MANAclust

MANAclust performs integration and unsupervised clustering of categorical and numeric clinical and multi-omic data to identify disease subsets and endotypes.

Key Features:

• Data Integration: Combines clinical parameters with multi-omic data, handling both categorical and numeric variables.
• Unsupervised Clustering: Applies unsupervised clustering techniques to group samples into molecularly and clinically distinct subsets.
• Feature Selection Algorithms: Incorporates feature selection algorithms that have been demonstrated to be highly accurate for selecting relevant variables from complex datasets.
• Validation on Real and Simulated Data: Has been applied to simulated datasets and real-world data including The Cancer Genome Atlas (TCGA) for method validation.

Scientific Applications:

• Disease Endotype Identification: Identifies endotypes—subtypes with distinct pathophysiological mechanisms—by integrating clinical and multi-omic profiles.
• Asthma Cohort Stratification: Resolves heterogeneous clusters within asthma cohorts, including groups resembling healthy controls and subsets associated with viral infections or allergies.
• Cancer Subtyping: Enables discovery of molecularly distinct patient groups when applied to cancer datasets such as TCGA.

Methodology:

Performs data integration of categorical and numeric clinical and multi-omic inputs, applies built-in feature selection algorithms, and conducts unsupervised clustering to delineate disease subsets.