inmoose.diffexp.meta_de

inmoose.diffexp.meta_de(de_results, alpha=0.05, min_common_genes=None)

Combine logFC and p-values of differential expression analyses

log-fold-changes are combined using a random-effect model, where the random effects variance is iteratively estimated with the Paule-Mandel method. Confidence intervals for the combined log-fold-change values are computed assuming a normal distribution and without scaling.

p-values are combined using Fisher’s combined probability test, then adjusted for multiple testing with Benjamini-Hochberg procedure.

Parameters:

  • de_results (list of DEResults) – the list of differential expression results to combine. Depending on the use-case, it can be results obtained with different tools on the same dataset, results obtained with the same tool on different datasets, or any combination thereof

  • alpha (float between 0 and 1) – significance level for the confidence intervals. Defaults to 0.05.

  • min_common_genes (int or None) – minimal number of genes all the elements of de_results need to have in common. Below this threshold, an error will be raised. If None, then all elements of de_results must have the same set of genes.

Returns:

a dataframe indexed by genes with the following columns:

  • "combined logFC": the combined log-fold-change

  • "combined logFC (CI_L)": the lower bound of the confidence interval for the combined log-fold-change

  • "combined logFC (CI_R)": the lower bound of the confidence interval for the combined log-fold-change

  • "adjusted combined pval": the combined p-value, adjusted for multiple testing

Return type:

pd.DataFrame