Informed-MiXeR

Informed-MiXeR applies a likelihood-based model to GWAS summary statistics and reference panels to estimate the number of causal variants and their effect-size distributions across functional genomic annotation categories, elucidating the genetic architecture of complex human traits and diseases.

Key Features:

• Functional annotation analysis: Analyzes annotation categories including protein-coding exons and regulatory regions to quantify their relative contributions to complex traits.
• Polygenicity and effect-size distribution: Estimates phenotype-specific differences in polygenicity and effect-size distributions across functional annotation categories.
• Likelihood-based estimation: Employs a likelihood framework to infer numbers of variants and their effect sizes from GWAS summary statistics.
• Input data: Operates on genome-wide association study (GWAS) summary statistics together with a reference panel.
• Simulation validation: Validated across a broad range of genetic architectures using extensive simulations.
• Implementation (Python): Implemented in Python.
• Visualization capabilities: Provides visualization routines to aid interpretation of annotation-specific polygenicity and effect-size results.

Scientific Applications:

• Disease-specific architecture analysis: Identifies genomic regions most associated with specific diseases, observing a predominance of protein-coding exons in type 2 diabetes and inflammatory bowel disease.
• Regulatory mechanism exploration: Highlights the importance of non-coding regulatory regions in schizophrenia, bipolar disorder, and attention-deficit/hyperactivity disorder, where fewer causal variants are located in protein-coding regions.

Methodology:

Uses GWAS summary statistics and a reference panel as input and employs a likelihood-based framework to estimate the number of variants affecting phenotypes and their effect-size distributions across functional annotation categories.