Background Model Functions

Import from chr3d.peak_based.background_model:

classify_pets


from chr3d.peak_based.background_model import classify_pets
 
classify_pets(
    bedpe_path: str,
    peaks_by_chr: Dict[str, List[Tuple[int, int]]],
    peak_ids_by_chr: Dict[str, List[str]],
    peaks_df: pl.DataFrame,
) -> pl.DataFrame

Classify PETs by peak overlap (P2P / P2D / D2D).

Returns: Polars DataFrame with classified PETs.

extract_templates


from chr3d.peak_based.background_model import extract_templates
 
extract_templates(
    pets_df: pl.DataFrame,
) -> List[Dict[str, Any]]

Build templates from cross-peak P2P PETs.

BackgroundSamplingPhase1


from chr3d.peak_based.background_model import BackgroundSamplingPhase1
 
sampler = BackgroundSamplingPhase1(
    n_bootstrap: int = 1000,
)
 
nb_params = sampler.fit(
    pets_df: pl.DataFrame,
    templates: List[Dict],
)

Sample background, fit Negative Binomial distribution.

calculate_pvalues


from chr3d.peak_based.background_model import calculate_pvalues
 
pvals = calculate_pvalues(
    pets_df: pl.DataFrame,
    nb_params: Dict,
    templates: List[Dict],
)

Compute PMF p-values for PETs.

apply_fdr_corrections


from chr3d.peak_based.background_model import apply_fdr_corrections
 
results = apply_fdr_corrections(
    pvals: pl.DataFrame,
    alpha: float = 0.05,
    method: str = "fdr_bh",
)

Apply multiple testing correction (Benjamini-Hochberg FDR or Bonferroni).