SGI: automatic clinical subgroup identification in omics datasets.

Area of Science:

• Computational biology
• Bioinformatics
• Data science

Background:

• Large-scale omics datasets present challenges in identifying distinct biological subgroups.
• Automated methods are needed to systematically analyze clinical parameters and outcomes within these datasets.

Purpose of the Study:

• To introduce the Subgroup Identification (SGI) toolbox, an R package for automated clinical subgroup detection.
• To demonstrate the SGI toolbox's capability in analyzing omics data with multiple clinical parameters.

Main Methods:

• The SGI toolbox employs hierarchical clustering trees combined with association testing and visualization.
• It supports processing of arbitrary numbers of clinical parameters and outcomes.
• A multi-block extension enables simultaneous analysis of multiple omics datasets from the same samples.

Main Results:

• The SGI toolbox was applied to a type 2 diabetes metabolomics study.
• Its utility was further demonstrated on copy number variation datasets from The Cancer Genome Atlas.
• The package provides a systematic framework for subgroup discovery.

Conclusions:

• The SGI toolbox offers a robust and flexible algorithm for identifying clinical subgroups in omics data.
• It facilitates deeper biological insights by integrating omics data with clinical information.
• The open-source R package is readily available for researchers.