grassp.tl.mr_score

Contents

grassp.tl.mr_score#

mr_score(data, condition_key, replicate_key, mcd_proportion=0.75, assume_centered=True, n_iterations=11, key_added='mr_scores', fraction_key=None)[source]#

M/R score for detecting protein translocation.

Implements the approach from Itzhak et al. 2016 where M is a moderated t-statistic (via Minimum Covariance Determinant) and R the measure of reproducibility between replicates.

Parameters:
data AnnData

AnnData object containing all conditions/replicates stacked in variables.

condition_key str

data.var column indicating the experimental condition (e.g. control, treatment).

replicate_key str

data.var column indicating biological replicates.

mcd_proportion float (default: 0.75)

Parameters forwarded to sklearn.covariance.MinCovDet().

assume_centered bool (default: True)

Parameters forwarded to sklearn.covariance.MinCovDet().

n_iterations int (default: 11)

Parameters forwarded to sklearn.covariance.MinCovDet().

key_added str (default: 'mr_scores')

Base name for the two new columns {key_added}_M and {key_added}_R written to data.obs.

fraction_key str | None (default: None)

data.var column naming the fraction of each sample. If given, control and treatment samples are paired by fraction, and each replicate must have the same fractions in both conditions. If None, they are paired by column order, so the fractions must appear in the same order in both conditions.

Return type:

None