Flag outlier PCG annotations in an export group — flag_PCG

For each protein-coding gene in an export group, aligns the amino-acid translations across samples and flags annotations that are likely mis-positioned: those whose start or stop extends past, or falls short of, the alignment's well-occupied core by more than a set number of residues (pointing at a start/stop codon placed too long or too short) and those that align poorly to the rest of the group (a low sequence-identity catch-all for badly annotated regions).

Usage

flag_PCG_outliers(
  group,
  db,
  start_aa = 10,
  stop_aa = 10,
  ident_pct = 60,
  genes = NULL
)

Arguments

group: Name of the export group.
db: Path to the project SQLite database.
start_aa: Start-offset threshold (amino acids). A sample is flagged when its start extends past, or falls short of, the alignment core by more than this many residues. Default 10.
stop_aa: Stop-offset threshold (amino acids). As `start_aa`, but for the stop end. Default 10.
ident_pct: Identity threshold (percent). A sample is flagged when its mean pairwise identity to the rest of the group is below this. Default 60.
genes: Optional character vector of gene names. When supplied, only these genes are aligned and flagged (the rest are skipped), e.g. to recompute a single gene edited via "Back to Review". Default `NULL` (all genes).

Value

A list with two elements:

flags: A tibble with one row per flagged (sample, gene): `ID`, `label`, `path`, `scaffold`, `gene`, `pct_identity`, `start_offset`, `stop_offset`, `start_flag`, `stop_flag`, `identity_flag`, `issue`.
alignments: A named list (by gene) of clustering-ordered aligned `AAStringSet` objects, for every gene that has a flagged sample.