INNER CODE UNIT · Python
pgen
antigenomics/vdjtools · python/vdjtools/cli/__init__.py:155
def pgen(
input: Path = typer.Argument(..., help="Table (TSV) or list of CDR3 sequences."),
model: Optional[str] = typer.Option(None, "--model", "-m", help="Built-in locus: TRA TRB TRG TRD IGH IGK IGL."),
source: str = typer.Option("olga", help="Built-in model set: olga | learned."),
model_path: Optional[Path] = typer.Option(None, help="Load a custom model directory instead of a built-in."),
column: Optional[str] = typer.Option(None, "--column", "-c", help="Sequence column (default: auto-detect / first)."),
v_col: Optional[str] = typer.Option(None, "--v-col", help="V-allele column to condition on (default: marginalize)."),
j_col: Optional[str] = typer.Option(None, "--j-col", help="J-allele column to condition on (default: marginalize)."),
seq_type: str = typer.Option("auto", "--type", help="auto | aa | nt."),
mismatches: int = typer.Option(0, help="Amino-acid Hamming ball: 0 exact, 1 sums all single-substitution neighbours."),
no_header: bool = typer.Option(False, "--no-header", help="Input is a bare sequence list (no header row)."),
out: Optional[Path] = typer.Option(None, "--out", "-o", help="Output TSV (default: stdout)."),
) -> None:
"""Compute generation probability (Pgen) for CDR3 sequences — like ``olga-compute_pgen``.
Appends a ``pgen`` column. V/J are marginalized unless ``--v-col``/``--j-col`` are given.
Nucleotide vs amino-acid is auto-detected per sequence; amino-acid input can also sum the
Hamming-distance-1 ball with ``--mismatches 1`` (fast, native).