DGU {IgGeneUsage} | R Documentation |
IgGeneUsage detects differential gene usage in immune repertoires that belong to two biological conditions.
DGU(usage.data, mcmc.warmup, mcmc.steps, mcmc.chains, mcmc.cores, hdi.level, adapt.delta, max.treedepth)
usage.data |
Data.frame with 4 columns: 'sample_id' = character identifier of each repertoire, 'condition' = character key representing each of the two biological conditions, 'gene_name' = character name of each gene to be tested for differential usage, 'gene_usage_count' = number of rearrangements belonging to a specific sample_id x condition x gene_name. Alternatively, usage.data can be a SummarizedExperiment object. See examplary data 'data(Ig_SE)' for more information. |
mcmc.chains, mcmc.warmup, mcmc.steps, mcmc.cores |
Number of MCMC chains (default = 4), number of cores to use (default = 1), length of MCMC chains (default = 1,500), length of adaptive part of MCMC chains (default = 500). |
hdi.level |
Highest density interval (HDI) (default = 0.95). |
adapt.delta |
MCMC setting (default = 0.95). |
max.treedepth |
MCMC setting (default = 12). |
The input to IgGeneUsage is a table with usage frequencies for each gene of a repertoire that belongs to a particular biological condition. For the analysis of differential gene usage between two biological conditions, IgGeneUsage employs a Bayesian hierarchical model for zero-inflated beta-binomial (ZIBB) regression (see vignette 'User Manual: IgGeneUsage').
glm.summary |
differential gene usage statistics for each gene. 1) es = effect size on differential gene usage (mean, median standard error (se), standard deviation (sd), L (low boundary of HDI), H (high boundary of HDI); 2) contrast = direction of the effect; 3) pmax = probability of differential gene usage |
test.summary |
differential gene usage statistics computed with the Welch's t-test (columns start with 't'), and Wilcoxon signed-rank test (columns start with 'u'). For both test report P-values, FDR-corrected P-values, Bonferroni-corrected P-values. Additionally, we report t-value and 95% CI (from the t-test) and U-value (from the Wilcoxon signed-rank test). |
glm |
stanfit object |
ppc.data |
two types of posterior predictive checks: 1) repertoire- specific, 2) gene-specific |
Simo Kitanovski <simo.kitanovski@uni-due.de>
LOO, Ig, IGHV_Epitopes, IGHV_HCV, Ig_SE
# input data data(Ig) head(Ig) # Alternative: # use SummarizedExperiment input data # data(Ig_SE) # run differential gene usage (DGU) M <- DGU(usage.data = Ig, mcmc.warmup = 500, mcmc.steps = 1500, mcmc.chains = 2, mcmc.cores = 2, hdi.level = 0.95, adapt.delta = 0.95, max.treedepth = 13)