Goodness-of-Fit Filters for Sparse Omics Data

Implements feature selection filters designed for sparse/zero-inflated omics data. Standard variance-based filters often discard biologically relevant features that are "turned off" (zero) in one condition but expressed in another.

Details

Two filters are provided: - **FilterGOF_KS**: Kolmogorov-Smirnov test comparing distributions between classes - **FilterHurdle**: Two-part hurdle model testing both zero-frequency and magnitude

These filters are designed for binary classification tasks in biomarker discovery.