hyper.gam

hyper.gam provides a supervised learning framework for deriving biomarkers based on single-cell distribution quantiles, enabling the estimation of quantile index predictors from immunohistochemistry images.

Key Features:

• Quantile Function Conversion: Single-cell data are converted into sample quantile functions for use as predictors.
• Scalar-on-Function Regression: Quantile functions are used as predictors in scalar-on-function regression models to estimate the integrand surface.
• Quantile Index Predictor Estimation: The estimated integrand surface defines quantile index predictors based on single-cell expression levels in a new test set.
• Integrand Surface Exploration: Visual tools enable exploration of the estimated integrand surfaces.

Scientific Applications:

• Functional Protein Biomarker Derivation: Derives biomarkers from proteins with meaningfully quantitative expression levels in immunohistochemistry images.
• Tumor Microenvironment Analysis: Accounts for heterogeneous protein expression levels in tissue for biomarker development.
• General Single-Cell Data Analysis: Applicable to other types of single-cell data beyond immunohistochemistry images.

Methodology:

Single-cell data are converted into sample quantile functions, which are then used as predictors in scalar-on-function regression models to estimate the integrand surface.