LOO: Leave-one-out analysis for quantitative evaluation of the...

In IgGeneUsage: Differential gene usage in immune repertoires

Description

IgGeneUsage detects differential gene usage in immune repertoires that belong to two biological conditions with its function DGU. To assert quantitatively the robustness of the estimated probability of DGU (pi), IgGeneUsage has a built-in procedure for a fully Bayesian leave-one-out (LOO) analysis. During each step of LOO, we discard the data of one of the repertoires, and use the remaining data to analyze for DGU with IgGeneUsage. In each step we recorded pi for all genes. Therefore, by evaluating the variability of pi for a given gene, we can we assert quantitatively its robustness.

Notice, however, that for datasets that include many repertoires (e.g. 100) LOO can be computationally costly.

Usage

LOO(usage.data, mcmc.warmup, mcmc.steps,
    mcmc.chains, mcmc.cores, hdi.level,
    adapt.delta, max.treedepth)

Arguments

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).

Details

IgGeneUsage invokes the function DGU in each LOO step. For more details see help for DGU or vignette 'User Manual: IgGeneUsage'.

Value

loo.summary

differential gene usage statistics for each gene of a given LOO step. 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; 4) loo.id (LOO step ID); 5 Neff (effective sample size), Rhat (potential scale reduction factor)

Author(s)

Simo Kitanovski <[e-mail]>

See Also

DGU, Ig, IGHV_Epitopes, IGHV_HCV, Ig_SE

Examples

# input data
data(Ig)
head(Ig)

# Alternative:
# use SummarizedExperiment input data
# data(Ig_SE)


# run leave-one-out (LOO)
L <- LOO(usage.data = Ig,
         mcmc.warmup = 500,
         mcmc.steps = 1500,
         mcmc.chains = 2,
         mcmc.cores = 1,
         hdi.level = 0.95,
         adapt.delta = 0.95,
         max.treedepth = 13)

IgGeneUsage documentation built on Nov. 8, 2020, 7:47 p.m.