allelicRatio: Fit beta-binomial across cell types

In Wancen/airpart: Differential cell-type-specific allelic imbalance

Fit beta-binomial across cell types

Description

This function performs additional inference on the allelic ratio across cell types, giving posterior mean and credible intervals per cell type. A Cauchy prior is centered for each cell type on the allelic ratio from the fused lasso across all genes in the gene cluster (or using a weighted means if the fused lasso is not provided).

Usage

allelicRatio(sce, formula, nogroup = FALSE, level = 0.95, DAItest = FALSE, ...)

Arguments

sce	A SingleCellExperiment containing assays ("ratio", "counts") and colData ("x", "part")
formula	The same formula object used in fusedLasso or used in ns when want to detect smooth continuous-axis estimates. eg. formula <- ratio ~ ns(x, df = 5)
nogroup	Indicate whether there is previous group step. Default is FALSE represents either Generalized fused lasso or nonparametric method is used to derive partition. nogroup == TRUE means the hypothesis that groups of cell types sharing a common regulatory context is not valid
level	the level of credible interval (default is 0.95)
DAItest	Indicate whether to do Likelihood Ratio test on differential allelic imbalance(DAI) or equivalent to heterogeneity.
...	arguments to pass to apeglm functions

Value

posterior mean ("ar") for allelic ratio estimate is returned in the rowData for each cell type, as well as the "s" value, "fsr" false sign rate and credible interval ("lower" and "upper"). One can use "fsr" < 0.005 or credible intervals contain 0.5 or not for AI test significance. "p.value" shows DAI test result and "adj.p.value" is false discovery rate corrected p values.

Examples

sce <- makeSimulatedData()
sce <- preprocess(sce)
sce <- geneCluster(sce, G = seq_len(4))
sce_sub <- wilcoxExt(sce, genecluster = 1)
sce_sub <- allelicRatio(sce_sub, DAItest = TRUE)

Wancen/airpart documentation built on March 12, 2023, 11:53 a.m.