Hi! This is my first post/question on biostars as I couldn't find my question after a few searches. Here it goes:
Tools like MAGMA (https://ctg.cncr.nl/software/magma) map SNPs to genes by proximity - e.g. if a disease-associated SNP is within a 5kb window to a gene, the gene-to-disease association score increases.
However crude that modeling sounds (at least to me), it seems to capture a lot of useful signal, as it predicts a lot of things correctly -> e.g. tissues (as defined by gene-sets obtained by DESeq2 from expression data) likely to be involved in disease (see https://www.nature.com/articles/s41588-018-0129-5).
Obviously, a lot of intergenic SNPs influence genes very far from themselves, or do not influence genes very close to themselves (especially those in gene-dense regions). Thus, this proximity SNP-to-gene mapping strategy may miss some noise (for ~Mbp distance SNP-to-gene contacts), and may incorporate some avoidable noise (SNPs close to genes they have no relationship with).
I think incorporating eQTL and Chromatin-capture data when doing this step (SNP to gene mapping) from the MAGMA analyses (and other GWAS gene set analysis tools) may or may not improve accuracy.
However, I'm not familiar with the R/Python tools available for exploring public Hi-C & eQTL data. Is there any tool where you can plug in a cell-type/tissue (e.g. naive CD4+ T cell or Whole Blood) & an intergenic SNP rsID, that retrieves the genes this SNP actually changes the expression of in the tissue of interest (from eQTL data), or that actually binds the promoters of (from Hi-C data), as opposed to just the closest genes to the rsID location?