normalization

[shimasadri]

I tried the following normalization in python
sc.pp.normalize_total(adata)
sc.pp.scale(adata, max_value=1e4)

but when I run compass I get the following warning message
RuntimeWarning: invalid value encountered in log2
result = getattr(ufunc, method)(*inputs, **kwargs)

How can I modify the normalization (in pyhton) to solve this?

Thank you

[DPLemonade]

Hi @shimasadri! COMPASS expects normalized counts as input so running sc.pp.normalize_total(adata) should already be fine (assuming that adata contains raw counts). Also note that COMPASS expects linear-scale data instead of log-transformed data.