Skip to main content
Fig. 2 | Genome Biology

Fig. 2

From: Modelling local gene networks increases power to detect trans-acting genetic effects on gene expression

Fig. 2

Benchmark of alternative eQTL association methods on simulated datasets. a, b Receiver operating characteristic (ROC) for alternative methods to detect eQTLs in sparse (a) and star-shaped simulated regulatory networks (b), assuming no confounding factors. c, d Power comparison when increasing the relative effect of confounding, either assuming sparse (c) or star-shaped network topologies (d). Compared are a standard LMM without conditioning (LMM), an LMM that exclusively conditions on true exogenous genes (Ideal-LMM), adjustment based on principal components (PC-LMM), adjustment based on principal components with selection (PCselect-LMM) and GNet-LMM. Power is defined as the area under the ROC curve for a false positive rate below 5 %

Back to article page