iscc.data.RNABatchHyperParams dataclass

RNABatchHyperParams(protocol: str = '10x', sigma_batch: float = 0.1, mu_lib: float = 2400.0, sigma_lib: float = 0.43, dispersion: float = 0.3, ambient_frac: float = 0.05, doublet_rate: float = 0.05, dropout_mid: float = 0.0, dropout_shape: float = 1.0, well_sigma: float = 0.0, depth_batch_sigma: float = 0.05, kappa: float = 50.0)

The scRNA assay's technical parameters — a defaults container.

Holds the protocol-typical magnitudes of the scRNA count model: the batch and library-size scales, the NB overdispersion and logistic dropout curve, and the ambient / doublet contamination rates. The scRNA assay builds one from its protocol preset and lets you override any field, and estimate_rna fits these same parameters from real data.

Attributes:

Name	Type	Description
protocol	str	Protocol preset the magnitudes come from (e.g. "10x", "smartseq3").
sigma_batch	float	Per-gene batch-factor LogNormal sd; beta_gb ~ LogNormal(0, sigma_batch^2), shared across every cell of the batch.
mu_lib	float	Mean library size (counts / cell) — the sequencing depth.
sigma_lib	float	Per-cell library-size LogNormal sd.
dispersion	float	NB overdispersion phi (var = mu + phi*mu^2); used only by the "nb" count model (phi <= 0 collapses to Poisson).
ambient_frac	float	Fraction of the library that is ambient ("soup") contamination.
doublet_rate	float	Fraction of barcodes that are doublets (two cells merged).
dropout_mid	float	Logistic dropout midpoint in expected-count space (0 disables dropout).
dropout_shape	float	Logistic dropout steepness (active only when dropout_mid > 0).
well_sigma	float	Per-cell "well" (plate-position) LogNormal sd (Smart-seq3 plate nesting).
depth_batch_sigma	float	Per-batch depth-shift LogNormal sd (so batches differ in depth).
kappa	float	Dirichlet concentration for the "dm" count model only (large kappa -> multinomial/Poisson-like, small kappa -> lumpy over-dispersed proportions).