ccml

ccml implements a two-step consensus clustering approach that accounts for unequal sample coverage to integrate multi-omics datasets and identify molecularly distinct subgroups.

Key Features:

• Handling Unequal Sample Coverage: Processes multiple predictive labels with varying sample coverages (biological replicates) to avoid bias from unequal data availability.
• Two-Step Consensus Clustering Strategy: Normalizes consensus weights by accounting for sample coverage and then applies regular consensus clustering to determine the final cluster configuration.
• Integration with Multi-Omics Data: Facilitates integration of diverse omics datasets for phenotyping and endotyping and complements algorithms such as Similarity Network Fusion.
• R-based Implementation: Provided as an R protocol/implementation for executing the consensus clustering strategy.

Scientific Applications:

• Karolinska COSMIC Cohort: Applied to chronic obstructive pulmonary disease analysis integrating 9-omics to identify molecularly distinct groups.
• U-BIOPRED Cohort: Applied to adult asthma subgrouping using a 24-omics handprint integrative analysis.

Methodology:

Normalizes consensus weights according to sample coverage and subsequently applies regular consensus clustering; implemented in R.