FalconData

FalconData(
    X,
    obs,
    modality,
    units=dict(),
    platform=None,
    species='Homo sapiens',
    uns=dict(),
)

A scoreable dataset.

Attributes

Name	Type	Description
X	pd.DataFrame	samples x features. Row index is the sample id, column index is the exact feature identifier the clocks match on (cg probe ids for arrays, chr:pos for RRBS, canonical marker names for chemistry).
obs	pd.DataFrame	Per-sample annotation, indexed identically to X. Conventional columns are age, sex, tissue, condition – conventional, not required, and each is only needed by the analyses that use it.
modality	Modality	What the numbers are. Governs which clocks are compatible and which preprocessing has already been applied.
units	dict[str, str]	For clinical chemistry only: canonical marker name to unit string. Empty for methylation, where beta is dimensionless by construction.
platform	str | None	For methylation: 27K, 450K, EPICv1, EPICv2, MSA, MammalMethylChip40, or None when unknown.

Methods

Name	Description
coverage	Fraction of features present and not entirely missing.
to_anndata	Convert to AnnData. Needs the anndata extra.
write_h5ad	Write the interchange format both languages read.

coverage

FalconData.coverage(features)

Fraction of features present and not entirely missing.

Presence alone is not enough: GEO series matrices routinely carry a probe column that is NaN for every sample, and counting that as covered is how a clock reports 80% coverage of features it cannot actually see.

to_anndata

FalconData.to_anndata()

Convert to AnnData. Needs the anndata extra.

write_h5ad

FalconData.write_h5ad(path)

Write the interchange format both languages read.