cNMTF

cNMTF integrates genotype, phenotype, variant-effect, and gene-network data using corrected non-negative matrix tri-factorization to prioritize genetic loci and uncover associations contributing to complex trait heritability.

Key Features:

• Integration of Omics Data: Merges genotype data with other omics datasets to jointly analyze relationships among genotypes, phenotypes, variant effects, and gene networks.
• Corrected Non-Negative Matrix Tri-Factorization: Applies corrected non-negative matrix tri-factorization as the core algorithm for multi-view data factorization and relationship discovery.
• Clustering Techniques: Employs clustering approaches to assess interconnectedness between genotypes, phenotypes, variant effects, and gene networks.
• Identification of Loci–Trait Associations: Evaluates the damaging impact of genetic variants and their network interactions to identify associations between specific loci and traits.
• Prioritization of Weak Loci: Enhances prioritization of weak loci that are often overlooked by standard regression models used in GWAS, addressing missing heritability.
• Replication and Validation: Replicated 129 genes reported in global GWAS and supported 226 out of 265 genes (85%), including novel associations such as NLGN1, lipid metabolism regulator DAB1, and pleiotropic gene CARM1.
• Population Structure Adjustment: Accounts for strong population structure by considering individuals' ancestry to ensure robustness across diverse cohorts.
• Flexibility and Adaptability: Allows incorporation of various omics data sources to tailor analyses to specific research needs.

Scientific Applications:

• Personalized Medicine: Provides integrative insights into complex disease mechanisms to inform candidate gene prioritization for precision medicine studies.
• GWAS Augmentation and Missing Heritability: Complements standard GWAS regression by capturing weak-effect loci and network-mediated associations contributing to missing heritability.
• Lipid Trait Genetics: Identifies regulators and pleiotropic genes relevant to lipid metabolism, exemplified by findings including DAB1 and CARM1.
• Cross-Population Genetic Analysis: Enables analyses that remain robust across diverse cohorts by adjusting for ancestry and population structure.

Methodology:

Uses corrected non-negative matrix tri-factorization and clustering to integrate genotype, phenotype, variant-effect, and gene-network data, evaluates damaging impacts of variants and network interactions, and adjusts for population structure by considering individual ancestry.