I have done kBET batch correction for three single cell RNA seq batches consisting of 15, 15 and 20 cells each. Studying gene expression of knock out and wild type mouse and the single cell RNA technology used is SMARTseq2. I have given the input matrix with columns as samples [50] and rows as genes [>20k] as per the kBET GitHub page: https://github.com/theislab/kBET.
I got the input matrix to kBET by doing transpose of count matrix.
The result after running the code was a plot [which showed rejection rate of 0.25] and batch.estimate. The batch estimate gave about three columns with 100 rows [5 of the rows are shown below] and I can't understand what it means. can anyone please provide their expertise in understanding how this data can be applyied to the counts matrix [to get batch corrected count matrix].
kBET.expected kBET.observed kBET.signif
0 0.32 1.5537704456392e-10
0 0.24 1.58665630389798e-06
0 0.36 6.02184968556685e-13
0 0.28 2.14351805194468e-08
0 0.28 2.14351805194468e-08
Thanks,
Prasanna
