Open haigdjambazian opened 2 years ago
Thanks. There's likely not enough information to partition the 100 cells. Usually, a larger n_cluster is preferred when the n_cells increase.
Yuanhua
Increasing the number of cells to 500 or 1000 caused the ELBO plot to look more normal. My data has two shoulders though but at least the plot is always increasing and has a plateau. Thanks!
Hi, I am working on a single cell DNA dataset (10x genomics) and I managed to run cellsnp-lite and vireo and now I am generating the clustering using the example here. I seem to get some clustering to appear (tried with 100 cells, with MINCOUNT=10,15,25,50,100) but the ELBO plot seems to be decreasing with added donnors instead of increasing to a plateau. The only case where I see an increase (ELBO jumps from 2 to 3 donnors) if for MINCOUNT=100. Can you help me interpret this? Thank you.