Mean expression can be negative in _get_perturbation_markers

[Zethson]

Report

A kind user: When I examined the function _get_perturbation_markers, I noticed that the mean expression for some groups could be negative, resulting in an invalid Log2FC value (NaN). This negative mean seems to arise from subtracting the expression values of genes in the Perturb-sample cell from the averaged expression of its 20 nearest neighbor cells in the Control sample.

Version information

No response