Skip to contents

Run ordinations on a distance matrix.

Source: R/distmat_ord_table.r
distmat_ord_table.Rd

Run ordinations on a distance matrix.

Usage

distmat_ord_table(dm, ord = "PCoA", k = 2L, seed = 0, ...)

Arguments

dm

A dist-class distance matrix, as returned from bdiv_distmat() or stats::dist(). Required.

ord

Method for reducing dimensionality. Options are:

"PCoA" -

Principal coordinate analysis; stats::cmdscale().

"UMAP" -

Uniform manifold approximation and projection; uwot::umap().

"NMDS" -

Nonmetric multidimensional scaling; vegan::metaMDS().

"tSNE" -

t-distributed stochastic neighbor embedding; Rtsne::Rtsne().

Multiple/abbreviated values allowed. Default: "PCoA"

k

Number of ordination dimensions to return. Either 2L or 3L. Default: 2L

seed

Random seed for the stochastic ordinations (NMDS, tSNE, UMAP), so that they are reproducible. PCoA is deterministic. Must be a non-negative integer. Default: 0

...

Additional arguments for the ord function. For tSNE, perplexity defaults to 30, or to (n - 1) / 3 for fewer than 91 samples.

Value

A tibble data.frame with columns .ord, .sample, .x, .y, and (optionally) .z. The attribute 'axes' is a tibble with one row per axis (.ord, .axis, .label, .var): for PCoA, .var is the percent of the variance (the positive eigenvalues) explained by that axis and .label reads e.g. "PCoA 1 (24.3%)"; for the other ordinations .var is NA.

See also

Other ordination: bdiv_ord_plot(), bdiv_ord_table()

Examples

    library(rbiom)

    dm  <- bdiv_distmat(hmp50, "bray")
    ord <- distmat_ord_table(dm, "PCoA")
    head(ord)
#> # Ordination: stats::cmdscale(dm, k = 2, eig = TRUE)
#> # A tibble:   6 × 4
#>   .ord  .sample     .x       .y
#>   <fct> <chr>    <dbl>    <dbl>
#> 1 PCoA  HMP01   -0.358  0.00236
#> 2 PCoA  HMP02   -0.420  0.0104 
#> 3 PCoA  HMP03   -0.382  0.0135 
#> 4 PCoA  HMP04   -0.388  0.00485
#> 5 PCoA  HMP05   -0.415 -0.0115 
#> 6 PCoA  HMP06   -0.414 -0.00703

    attr(ord, 'axes')
#> # A tibble: 2 × 4
#>   .ord  .axis .label          .var
#>   <fct> <chr> <chr>          <dbl>
#> 1 PCoA  .x    PCoA 1 (21.6%)  21.6
#> 2 PCoA  .y    PCoA 2 (14.7%)  14.7