BOOME

BOOME applies the BOOsting algorithm for Measurement Error to correct measurement error and perform variable selection in ultrahigh-dimensional gene expression datasets with binary responses.

Key Features:

• Measurement Error Correction: Corrects measurement errors in both response variables and covariates to reduce bias in downstream analyses.
• Variable Selection and Parameter Estimation: Uses a boosting procedure to select informative variables while simultaneously estimating model parameters.
• Model Support: Targets logistic regression and probit models for analysis of binary responses under measurement error.
• Ultrahigh-Dimensional Data Handling: Designed to operate with ultrahigh-dimensional predictors typical of gene expression studies.
• Python Implementation: Provided as a Python package implementing the BOOsting algorithm for Measurement Error.

Scientific Applications:

• Gene Expression Data Analysis: Corrects measurement errors and selects predictors in ultrahigh-dimensional gene expression datasets to improve inference.
• Robust Regression with Binary Outcomes: Enhances robustness of logistic and probit regression models when responses and predictors are contaminated by measurement error.

Methodology:

BOOME employs the BOOsting algorithm for Measurement Error, an iterative boosting algorithm that refines model parameters while correcting measurement error effects in responses and predictors.