ErrorEstimationConfig

class pyprophet._config.ErrorEstimationConfig(parametric: bool = False, pfdr: bool = False, pi0_lambda: float | List[float] = (0.1, 0.5, 0.05), pi0_method: str = 'bootstrap', pi0_smooth_df: int = 3, pi0_smooth_log_pi0: bool = False, lfdr_truncate: bool = True, lfdr_monotone: bool = True, lfdr_transformation: str = 'probit', lfdr_adj: float = 1.5, lfdr_eps: float = np.float64(1e-08))[source]

Bases: object

Configuration for global and local FDR (false discovery rate) estimation.

parametric

Whether to use parametric estimation of p-values.

Type:: bool

pfdr

Whether to compute positive FDR (pFDR) instead of traditional FDR.

Type:: bool

pi0_lambda

Lambda range or fixed value for pi0 estimation (e.g., [0.1, 0.5, 0.05] or [0.4, 0.0, 0.0]).

Type:: Any

pi0_method

Method to estimate pi0; either ‘smoother’ or ‘bootstrap’.

Type:: str

pi0_smooth_df

Degrees of freedom for smoothing function in pi0 estimation.

Type:: int

pi0_smooth_log_pi0

Whether to apply smoothing on log(pi0) estimates.

Type:: bool

lfdr_truncate

If True, truncate local FDR values above 1 to 1.

Type:: bool

lfdr_monotone

If True, enforce monotonic increase of local FDR values.

Type:: bool

lfdr_transformation

Transformation of p-values; either ‘probit’ or ‘logit’.

Type:: str

lfdr_adj

Smoothing bandwidth adjustment factor in local FDR estimation.

Type:: float

lfdr_eps

Threshold for trimming empirical p-value distribution tails.

Type:: float

__eq__(other): Return self==value.

__hash__ = None

__init__(parametric: bool = False, pfdr: bool = False, pi0_lambda: float | List[float] = (0.1, 0.5, 0.05), pi0_method: str = 'bootstrap', pi0_smooth_df: int = 3, pi0_smooth_log_pi0: bool = False, lfdr_truncate: bool = True, lfdr_monotone: bool = True, lfdr_transformation: str = 'probit', lfdr_adj: float = 1.5, lfdr_eps: float = np.float64(1e-08)) → None

__repr__()[source]: Return repr(self).

__str__()[source]: Return str(self).

__weakref__: list of weak references to the object (if defined)