MR Atlas

Our Goal: Explore the genetic causal relationship between multiple organs and clinical outcomes

Data

Mendelian randomization and genetic correlation studies (LD score regression, LAVA and PRS) were with the following data types:

  • Brain imaging traits

301 imaging traits from three brain MRI modalities, including 1) 101 regional brain volume; 2) 110 diffusion tensor imaging (DTI) parameters from diffusion MRI; 3)90 functional connectivity traits from resting state functional MRI (fMRI).

  • Heart imaging traits

82 imaging traits from six heart regions, including 1) Ascending aorta (AAo); 2) Descending aorta (DAo); 3) Left ventricle (LV); 4) Right ventricle (RL); 5) Left atrium (LA); 6)Right atrium (RA). 

  • Abdominal imaging traits

11 abdominal imaging traits, measuring volume, fat, and iron content, cover 6 different organs or tissue, including pancreas, liver, spleen, lung, kidney and adipose tissue. More detailed information about the dataset can be found in Liu., et al (2021) (https://doi.org/10.7554/eLife.65554).

  • skeleton imaging traits

8 imaging traits include all long bone lengths as well as hip and shoulder width. More detailed information about the dataset can be found in Kun., et al (2023) ( https://doi.org/10.1101/2023.01.03.521284). 

  • FinnGen clinical endpoints

We selected 88 clinical endpoints from Finngen database (R7) (https://www.finngen.fi/en), where all the selected clinical endpoints have ncase>10,000, except some important diseases, such as Alzheimer’s diseases (ncase > 6000).

Methods

 We assessed the performance of 8 MR methods: Inverse variance weighted (fixed effect), Inverse variance weighted (multiplicative random effect), MR-Egger (used as test for pleiotropy), Weighted Median, Weighted， Mode, DIVW, GRAPPLE, MR-RAPS.

Quality control: 

  • Number of Instrumental variables (IV) > 6
  • Combine robust MR methods and pleiotropy test to further select results 

Study Design and Summary of Findings