Struct DiffSpliceQl 

pub struct DiffSpliceQl<'a> {
    pub df_prior: &'a [f64],
    pub df_residual: &'a [f64],
    pub deviance: &'a [f64],
    pub legacy_zeros: bool,
    pub average_ql_dispersion: Option<f64>,
}

The quasi-likelihood quantities diff_splice needs from glmQLFit.

Passing Some switches the exon and gene tests from chi-squared to moderated F. Fields map onto crate::glm::ql_fit::QlFit:

field	legacy pipeline	current pipeline
df_residual	df_residual_zeros	df_residual_adj
deviance	deviance	deviance_adj
legacy_zeros	true	false
average_ql_dispersion	None	average_ql_dispersion

The fit’s s2_post is absent: diffSpliceDGE discards the exon-level squeezing and squeezes again at the gene level.

Fields

df_prior: &'a [f64]

Prior degrees of freedom from the fit, one value or one per exon. Only its length is read, to default robust.

df_residual: &'a [f64]

Residual degrees of freedom per exon that go with deviance.

deviance: &'a [f64]

Residual deviance per exon that goes with df_residual.

legacy_zeros: bool

Whether df_residual came from df.residual.zeros.

Switches on edgeR’s chi-squared floor: an exon in a gene whose posterior quasi-dispersion fell below one gets no p-value smaller than the unmoderated chi-squared gives. Only the legacy pipeline (which drops structural zeros from the df) can land there spuriously.

average_ql_dispersion: Option<f64>

average.ql.dispersion, which divides the dispersion before the reduced model is refitted so both models share a scale. None on the legacy pipeline.

Trait Implementations

impl<'a> Clone for DiffSpliceQl<'a>

fn clone(&self) -> Self

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl<'a> Copy for DiffSpliceQl<'a>

impl<'a> Debug for DiffSpliceQl<'a>

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.