search

AltschulerWu-Lab / scScope

48 stars 12 forks source link

ValueError: Input contains NaN, infinity or a value too large for dtype('float64'). #8

Closed xuebaliang closed 3 years ago

xuebaliang commented 5 years ago

I recently run your code on some other dataset. But when I take out the latent representation, I find it has many nan value. Do you meet it? What's your solution?

feng-bao-ucsf commented 4 years ago

I recently run your code on some other dataset. But when I take out the latent representation, I find it has many nan value. Do you meet it? What's your solution?

I did not meet this circumstance. You can try to perform data filtering before inputing them to scScope, eg removing genes with lots of zero entris, normalizing data using library size. You can follow our data preprocessing steps in demo.py.