SLR

SLR models array comparative genomic hybridization (aCGH) data using a smoothed logistic regression framework that integrates spatial smoothing and sparsity to classify samples by DNA copy number variations.

Key Features:

• Spatial Information Utilization: Incorporates spatial characteristics of aCGH probes via a mixed logistic regression model with a random component that controls smoothness and sparsity.
• Mixture Distribution: Employs a mixture distribution within the random component to balance sparsity (feature selection) and smoothness across spatially related probes.
• Iterative Weighted Least-Squares Algorithm: Estimates model parameters using an iterative weighted least-squares algorithm accelerated by singular value decomposition (SVD).
• Performance Evaluation: Validated on simulated and real aCGH datasets using leave-one-out cross-validation and reported lower misclassification error rates compared to existing methods.

Scientific Applications:

• Genomic studies: Uses probe spatial context to improve interpretation of DNA copy number variations in aCGH datasets.
• Cancer genomics: Enhances classification of tumor samples from aCGH data based on copy number alterations.
• Genetic disorder diagnostics: Supports classification of samples with pathogenic copy number variants from aCGH profiles.
• Personalized medicine: Provides more precise aCGH-based sample classification to inform understanding and treatment strategies.

Methodology:

Uses a mixed logistic regression model with a random component incorporating a mixture distribution; parameter estimation via an iterative weighted least-squares algorithm with singular value decomposition; performance assessed by leave-one-out cross-validation on simulated and real aCGH datasets.