mRNA expression quantitative trait loci used in the manuscript titled: MR-link-2: pleiotropy robust cis Mendelian randomization validated in three independent reference datasets of causality

# eQTLGen Data for MR-link-2This repository contains 54 Parquet files generated by the eQTLGen Consortium. These datasets were used in the [MR-link-2 paper]() to identify the correct direction of causality.---## Important Disclaimer- **These data are provided for reproducing the results of the MR-link-2 paper.**- There is an **upcoming publication** from the eQTLGen Consortium that will feature **larger sample sizes** and **additional gene expression phenotypes**. If you require more comprehensive datasets, please refer to the forthcoming publication once it is available.---## Files Overview- **Number of files:** 54 - **Format:** Parquet - **Content:** Cell-type marker gene eQTL data generated by the eQTLGen Consortium - **Harmonization:** Each file has been harmonized to the UK10K genotype reference and is in b37 ---## Data FormatThe Parquet files follow the **GWAS format specification**, which also makes them compatible with the MR-link-2 method. The schema of each file is as follows:```plaintextchromosome | stringbase_pair_location | uint64effect_allele | stringother_allele | stringbeta | float64standard_error | float64effect_allele_frequency | float64p_value | float64variant_id | stringrsid | stringn | uint64```**Field Descriptions:**- **chromosome**: The chromosome on which the variant is located (e.g., "1", "2", "X"). - **base_pair_location**: The base pair position of the variant on the chromosome. - **effect_allele**: The allele for which the effect size (`beta`) is reported. - **other_allele**: The non-effect allele. - **beta**: The estimated effect size of the effect allele on the gene expression. - **standard_error**: The standard error of the effect size estimate. - **effect_allele_frequency**: The frequency of the effect allele in the reference dataset (UK10K). - **p_value**: The p-value for the association. - **variant_id**: A unique identifier for the variant (often in the format `chr:pos:ref:alt`). - **rsid**: The dbSNP rs identifier (if available). - **n**: The sample size for the specific variant-gene association.------## Getting Started1. **Dependencies**: - A Parquet-reading library (e.g., [pyarrow](https://arrow.apache.org/docs/python/) in Python) to load these files. - Any standard GWAS/QTL analysis tool or MR-link-2 method pipeline.2. **Loading Example (Python)**: ```python import pyarrow.parquet as pq # Example: load the first Parquet file file_path = "eqtlgen_marker1.parquet" table = pq.read_table(file_path) df = table.to_pandas() # Inspect the first few rows print(df.head()) ```3. **Analysis**: - After loading, the data can be integrated into your pipelines for QTL mapping, colocalization studies, or Mendelian randomization analyses with MR-link-2.---## Contact and SupportIf you have any questions regarding the usage or interpretation of these datasets in the context of MR-link-2:- **MR-link-2 Support**: [https://github.com/adriaan-vd-graaf/mrlink2](https://github.com/adriaan-vd-graaf/mrlink2)Please direct inquiries about the upcoming eQTLGen dataset to the eQTLGen Consortium.