in metacells/pipeline/collect.py, _collect_metacell() performs a weighted geomean for the regularized expression (by the way, why not use scipy.stats.mstats.gmean with the weights parameter?), yet performs an unweighted geomean for the regularization values, i.e., current code does:
regularization_of_metacell = ss.mstats.gmean(regularization_per_cell_of_metacell)
while seems to me the code should be:
regularization_of_metacell = ss.mstats.gmean(regularization_per_cell_of_metacell, weights=weight_per_cell_of_metacell)
in
metacells/pipeline/collect.py
,_collect_metacell()
performs a weighted geomean for the regularized expression (by the way, why not usescipy.stats.mstats.gmean
with the weights parameter?), yet performs an unweighted geomean for the regularization values, i.e., current code does:regularization_of_metacell = ss.mstats.gmean(regularization_per_cell_of_metacell)
while seems to me the code should be:regularization_of_metacell = ss.mstats.gmean(regularization_per_cell_of_metacell, weights=weight_per_cell_of_metacell)