Fig. 1: Overview of the study.

a The CoDeS3D pipeline⁵⁶ was used to construct the Brain Cortex Gene Regulatory Network (BC-GRN) by integrating Hi-C data from cortex cells and eQTL data from brain cortex tissue to map common SNPs to their target genes. The sceQTL-gene associations that passed the Benjamini–Hochberg FDR correction threshold constitute the BC-GRN. b Transcriptome-wide Mendelian randomization (MR) was used to estimate the causal effect of the expression of the brain cortex genes (exposures) on PD using sceQTLs as IVs. c The multimorbid3D pipeline was used to connect the PD-risk genes (identified by MR analysis in b) and PD-associated GWAS SNPs (Nalls et al., 2019) to co-occurring traits and the genetic and biological interaction that connects them (see ‘Methods’). Biological pathways enriched for the PD-causal and PD-associated network were detected. d Medical conditions that co-occur with PD (ICD10 code – G20) were identified within the clinical records of ~2 million NZ public health patients (January 2016 and December 2020) using the comorbidity R package. ICD10 codes for co-occurring conditions were converted to MeSH terms using the Unified Medical Language System application programming interface. The MeSH terms for the observed co-occurring conditions were compared to the MeSH terms for the genes that are linked to PD risk, following disease-gene mapping (R interface of DisGeNET (disgenet2r⁶⁴)).