Skip to main content
Figure 1 | Genome Biology

Figure 1

From: Effectively identifying regulatory hotspots while capturing expression heterogeneity in gene expression studies

Figure 1

GraphicalFigures: format model of ICE and NICE with correction for confounding effects. (a) Genetic and spurious associations. SNP 1 has a genetic effect on the first three genes (blue arrows). SNP 2 has no genetic effect on any of the genes; however, it has spurious associations with many of the genes, because, by chance, it happens to be correlated with a confounding factor (red arrows). (b) ICE [18] models the confounding effects by estimating the global correlation structure of the expression levels of all genes (gray block). However, this eliminates genetic associations in addition to confounding effects since any regulatory hotspots, such as the first three genes, will also be captured in the global correlation structure and be eliminated. (c) NICE uses only a subset of genes to model the global correlation structure between expression levels used to correct for confounding factors. Since any subset of genes will capture the confounding effects, by using the bottom four genes (gray block) we can eliminate the confounding effects but preserve the genetic effects. SNP, single nucleotide polymorphism.

Back to article page