Fuzzy search DSSTox by string similarity

Finds approximate matches for chemical names using DuckDB's native string distance functions. Searches are length-blocked to keep performance reasonable against the 15M-row table.

Usage

dss_fuzzy(
  query,
  method = c("jaro_winkler", "levenshtein", "damerau_levenshtein", "jaccard"),
  threshold = NULL,
  cols = "PREFERRED_NAME",
  limit = 20L,
  con = NULL
)

Arguments

query

A single character string to fuzzy-match.

method

Distance method. One of "jaro_winkler" (default), "levenshtein", "damerau_levenshtein", or "jaccard".

threshold

Similarity/distance cutoff. For jaro_winkler and jaccard, a minimum similarity (0–1, default 0.85). For levenshtein and damerau_levenshtein, a maximum edit distance (default 3).

cols

Which parent_col values to search. Defaults to "PREFERRED_NAME". Set to NULL to search all columns (slower).

limit

Maximum results to return. Default 20.

con

An optional DBI::DBIConnection.

Value

A tibble with DTXSID, parent_col, values, and a similarity or distance score column, ordered by best match first.

See also

Other dsstox: dss_cas(), dss_connect(), dss_diag_freshness(), dss_disconnect(), dss_install(), dss_query(), dss_resolve(), dss_search(), dss_synonyms()

Examples

if (FALSE) { # \dontrun{
dss_fuzzy("Atrazin")
dss_fuzzy("Clorpyrifos", threshold = 0.75)
dss_fuzzy("CC(=O)O", method = "levenshtein", cols = "SMILES", threshold = 2)
} # }