variable LCV

variable LCV infers and visualizes phenome-wide partial genetic causality across the human phenome using the latent causal variable (LCV) methodology applied to genetic data.

Key Features:

• Latent Causal Variable (LCV) methodology: Provides an alternative to Mendelian randomization by addressing statistical power constraints and horizontal pleiotropy.
• Automated phenome-wide testing pipeline: Executes phenome-wide causal inference across large sets of trait pairs.
• Estimation of partial genetic causality: Estimates the genetic causality proportion between traits.
• Filtering and multiple testing correction: Applies filtering to generate a meaningful set of estimates and corrects for multiple testing to ensure statistical robustness.
• Causal architecture plots: Produces causal architecture plots that visualize phenome-wide partial genetic causality estimates.
• Applications to specific traits: Has been applied to body mass index (BMI), lipid traits, dental caries, and periodontitis to evaluate upstream versus downstream relationships.
• Integration with MASSIVE pipeline: Integrates with the MASSIVE pipeline from the Complex-Traits Genetics Virtual Lab for automated processing.

Scientific Applications:

• Hypothesis-free phenome-wide causal inference: Enables systematic, hypothesis-free estimation of genetic causality proportion across numerous trait pairs.
• Identification of upstream risk factors: Supports findings that BMI and lipid traits act as upstream causal risk factors influencing multiple conditions.
• Assessment of downstream consequences: Identifies dental caries and periodontitis primarily as downstream consequences rather than causal agents of systemic ill health.
• Complement to Mendelian randomization: Serves as an alternative or complement to Mendelian randomization when MR is limited by power or horizontal pleiotropy.

Methodology:

Applies the latent causal variable (LCV) method to estimate partial genetic causality between trait pairs, followed by filtering, multiple testing correction, and visualization via causal architecture plots, integrated into the MASSIVE pipeline.