fix: normalise Kleborate column names in import_kleborate() to prevent silent data loss

[efosternyarko]

Problem

Different versions of Kleborate (and the same version depending on whether --strain is used) produce inconsistently capitalised column names. For example:

Column	Kleborate v2/v3 (--strain)	Kleborate v3 (default)
Aminoglycosides	Agly_acquired	AGly_acquired
Chromosomal beta-lactam	Bla_Chr	Bla_chr

Because import_kleborate() uses select(any_of(kleborate_class_table$Kleborate_Class)), columns whose names differ only in capitalisation are silently dropped — no warning, no error. Entire drug classes disappear from the returned genotype table.

Discovered while running the function on real Klebsiella pneumoniae data from The Gambia: Aminoglycosides showed 0 markers and 0% sensitivity, tracing back to Agly_acquired being silently skipped.

Fix

Add a case-insensitive rename step immediately after the sample_col rename (line 400) and before the select(any_of(...)) call. Any input column whose name matches an expected column name case-insensitively — but not exactly — is renamed to the expected capitalisation. An informative message() is emitted when this happens so users are aware of the discrepancy in their Kleborate output.

expected_cols_lower <- setNames(
  kleborate_class_table$Kleborate_Class,
  tolower(kleborate_class_table$Kleborate_Class)
)
cols_to_rename <- names(in_table)[
  tolower(names(in_table)) %in% names(expected_cols_lower) &
  !(names(in_table) %in% kleborate_class_table$Kleborate_Class)
]
if (length(cols_to_rename) > 0) {
  rename_vec <- setNames(cols_to_rename, expected_cols_lower[tolower(cols_to_rename)])
  in_table <- in_table %>% rename(!!!rename_vec)
  message(
    "Normalised Kleborate column name(s) to match expected capitalisation: ",
    paste(cols_to_rename, "->", names(rename_vec), collapse = ", ")
  )
}

Testing

Verified against:

• Kleborate output with Agly_acquired / Bla_Chr (older capitalisation): aminoglycosides now correctly returned with 15 markers across 87 isolates
• Kleborate output already using AGly_acquired / Bla_chr (expected capitalisation): no rename, no message, behaviour unchanged

[efosternyarko]

Closing — the capitalisation mismatch (Agly_acquired / Bla_Chr) was in a manually prepared Excel file used in our analysis, not from Kleborate or Pathogenwatch output. The kleborate_classes lookup is correct. Will fix the column names in our data file locally. Sorry for the noise!