iClusterPlus

iClusterPlus performs integrative clustering of multi-type omics data to identify clinically relevant tumor subtypes and driver molecular alterations in cancer genomic studies such as The Cancer Genome Atlas (TCGA).

Key Features:

• Multi-omics integration: Integrates genomic, transcriptomic, epigenomic, and proteomic data for joint analysis.
• Data-type support: Models both continuous and discrete types of omics data.
• Latent-variable dimension reduction: Uses a small number of latent variables to capture shared structure across datasets and enable joint dimension reduction.
• Feature selection: Employs Bayesian variable selection to identify omics features that drive sample clustering.
• Scalability to large studies: Applicable to large-scale genomic profiling studies, including TCGA datasets and simulated data used for validation.

Scientific Applications:

• Tumor subtype discovery: Identification of clinically relevant tumor subtypes from integrated multi-omics profiles.
• Driver alteration and biomarker identification: Detection of driver molecular alterations and candidate biomarkers for precision medicine applications.
• Method validation: Benchmarking and validation using TCGA datasets and simulated data to assess clustering performance.

Methodology:

Implements a fully Bayesian latent variable framework that models continuous and discrete omics data using a few latent variables for joint dimension reduction, applies Bayesian variable selection to identify relevant features, and clusters samples within the latent variable space.