The topmost level represents the 15 phenotypes for which we investigated non-coding variation from the NHGRI-GWAS catalog.
Each phenotype has 4 sub-directories for the different levels of analysis
1. SNPGeneInteractions
These are individual SNP (col 1) gene (col2) interactions that were mapped to our significant interactions for each of the 55 cell types. Each cell type is a file.

2. NodeScores
These files indicate the gene (col 1) 
The score of the gene if it was directly targeted by a SNP (in SNPGeneInteractions). This score is 0 if it is not directly targeted by a SNP. (col 2)
The diffused score of all genes directly and indirectly targeted by a SNP (col 3)

3. EdgeWeights
These files are the diffused edge score for the edges that are in the subgraph for the phenotype (all genes that are direct targets of SNPs or in the top 1% of diffused node scores). Each file represents node 1, node 2 and the diffused edge score.

4. ClusterAssignments
The result of the application of multi-task graph clustering. Each file is simply the gene name and the cluster assignment for each of the 55 cell lines.

**Roadmap Predictions**
We additionally provide the significant interactions for all 55 cell lines in the RoadmapPredictions folder.

**Transitioning gene sets**
These pages display the transitioning gene sets from the result of MTGC for all 15 phenotypes as heatmaps. Additionally, we provide heatmaps for the SNP-gene interactions that are present in the transitioning gene set (if any).

**NetworkFigures**
We visualize the cluster assignments on each network. Each file is a phenotype containing the networkx figures for all 55 cell lines. Color represents cluster assignment, triangle nodes are direct targets and circular nodes are downstream genes.