Skip to contents

Rarefy Counts to a Constant Depth

Source: R/rarefy.r
rarefy.Rd

This function reduces the number of observations (reads) in each sample to a fixed integer value (depth). Samples with fewer observations than the specified depth are discarded.

Rarefaction is a common technique in microbiome analysis used to account for uneven sequencing effort across samples. By standardizing the library size, it allows for fair comparisons of alpha and beta diversity metrics.

The transformation is recorded in the object's history (biom$code).

Usage

rarefy(
  biom,
  depth = NULL,
  seed = 0L,
  inflate = FALSE,
  clone = TRUE,
  cpus = n_cpus()
)

Arguments

biom

An rbiom object, or any value accepted by as_rbiom().

depth

The number of observations to keep per sample. Must be an integer greater than 0.

  • If NULL (the default), a depth is automatically selected that retains at least 10% of the dataset's total abundance while maximizing the number of samples kept. See suggest_rarefy_depth() for the specific heuristic used. The selected depth is recorded in biom$code.

  • Samples with total counts less than depth will be dropped from the result.

seed

Random seed for selecting which observations to keep. The same seed, depth, and input always give the same result. Must be a non-negative integer. Default: 0

inflate

Logical. Handling for non-integer data (e.g. relative abundances).

  • FALSE (Default): The function will error if non-integers are detected. Rarefaction requires discrete counts (integers).

  • TRUE: The function will automatically rescale (inflate) non-integers to integers using biom_inflate() before rarefying. This is useful for 'shoehorning' metagenomic relative abundance data into diversity functions that strictly require integers.

clone

Create a copy of biom before modifying. If FALSE, biom is modified in place as a side-effect. See speed ups for use cases. Default: TRUE

cpus

The number of CPUs to use. Set to NULL to use all available, or to 1 to disable parallel processing. Does not affect results. Default: ecodive::n_cpus() (all available cores)

Value

An rbiom object.

Details

Normalizes the library sizes of a dataset by randomly sub-sampling observations from each sample to a specific depth.

See also

suggest_rarefy_depth() for details on the default depth selection.

Other transformations: biom_inflate(), biom_relativize(), biom_rescale(), modify_metadata, slice_metadata, subset, with

Examples

    library(rbiom)

    biom <- hmp50[1:5]
    sample_sums(biom)
#> HMP01 HMP02 HMP03 HMP04 HMP05 
#>  1660  1371  1353  1895  3939 

    # Rarefy to the lowest sample depth
    # (All samples are kept, but counts are reduced)
    biom_rare <- rarefy(biom, depth = min(sample_sums(biom)))
    sample_sums(biom_rare)
#> HMP01 HMP02 HMP03 HMP04 HMP05 
#>  1353  1353  1353  1353  1353 

    # Auto-select depth (may drop samples with low coverage)
    biom_auto <- rarefy(biom)
    sample_sums(biom_auto)
#> HMP01 HMP02 HMP03 HMP04 HMP05 
#>  1353  1353  1353  1353  1353 

    biom_auto$code
#> # Generated by rbiom 4.0.0
#> library(rbiom)
#> 
#> biom <- hmp50
#> biom <- biom[1:5]
#> biom <- rarefy(biom, depth = 1353L)