CLARITY

CLARITY compares similarity matrices from heterogeneous datasets to quantify conserved and divergent (dis)similarities and relationships between entities across data sources.

Key Features:

• Quantification of Consistency: Quantifies the consistency of (dis)similarities between entities across different datasets.
• Identification and Interpretation of Inconsistencies: Identifies where inconsistencies arise between similarity matrices and aids interpretation of these discrepancies.
• Non-Parametric Approach: Employs a non-parametric methodology that is robust to noise and differences in scaling and makes minimal assumptions about data generation.
• Decomposition of Similarities: Decomposes similarities into a 'structural' component akin to clustering and an underlying 'relationship' between these structures.
• Structural Comparison: Focuses on the predictability of one similarity matrix from another's structure to highlight conserved or divergent patterns.
• Significance Assessment: Assesses significance using re-sampling techniques tailored to each dataset.

Scientific Applications:

• Gene Methylation vs. Expression: Analyzing relationships between gene methylation patterns and gene expression levels.
• Evolution of Language Sounds vs. Word Use: Investigating parallels between changes in language sounds and their usage over time.
• Country-Level Economic Metrics vs. Cultural Beliefs: Exploring connections between economic indicators and cultural beliefs across countries.

Methodology:

Decomposes similarity matrices into structural and relationship components, applies a non-parametric comparison that assesses predictability of one matrix from another's structure, and evaluates significance via dataset-tailored re-sampling; the approach is robust to noise and scaling differences.