intCC

intCC performs integrative clustering of multiomics datasets to identify latent cluster structures for applications such as cancer subtype discovery.

Key Features:

• Integrative clustering framework: intCC applies unsupervised integrative clustering across multiple datasets to support subtype discovery in complex diseases.
• Weighted integrative clustering: It combines ensemble methods, consensus clustering, and kernel learning to weight and integrate information across diverse data types.
• Consensus matrix derivation: The method applies consensus clustering per dataset using user-specified clustering methods and numbers of clusters (per-dataset and global) to derive consensus matrices for integration.
• Validation via simulation and TCGA pan-cancer case study: Performance has been evaluated using simulation studies and a case study on TCGA pan-cancer datasets.

Scientific Applications:

• Subtype discovery in cancer: Analysis of high-throughput multiomics data to discover and characterize cancer subtypes.
• Complex disease research: Integration of diverse omics datasets to investigate biological heterogeneity in complex diseases beyond cancer.

Methodology:

Users provide multiple datasets with specified clustering methods and clustering parameters including per-dataset and global numbers of clusters. Consensus clustering is applied to each dataset to generate consensus matrices. These matrices are integrated using a weighted approach that incorporates ensemble methods, consensus clustering, and kernel learning. The integrated consensus is used to identify latent cluster structures.