stepwiseCM: An R Package for Stepwise Classification of Cancer Samples Using Multiple Heterogeneous Data Sets

Area of Science:

• Bioinformatics
• Computational Biology
• Cancer Research

Background:

• Cancer sample classification often relies on multiple data types.
• Integrating heterogeneous data for classification can be complex and costly.
• Existing methods may require extensive data collection, increasing costs and patient burden.

Purpose of the Study:

• To introduce the stepwiseCM R/Bioconductor package for efficient cancer sample classification.
• To present a novel algorithm that leverages distinct classification powers from two heterogeneous datasets without direct combination.
• To develop a cost-efficient classification strategy that minimizes the need for expensive or invasive measurements.

Main Methods:

• Utilizes the stepwiseCM package in R/Bioconductor for classification.
• Employs an algorithm that captures unique classification power from two heterogeneous data types.
• Implements functions for projecting neighborhood information between data spaces to identify samples benefiting from additional measurements.
• Connects heterogeneous data spaces via indirect mapping.

Main Results:

• The stepwiseCM package enables efficient classification of cancer samples using two distinct data types.
• The algorithm effectively utilizes the classification power of each data type independently.
• Demonstrates utility in combining clinical covariates with high-dimensional data (e.g., gene expression) or two high-dimensional types (e.g., DNA copy number and mRNA expression).
• Identifies samples that would benefit most from additional covariate measurements, optimizing data collection.

Conclusions:

• The stepwiseCM package offers a cost-efficient and effective approach to cancer sample classification.
• This method reduces the need for extensive data collection, leading to faster diagnoses and decreased patient anxiety.
• The package addresses key limitations of existing methods, enabling broader applications in cancer research and clinical settings.