pylimma.pred_fcm

pylimma.pred_fcm(fit, coef=1, var_indep_of_fc=True, all_de=True, prop_true_null_method='lfdr', *, key='pylimma')[source]

Predictive (empirical-Bayes shrunken) fold changes.

Port of R limma’s predFCm (predFCm.R). Uses the eBayes posterior variance together with an estimated proportion of true nulls to shrink log-fold-changes toward zero.

Parameters:

  • fit (AnnData, MArrayLM, or dict) – Fit from lm_fit() + e_bayes(). For AnnData input, the fit is read from adata.uns[key].

  • coef (int, default 1) – Zero-based coefficient index (R uses 1-based).

  • var_indep_of_fc (bool, default True) – If True, assume the prior variance is independent of fold-change magnitude.

  • all_de (bool, default True) – If True, assume all genes are differentially expressed.

  • prop_true_null_method ({"lfdr", "convest", "mean", "hist"}) – Forwarded to prop_true_null().

  • key (str, default "pylimma") – Key for fit results in adata.uns (AnnData input only).

Return type:

ndarray