Struct Normalize 

pub struct Normalize;

Normalization methods

Each method transforms expression values in a specific way, returning a new ExpressionMatrix. Methods are pure functions: they do not modify the input matrix.

Implementations

impl Normalize

pub fn log2(matrix: &ExpressionMatrix) -> Result<ExpressionMatrix>

Log2 transformation: log2(x+1)

Applies log2(x + 1) to all non-null expression values. This transformation compresses the dynamic range and handles zero values gracefully (log2(0+1) = 0).

Errors

Returns an error if Arrow data construction fails.

pub fn quantile(matrix: &ExpressionMatrix) -> Result<ExpressionMatrix>

Quantile normalization across samples

Normalizes the distribution of expression values across all samples to have the same distribution (the average quantiles across samples). This ensures that differences in expression are due to biology, not technical variation.

Algorithm

1. Sort values within each sample (column) and compute mean ranks
2. Replace each value with the mean of values at that rank across samples
3. Unsort to restore original gene order

Errors

Returns an error if Arrow data construction fails.

pub fn z_score_per_gene(matrix: &ExpressionMatrix) -> Result<ExpressionMatrix>

Z-score normalization per gene (row-wise)

For each gene (row), computes: (x - mean) / std where mean and std are calculated across all samples for that gene. Genes with zero variance (std = 0) are left unchanged.

Errors

Returns an error if Arrow data construction fails.