Low values for iLISI score

[Mari123i]

Hello,

I graph connectivity, kBET and iLISI on the same dataset. While the former two metrics seem to reflect a pretty solid batchmixing, iLISI is ridiculously low. Is there anything that needs to be done in terms of scaling?

This is my method call:

Thank you for your help!

[YY-SONG0718]

Dear developers,

I have encountered the same issue. Between integrated and unintegrated data, iLISI scores are indifferent, or the difference is at 16 digits scale (the precision of python floating point numbers). In contrast, all other metrics show informative differences.

Could you please suggest if and how to use iLISI scores in this case?

Thank you very much.