Struct ArrayWeightParams 

pub struct ArrayWeightParams {
    pub method: ArrayWeightMethod,
    pub prior_n: f64,
    pub max_iter: usize,
    pub tol: f64,
    pub var_design: Option<Vec<f64>>,
    pub n_var_coef: usize,
}

Knobs for array_weights.

Fields

method: ArrayWeightMethod

Which sweep to run. See ArrayWeightMethod.

prior_n: f64

Prior number of genes pulling the weights towards one.

Enters as a ridge prior_n * Z2' Z2 on the information and, for the REML sweep, a matching prior_n * (w - 1) on the score. limma’s default is 10.

max_iter: usize

Iteration budget for the REML sweep. Ignored by the gene-by-gene sweep.

tol: f64

Convergence tolerance for the REML sweep, on the score-step product scaled by the number of variance coefficients and effective gene count. limma’s default is 1e-5.

var_design: Option<Vec<f64>>

Variance design, row-major n_samples * n_var_coef.

None uses contr.sum(n_samples): one free weight per sample, logs summing to zero. A coarser basis, such as group indicators, pools samples.

n_var_coef: usize

Number of columns in var_design. Ignored when var_design is None.

Implementations

impl ArrayWeightParams

pub fn new( method: ArrayWeightMethod, prior_n: f64, max_iter: usize, tol: f64, var_design: Option<Vec<f64>>, n_var_coef: usize, ) -> Self

Builds a parameter set.

Params

• method - Which sweep to run
• prior_n - Prior gene count shrinking the weights towards one, limma’s 10
• max_iter - REML iteration budget, limma’s 50
• tol - REML convergence tolerance, limma’s 1e-5
• var_design - Optional variance design, row-major n_samples * n_var_coef
• n_var_coef - Number of columns in var_design

Returns

The parameters. Validation happens in array_weights, which knows the sample count these have to agree with.

Trait Implementations

impl Clone for ArrayWeightParams

fn clone(&self) -> Self

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for ArrayWeightParams

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

Formats the value using the given formatter.

impl Default for ArrayWeightParams

fn default() -> Self

limma’s defaults, with the method fixed at ArrayWeightMethod::Reml.

limma’s method = "auto" resolves to "reml" for a complete, unweighted matrix and "genebygene" otherwise. There is no Auto variant, so the default matches a bare arrayWeights(y, design); pass ArrayWeightMethod::GeneByGene for weights or missing values.