GWAS Imputation Toolkit: GTEx v8 QTL Models and 1000G Reference Data

Alvaro Barbeira's sample archive provides the necessary components to impute GWAS summary statistics for integration with PrediXcan MASHR-M models from the GTEx v8 project. Its contents include a 1000 Genomes reference panel in hg38, precomputed LD regions, and coordinate mapping tables for harmonizing genomic assemblies. This dataset is designed as a tutorial resource for performing transcriptome-wide association studies.

Use Cases

• Imputing missing summary statistics for GWAS using the provided 1000 Genomes reference panel.
• Harmonizing genomic coordinates across different human genome assemblies (e.g., hg17 to hg38) using the precomputed liftover chains and coordinate maps.
• Performing PrediXcan/S-MultiXcan analysis for gene-trait association by applying the included MASHR-M eQTL and sQTL models.
• Filtering genetic variants for European populations based on the provided MAF > 0.01 annotation file.
• Defining approximately independent linkage disequilibrium (LD) regions for analysis using the provided Berisa-Pickrell region definitions.

Strengths

• Includes a specific, high-quality reference panel: 1000 Genomes Project genotypes in hg38 parquet format.
• Contains detailed, precomputed components like LD regions (Berisa-Pickrell) and genome assembly coordinate maps, reducing computational setup.
• Integrates with a major public resource, using QTL models derived from the GTEx v8 dataset.

Limitations

• Key metadata like row count, file size, and specific column names are not provided across all sources.
• The dataset's primary purpose is described as a tutorial, suggesting it may contain a limited sample GWAS rather than a production-scale resource.
• The 'Last updated' date is consistently listed as unknown, making data freshness unclear.

Provenance

Source

Author: Alvaro Barbeira. Organization: University of Chicago. Sample GWAS data from Nikpay et al. (Nat Gen 2016).

Collection Method

Likely compiled and structured as a support package for the MetaXcan/PrediXcan software tools.

Geography

Genetic data is filtered for variants with MAF > 0.01 in Europeans; reference panel is from the global 1000 Genomes Project.