennan chang chang91 Hi Dr. Brain,
Thank you so much. That makes sense to me.
Best,
Wennan Also, please note that the data you are referring to are the data in express lane queue. These are not intended to represent a real dataset, but only to give you an opportunity to efficiently verify that your Docker image is working and outputting a prediction file with the correct format.
Best,
Brian Hi @chang91 ,
The validation data will consist of several datasets--each should be ~100 samples.
The leaderboard datasets are much smaller. There will be 14 of them will the following sizes:
n=5
13
14
14
15
15
21
25
29
34
47
56
173
198
These will be split across the two leaderboard rounds.
Regards,
Brian
Drop files to upload
How many sample will provide in the validation phage? page is loading…