biobouncer
Validate Biological Identifiers and Inputs
Gene symbols, ontology terms, variant formats, and database
accessions are validated through one small application programming
interface (API), either offline against identifier patterns and bundled
snapshots or live against a source web service. Results are designed to
match those of the companion 'Python' package for the same inputs. Input
order and length are preserved, and errors are reported explicitly rather
than as a silent failure. Identifier patterns follow the registry and
nomenclature described in Hoyt et al. (2022)
biobouncer (R package) 
This directory holds the R package for biobouncer. See the repository root
README.md for what biobouncer does and PLAN.md for the architecture.
Offline pattern and cache modes, live remote mode, and existence mode
(snapshot first, then remote) work across 46 sources.
Usage
library(biobouncer)
# List what can be checked.
sources()
source_info()
# pattern mode: is the string well-formed?
check_id(c("MONDO:0005148", "mondo:5148", "GO:0006915"), source_db = "mondo")
# cache mode: does the id exist in a pinned snapshot?
check_id("MONDO:0005148", source_db = "mondo", how = "cache", version = "sample")
# Snapshot management.
biobouncer_snapshots()
biobouncer_cache_dir()
biobouncer_pull("go") # download a full snapshot into the cache directory
# Just the verdict.
is_valid_id("P04637", source_db = "uniprot")
check_id() returns a tibble with one row per input and the columns input,
valid, normalized, suggestion, source_db, version, species, how,
and error (the reason a remote check was left indeterminate, else NA).
Validation frameworks
Adapters wrap the core classifier so it plugs into common validation frameworks. They never reimplement any checks.
# checkmate style: check, assert, or test.
check_valid_id(c("MONDO:0005148", "mondo:5148"), "mondo")
assert_valid_id("MONDO:0005148", "mondo")
test_valid_id("MONDO:0005148", "mondo")
# shinyvalidate rule.
iv <- shinyvalidate::InputValidator$new()
iv$add_rule("term", sv_biobouncer("mondo"))
# Data-frame validation. id_predicate() returns an elementwise predicate.
is_mondo <- id_predicate("mondo")
# assertr:
df |> assertr::assert(is_mondo, term)
# validate:
rules <- validate::validator(good_terms = is_mondo(term))
validate::confront(df, rules)
Development
From the repository root:
# generate NAMESPACE and man pages
roxygen2::roxygenise("pkg-r")
# run the tests
testthat::test_dir("pkg-r/tests/testthat")
Shared source definitions and the conformance corpus live in shared/ at the
repository root and are vendored into inst/extdata/ by
python tools/sync_shared.py. Do not edit the vendored copies by hand.