Function postprocess_seurat_dispersions

src/memory/processing/hvg/mod.rs:184–198

Post-process dispersion statistics for Seurat method to handle edge cases. Handles bins with single genes where standard deviation cannot be computed. Following the original Seurat implementation, single-gene bins get special treatment to ensure they can still contribute to HVG selection. ## Parameters `bin_means` - Mean dispersion values for each bin `bin_stds` - Standard deviation of dispersio

(
    bin_means: &mut [f64],
    bin_stds: &mut [f64],
)

/// - Sets std = mean and mean = 0
/// - This results in normalized dispersion = 1 for these genes
fn postprocess_seurat_dispersions(
    bin_means: &mut [f64],
    bin_stds: &mut [f64],
) -> anyhow::Result<()> {
    // This matches Python's _postprocess_dispersions_seurat
    for i in 0..bin_means.len() {
        if bin_stds[i].is_nan() {
            // For single-gene bins, set std = mean and mean = 0
            // This effectively sets normalized dispersion to 1
            bin_stds[i] = bin_means[i];
            bin_means[i] = 0.0;
        }
    }
    Ok(())
}

/// Create equal-width bins for gene expression levels.
///