Compare refseq sequences across all objects in a list_phyloseq

For each pair of phyloseq objects in x, computes pairwise sequence similarity between @refseq slots using k-mer Jaccard similarity, then groups near-identical sequences across all objects via connected components (union-find). Returns a list of results per threshold, and one Venn diagram per threshold (requires ggVennDiagram).

Similarity is approximated as p ≈ (2j / (1 + j))^(1/k) where j is the Jaccard index of shared k-mer presence and k is the k-mer size. The approximation is accurate within ~2 percentage points for identities in the 90–100% range.

Usage

refseq_comp_lpq(x, thresholds = c(0, 1, 2, 3, 5, 10), k = 7, verbose = TRUE)

Arguments

x

(required) A list_phyloseq object. All phyloseq objects must have a populated @refseq slot. Supports 2–7 objects (Venn diagrams are suppressed for > 7).

thresholds

(numeric vector, default c(0, 1, 2, 3, 5, 10)) Maximum dissimilarity thresholds in percent. 0 means exact string matches only; 10 means sequences within 10% divergence are grouped together.

k

(integer, default 7) K-mer size for building presence/absence profiles. Larger k increases specificity; smaller k increases sensitivity for divergent sequences.

verbose

(logical, default TRUE) If TRUE, print progress messages.

Value

A list of class "refseq_comp_lpq_result" with:

venn_plots

Named list of ggVennDiagram objects (one per threshold, names follow "<threshold>_pct"), or NULL if ggVennDiagram is not installed or N > 7.

results

List of per-threshold results, each containing threshold, venn_data (named list of component IDs per set), and n_shared (named list of shared cluster counts per pair).

labels

Character vector of phyloseq object names.

thresholds

The thresholds used.

k

The k-mer size used.

Details

Author

Adrien Taudière

Examples

lpq <- list_phyloseq(list(
  a = data_fungi_mini,
  b = prune_taxa(taxa_names(data_fungi_mini)[1:20], data_fungi_mini)
))
#> ℹ Building summary table for 2 phyloseq objects...
#> ℹ Computing comparison characteristics...
#> ℹ Checking sample and taxa overlap...
#> ℹ Detected comparison type: REPRODUCIBILITY
#> ℹ 137 common samples, 20 common taxa
#> ✔ list_phyloseq created (REPRODUCIBILITY)
res <- refseq_comp_lpq(lpq)
#> Building k-mer profiles (k = 7) for 2 objects...
#> Computing 1 pairwise similarity matrices...
#>   a vs b