Epistatic interactions of genetic loci associated with age-related macular degeneration

Figures

• 

  Figure 1. Categorizing independent AMD-associated variants. (a) AMD-associated loci as defined in ref. 2 were assigned to three biological pathways discussed to be relevant in AMD pathobiology, particularly the “Extracellular structure organization”, “Cholesterol transport” and “Regulation of plasma lipoprotein”, and “Immune system process”. A locus was assigned to a pathway if a gene located in this locus (locus definition: index variant(s) and proxies (r2 > 0.5; +/- 100 kb)2 is functionally associated by GO-Terms. Subsequently, the independent signals within these loci were correlated with the liver eQTL dataset reported by Strunz et al.7. (b) Distribution of independent AMD-associated variants over the three biological pathways identified. Overlapping circles indicate independent variants that were assigned to more than one category. Details are given in Supplementary Table S1.

• 

  Figure 2. Schematic project workflow. (a) AMD-associated genetic variants were used to generate unique variant combinations. (b) Based on each combination, a genetic risk score (GRS) was calculated for 588 individuals represented in the liver eQTL dataset7. According to the individual genetic risk profile, each person was graded into risk groups, with a low-risk group containing the 30%-quantile of the lowest scores and a high-risk group containing the 30%-quantile of the highest scores. Consequently, expression of 24,123 genes in liver tissue were compared between the risk groups.

• 

  Figure 3. eCombinations altering expression of LIPC. (a) The boxplot compares LIPC expression between the AMD low and high-risk groups based on the genetic profile of the two genetic variants rs2070895 (orange, located in the LIPC locus) and rs17231506 (blue, located in the CETP locus). The y-axis corresponds to the relative level of gene expression. (b) The tree chart shows related variant combinations influencing the expression of LIPC. The chart starts with a single variant combination at the top and adds one variant per step downwards. Every box represents a set of variants used to determine risk groups. The P-value (P) and the effect size (ES) display the relationship between calculated risk groups and gene expression. Combinations highlighted with an asterisk remain significant after adjustment for multiple testing (Q-value < 0.05). Color code of variants refer to their locus affiliation.

• 

  Figure 4. Interaction of genetic variants assigned to the immune pathway. (a) Number of eCombinations (P-value < 10–7) influencing the expression of the respective eGene given on the Y-axis. eCombinations were generated from genetic variants assigned to the immune pathway by Fritsche et al.2 and only eGenes which were detected at least three times are shown. Applying these thresholds, 438,219 significant results were obtained and related to expression changes of 25 eGenes. Black bars highlight eGenes that overlap with a known AMD locus2. (b) Color-coded effect sizes (ES) reflecting the correlation of eCombinations and gene expression. Colored fields show the absolute ES of an eCombination. White fields indicate a non-significant combination. The X-axis displays the number of genetic variants included in the respective eCombination. The orange line separates three subtypes of results, specifically one group including the CFH-related genes with one genetic variant being sufficient for eQTL detection and high absolute ES (highest absolute ES > 1.8), one group with a single genetic variant being sufficient for eQTL detection but lower ES (highest absolute ES < 1), and one group where an effect…