TOP function reference

This is a practical map of the supported TOP interface. In R, open the full argument-level help with ?function_name, for example ?top_fit.

Recommended workflow

Function	Use it for
top_fit()	Fit one validated two-stage polytomous logistic-regression model.
top_fit_custom()	Fit a model with a scientifically defined, custom second-stage design.
top_score_support()	Fit and retain the null-model support used for a multi-variant score scan.
top_score_scan()	Score-test a matrix of variants, reusing top_score_support() output.
top_followup()	Fit detailed models for selected variants after a scan.
top_meta()	Combine aligned score vectors and information matrices from independent studies.

top_fit() returns a top_model. Its principal components are second.stage.test, first.stage.test, and global.test; coef(), vcov(), summary(), and print() are available for the fitted object.

Input and design helpers

Function	Use it for
GenerateZstandard()	Create TOP's standard second-stage design from a tumor-characteristic matrix.
GenerateZDesignBaselineonly()	Construct a baseline-only design.
GenerateZDesignAdditive()	Construct a design with marker main effects.
GenerateZDesignPairwiseInteraction()	Construct a design with marker main effects and pairwise interactions.
GenerateZDesignSaturated()	Construct a design with a separate effect for each retained subtype.
GenerateZDesignCombination()	Construct a specified combination of second-stage effects.

For the standard top_*() functions, supply an outcome y whose first column is case-control status (0/1) and whose remaining columns are tumor characteristics. Controls must have NA tumor characteristics. Supply a numeric missing_value when cases use a distinct missing-tumor code.

Legacy API

Existing scripts can continue to use TwoStageModel(), ScoreTest(), ScoreTestMixedModel(), and ScoreMetaAnalysis(). New work should prefer the top_*() functions above: they validate inputs and return consistently named objects. The historical low-level helpers remain available for reproducibility, but they are not the recommended starting point for new analyses.

For the complete legacy reference, browse the package's Rd source files.