Fig. 1: Data overview and study design.

A An overview of the data included in our study. For imaging phenotypes, we made use of a broad range of multi-organ imaging phenotype data including brain imaging traits such as regional brain volume from structural MRI, diffusion tensor imaging (DTI) parameters from diffusion MRI, and functional connectivity and activity traits from resting and task fMRI; CMR traits such as measurements of the aorta (ascending/descending aorta) and heart chambers (left ventricle, right ventricle, left atrium, and right atrium); abdominal MRI traits such as abdominal organ traits (liver, kidney, lung, spleen, and pancreas) and abdominal composition measurements of fat and muscle. For genetic data, we included whole exome sequencing data and focused on rare coding variants (minor allele frequency [MAF] <0.01). Only European individuals with imaging traits were included in the present study (n = 54,365). Some elements in this figure were created in BioRender: Fan, Y. (2025) https://BioRender.com/oz6gd9i, Fan, Y. (2025) https://BioRender.com/f2ok1xo, Fan, Y. (2025) https://BioRender.com/x77lwka, Fan, Y. (2025) https://BioRender.com/29marqh, Fan, Y. (2025) https://BioRender.com/a8yoefm, and Fan, Y. (2025) https://BioRender.com/19zbtn5. B An overview of our study design. We adopted an internal discovery-replication procedure and finalized the summary statistics based on the joint sample (phases 1 to 5 sample). Specifically, we conducted association tests on phases 1 to 3 sample, phases 4 to 5 sample, and phases 1 to 5 sample respectively. We examined (i) whether the detected signals from phases 1 to 3 sample had concordant directions and remained significant in phases 4 to 5 sample; (ii) whether the detected signals from phases 1 to 3 sample had concordant directions and obtained stronger evidence (smaller P-values) in the joint sample. Based on this validated procedure, the significant results and other downstream analyses used the summary statistics generated from this joint sample. The average n indicates the average sample size used in the tests across all 596 phenotypes. Some elements in this figure were created in BioRender: Fan, Y. (2025) https://BioRender.com/n0j8gb2 and Fan, Y. (2025) https://BioRender.com/wawasqc.