Methylation variance as a function of sequencing depth

For each coverage level (or coverage bin), computes the variance of methylation beta values across sites at that depth. This is useful for diagnosing whether low-coverage sites have inflated methylation variance and for setting appropriate coverage thresholds.

Usage

varianceByDepth(object, coverage_bins = NULL, mod_type = NULL)

Arguments

object

A commaData object.

coverage_bins

Integer vector specifying the coverage levels to include. If NULL (default), all unique coverage levels observed across all samples are used. Useful to pass 5:30 to focus on a specific depth range.

mod_type

Character string or NULL. If provided, only sites of the specified modification type are included. Default: NULL (all types).

Value

A data.frame with one row per (coverage level, sample), containing:

coverage

Sequencing depth (integer).

sample_name

Sample identifier.

variance

Variance of beta values at sites with exactly this coverage level. NA if fewer than 2 sites are at this level.

n_sites

Number of sites at this coverage level.

See also

coverageDepth, methylomeSummary

Examples

data(comma_example_data)
vd <- varianceByDepth(comma_example_data, coverage_bins = 5:30)
head(vd)
#>   coverage sample_name    variance n_sites
#> 1        5      ctrl_1          NA       0
#> 2        6      ctrl_1          NA       0
#> 3        7      ctrl_1          NA       0
#> 4        8      ctrl_1          NA       0
#> 5        9      ctrl_1          NA       0
#> 6       10      ctrl_1 0.002581064       2