`addMDS` drops smallest sample when `sample = min(colSums(assay(tse, "counts")))`

[0xMuluh]

Reported by Anna Kaisanlahti:

• addAlpha(sample = min(colSums(assay(tse, "counts"))) keeps the smallest-depth sample.
• addMDS(sample = min(colSums(assay(tse, "counts"))) drops that sample unless sample is set noticeably lower.

Example: smallest sample depth = 5887 reads; addMDS(sample = 5887) drops it, but sample = 5390 keeps it. This suggests inconsistent subsampling logic between addAlpha() and addMDS() (or their underlying helpers).

Steps to reproduce

min_reads <- min(colSums(assay(tse, "counts")))

# 1) Alpha: smallest sample retained
tse_alpha <- addAlpha(tse, sample = min_reads, niter = 100)

# 2) Beta: check if smallest sample still present
tse_mds <- addMDS(tse, sample = min_reads, niter = 100)

Expected behavior

addMDS() should treat sample consistently with addAlpha() / rarefyAssay(): using the true minimum depth should not drop that smallest sample.

Actual behavior

addMDS() excludes the smallest sample when sample = min_reads; only lower values keep it.

Request

• Confirm if this behavior is intentional.
• If not, align addMDS() with addAlpha() and/or add a clear warning when sample will cause sample exclusion.