qpgraph

qpgraph estimates eQTL networks from high-throughput gene-expression and genotyping assay data by modeling conditional independences to identify direct genotype–phenotype associations while accounting for confounding and unwanted variation.

Key Features:

• Mixed Graphical Markov Models: Uses mixed graphical Markov models to capture higher-order conditional independences and adjust for indirect effects from genetic, molecular, and environmental perturbations.
• q-Order Correlation Graphs: Employs q-order correlation graphs to model how additive genetic effects propagate through gene networks based on gene–gene correlations.
• Adjustment for Confounding and Unwanted Variation: Adjusts for confounding factors and sources of unwanted variation in gene-expression data using statistical modeling.
• Sparse Network Estimation: Produces sparse eQTL network structures that highlight direct genetic and regulatory associations and facilitate comparison of genetic control across chromosomes.

Scientific Applications:

• Direct genetic and regulatory association mapping: Identifies direct genotype–phenotype associations and regulatory links within gene-expression networks.
• eQTL contribution to hub gene variability: Quantifies how eQTLs contribute to expression variability of network hub genes.
• Study of gene regulation and genetic architecture: Supports analyses of gene regulation, genetic architecture, and mechanisms underlying phenotypic traits.
• Chromosome-wise comparison of genetic control: Facilitates comparison of genetic control mechanisms across different chromosomes.

Methodology:

qpgraph adjusts for confounding and unwanted variation and leverages mixed graphical Markov models together with q-order correlation graphs to estimate sparse eQTL networks.