FunGen-xQTL Resource File Format Description

Overview

This repository provides Quantitative Trait Loci (QTL) analysis and GWAS integration results from the FunGen-xQTL project, across diverse aging human brain regions and cell types. The analyses span multiple cohorts including ROSMAP, Knight ADRC, Mount Sinai Brain Bank, MiGA, MetaBrain, and STARNET.

For comprehensive metadata of all analyses with harmonized context definitions, see our analysis meta-data table.

Fine-mapping and integrative results for xQTL and AD GWAS

The following table catalogs all fine-mapped QTL and AD GWAS integration results, organized by molecular modality, cohort, tissue context, and analytical method. Each entry represents fine-mapped genetic loci for corresponding molecular phenotypes, with top loci and variants exported in standardized BED format.

• AD_xQTL_colocalization results include “AD_Bellenguez_2022”, “AD_Wightman_Excluding23andMe_2021”, “AD_Wightman_ExcludingUKBand23andME_2021”, “AD_Kunkle_Stage1_2019”, “AD_Wightman_Full_2021”

Data Format Specifications

Each exported file follows a standardized BED format with compressed storage (.gz), containing top loci from respective QTL analyses. Files are optimized for downstream genomic and transcriptomic research applications.

SuSiE/mvSuSiE fine-mapping columns

ColocBoost-specific Columns

TWAS-Specific Columns

cTWAS-Specific Columns

For detailed documentation, please refer to cTWAS (causal-TWAS)

fSuSiE-Specific Columns

SuSiE-COLOC Colocalization Columns

Between xQTL and AD, pairwise.

Consolidated single context xQTL loci

FunGen_xQTL_single_context_top_loci_integrated.rds

All single-context SuSiE fine-mapping results in one R data frame: 13,393,657 rows, one per variant and molecular trait. It covers 51 contexts (eQTL, productive and unproductive sQTL, pQTL and glycoQTL), 634,382 molecular traits, 18,638 genes and 5,147,485 variants. Results from fSuSiE (mQTL, haQTL, caQTL) are not in this file.

FunGen_xQTL_AD_loci_context_method_summary.fsusie.rds

AD GWAS fine-mapped variants with the xQTL evidence found for them, in one R data frame: 47,722 rows, one per variant, AD GWAS, gene and type of xQTL analysis. It covers 1,517 variants in 56 regions, 143 genes, 97 xQTL contexts and eight AD GWAS. Despite the file name, the rows come from all analysis types, not only fSuSiE.

Usage Guidelines

Prioritization Strategies

• Causal variant identification: Filter by cs_coverage_0.95 > 0 or PIP > 0.1 to identify high-confidence causal variants
• Significant results only: Restrict analysis to rows with non-missing event_ID values for significance-filtered datasets
• Gene-level visualization: Group results by context and gene_ID to examine credible sets across studies
• Multi-context analysis: Rows with missing event_ID from multi-context analyses represent variants not passing LFSR threshold but retained for completeness

Integration Notes

• All files use standardized variant naming convention for cross-dataset integration
• Credible set columns (cs_coverage_*) indicate membership probability thresholds with varying purity requirements
• For epigenetic marks in fSuSiE results, note that epi_mark_effects are interpolated from neighboring grid effects rather than direct measurements

xQTL Association Results Summary

The following table provides paths to marginal QTL association results, including linear regression (LR), quantile regression (QR), and interaction analyses testing for genotype-by-sex, genotype-by-APOE, and other context-dependent effects.

For detailed methodology and implementation, refer to our analysis protocol documentation.